v:ekn ranking list

does anyone know what happened to the v:ekn ranking list? http://206.65.59.246/vtesregistry/topranking.tpl the last time my ranking was updated was after a tournament in september. i took part in 2 tourneys after that, one on november 24th and the other on january 5th 2002, but the ranking hasn't changed a bit. when was the last last time people have noticed any changes? regards pallando

"pallando" <pall...@NOSPAM.gmx.at> wrote in message news:a2eb4m$j2g$1...@bird.wu-wien.ac.at... [ quoted text not captured ] I participated ina tourney 30th of December and my ranking has been changed according to that. However it hasn't been updated since the tourney I played 12th of Jan. Roger

[ quoted text not captured ] The results of my tounament held January 13 are updated on the V:EKN registry. (Remy Auclair is now on the Top 20 players...) Maybe the informations of the tournaments you speak of have not been sent to Todd ? Maybe you could ask the organisator of those tournanent about it ? Bye, Khaleph the Setite http://veknquebec.web-page.net

kha...@hotmail.com (Khaleph Alendis) wrote in message news:<3c4b4763...@news.oricom.ca>... [ quoted text not captured ] Following this, I was a little upset that my ranking has dropped 57 places even though I'm sure I've been in the top 3 in all tournaments since updated. How the hell did that happen???

> Following this, I was a little upset that my ranking has dropped 57 > places even though I'm sure I've been in the top 3 in all tournaments > since updated. > > How the hell did that happen??? Do you play the same people always? Isn't the system set up that people with high scores beating people with very low scores get smaller adjustments to their ratings? That combined with new players getting ranked it makes sense. You might get top 3rd, but there are possibly people that are pulling off top 2, or first. If they are comming off of poor previous ratings their potential to increase their ratings are much greater then yours. -- Aaron The Nosferatu Stuff

"Barny Baker" <barny...@synergygroup.co.uk> schrieb im Newsbeitrag news:a9abcc59.02012...@posting.google.com... > > Following this, I was a little upset that my ranking has dropped 57 > places even though I'm sure I've been in the top 3 in all tournaments > since updated. > > How the hell did that happen??? this is easy to explain. you most probably had a ranking beyond 3300, maybe even close to 3350. at this stage you need a table win in almost every game you play just too keep your rating. also the games are applied in the order they were played. the rating doens't care if you came out 2nd or 3rd in a tourney, specially if you failed to gain more than 0 or 1 vps in the final. in a big tournament it is better for your rating if you finish in 6th place with a strong 3rd round than if you go into the final and don't get any vps there. regards, pallando

barny...@synergygroup.co.uk (Barny Baker) wrote in message news:<a9abcc59.02012...@posting.google.com>... > kha...@hotmail.com (Khaleph Alendis) wrote in message news:<3c4b4763...@news.oricom.ca>... [snipped some previous entries] > > Following this, I was a little upset that my ranking has dropped 57 > places even though I'm sure I've been in the top 3 in all tournaments > since updated. > > How the hell did that happen??? Not sure why you'd drop so much. It might help to see the rankings of the players close to yours? I was thinking it'd be really nice if when you look up your own ranking you could see the 10 or so players above and below your rank also displayed? (is not too hard an sql statement and would be interesting for players to see methinks, though maybe database load is an issue?) Ability to see ranks of all players for a given city/region in one table (rather than links to individual players' info and rank. Ie a table like the top 20 being displayed if selected as an option?) Simidh, Prince of Adelaide

barny...@synergygroup.co.uk (Barny Baker) wrote in message news:<a9abcc59.02012...@posting.google.com>... > > Following this, I was a little upset that my ranking has dropped 57 > places even though I'm sure I've been in the top 3 in all tournaments > since updated. > > How the hell did that happen??? Could it be because you suck Baker? Marty '17th place in the World' Cumberland :)

"Barny Baker" <barny...@synergygroup.co.uk> wrote in message news:a9abcc59.02012...@posting.google.com... > Following this, I was a little upset that my ranking has dropped 57 > places even though I'm sure I've been in the top 3 in all tournaments > since updated. > How the hell did that happen??? You keep losing that final round, I guess. Your ranking goes up (potentially) during each of the preliminary rounds, then goes down (significantly) during the final round simply because it is played last, and entered last in the Archon spreadsheet. For the purposes of maintaining rank, it is better to qualify for the final and not play than it is to qualify for the final and do anything other than win. Carpe noctem. Lasombra http://www.TheLasombra.com -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG

Barny Baker wrote: > Following this, I was a little upset that my ranking has dropped 57 > places even though I'm sure I've been in the top 3 in all tournaments > since updated. > > How the hell did that happen??? > It happened because the rating-system is thoroughly broken - or more correctly, it's not made for handling a multiplayer environment. I suggested earlier that now is a golden opportunity to change it slightly as we now have the system with Game Wins. Keep the algorithm, it's more than well-proven (standard issue international chess-rating). The one big, and I admit, it's big, disadvantage is that current ratings may need to be reset, but I actually don't think so. Suggestion in concrete terms: If a table creates a Game Win then everyone looses against the person getting the Game Win. If a table doesn't create a Game Win then everyone draws. That's it, nothing else, no comparision between second and third place and you only win when according to the new rules you actually Game Win. The difference from today is that bad seating, bad luck, bad breath and bad mana in general should cost you a maximum of aprox 30 points during any one game. It also reflects the rules of the game better as there really (at a five player table) should be one winner and four loosers. Sten During

Sten During wrote: > > Barny Baker wrote: > > > Following this, I was a little upset that my ranking has dropped 57 > > places even though I'm sure I've been in the top 3 in all tournaments > > since updated. > > > > How the hell did that happen??? > > > > It happened because the rating-system is thoroughly broken - or more > correctly, it's not made for handling a multiplayer environment. It works at least as well as the "solution" you propose, which largely ignores the multiplayer aspect. > I suggested earlier that now is a golden opportunity to change it > slightly as we now have the system with Game Wins. Keep the > algorithm, it's more than well-proven (standard issue international > chess-rating). > > The one big, and I admit, it's big, disadvantage is that current > ratings may need to be reset, but I actually don't think so. > > Suggestion in concrete terms: > > If a table creates a Game Win then everyone looses against the person > getting the Game Win. > > If a table doesn't create a Game Win then everyone draws. This is not logical. Getting 2VP is harder than getting 0VP, and should be rated accordingly. > The difference from today is that bad seating, bad luck, bad breath > and bad mana in general should cost you a maximum of aprox 30 points > during any one game. It also reflects the rules of the game better > as there really (at a five player table) should be one winner and > four loosers. That's handled properly by the current system (the one winner scores a win vs. each of the non-winners). The current system merely has the added value of rating the non-winners play properly amongst themselves (a 1VP non-winner is rated better than a 0VP non-winner, unlike your proposed system in which 1VP ties with 0VP - a situation not true in tournaments, where VPs are used as tiebreakers after GWs). -- LSJ (vte...@white-wolf.com) V:TES Net.Rep for White Wolf, Inc. Links to revised rulebook, rulings, errata, and tournament rules: http://www.white-wolf.com/vtes/

The Lasombra <thela...@hotmail.com> wrote in message news:bca55d83bfbb07846ee...@mygate.mailgate.org... > > For the purposes of maintaining rank, it is better to qualify for > the final and not play than it is to qualify for the final and do > anything other than win. ???? WTF. Surely this is a screwed up way of scoring ? Rob

[ quoted text not captured ] No. Of course, losses against finalists tend to do less damage to one's rating, since the finalists tend to have higher ratings (for having won enough GWs and VPs to get to the finals). See also <insert your favorite game with a zero-sum ratings system>. [ quoted text not captured ]

martin.c...@southbeds.gov.uk (Marty) wrote in message news:<a1d15b7.02012...@posting.google.com>... [ quoted text not captured ] Ok, further to my last message (above), I'd just like to go on record as saying Barny really is that good. But it amazes me how someone who gets to EVERY final can drop 50 places and someone who does reasonable well once or twice can rocket up the ranking ladder.... How on earth do they work this thing? Marty 'Yes, technically I am 17th place but Barny should at least be 3rd' Cumberland P.S. There ya go Barn, that better?

"pallando" <pall...@NOSPAM.gmx.at> wrote in message news:a2eb4m$j2g$1...@bird.wu-wien.ac.at... [ quoted text not captured ] I am constantly updating ratings as events come in. If your rating hasn't changed in a while, contact your event coordinator and have them (re)send me the results. If I do not get the results - then they don't get updated. Take care, Todd Banister

> Following this, I was a little upset that my ranking has dropped 57 > places even though I'm sure I've been in the top 3 in all tournaments > since updated. > > How the hell did that happen??? The reason why your rating fell is you didn't do to very well at the Dec. 9th event. Going into the Dec 9th event: 3288 Rating after round 1: 3301 Rating after round 2: 3240 Rating after round 3: 3168 Rating after the finals: 3122 Then, at the Dec 23rd event: Rating going into event: 3122 Rating after round 1: 3161 Rating after round 2: 3184 Rating after round 3: 3206 Rating after the finals: 3157 Take care, Todd Banister

Sten During <ya...@netg.se> wrote in message news:<3C4C3D5...@netg.se>... [ quoted text not captured ] Well having listened to all these points with great interest I have concluded that the ranking system should be shoved somewhere with an extra hard shove, by that new Worlds Strongest Man (Sven or Olaf or something). I will however admit that when I was briefly first and a steady 2nd or third I thought the system was very good. I dont think a game wins league would work, you need something where every player can compare themselves with the leaders, I suppose a rank system based purely on final position in the tournament with weighting for numbers attending the tournament and a table wins consideration could be looked at, but then again. Barny Baker 'Only ever been beaten once by Martin Cubberley in tournament play and he can Shove it also'

LSJ wrote: > Sten During wrote: > >>Barny Baker wrote: >> >> >>>Following this, I was a little upset that my ranking has dropped 57 >>>places even though I'm sure I've been in the top 3 in all tournaments >>>since updated. >>> >>>How the hell did that happen??? >>> >>> >>It happened because the rating-system is thoroughly broken - or more >>correctly, it's not made for handling a multiplayer environment. >> > > It works at least as well as the "solution" you propose, which largely > ignores the multiplayer aspect. > That was kind of the idea behind my suggestion ;) > >>I suggested earlier that now is a golden opportunity to change it >>slightly as we now have the system with Game Wins. Keep the >>algorithm, it's more than well-proven (standard issue international >>chess-rating). >> >>The one big, and I admit, it's big, disadvantage is that current >>ratings may need to be reset, but I actually don't think so. >> >>Suggestion in concrete terms: >> >>If a table creates a Game Win then everyone looses against the person >>getting the Game Win. >> >>If a table doesn't create a Game Win then everyone draws. >> > > This is not logical. Getting 2VP is harder than getting 0VP, and should > be rated accordingly. Different, yes. Explicitly illogical, no. The new rules actually does state that getting 3-0-0 VP during the prelims should be rated higher than getting 2-2-2 (in case of VERY bad luck), and from some of the reactions online a lot of people seem to think that this is illogical as well. To me it's equally illogical to apply a ratingsystem designed for games with only two sides to a multiplayer game, especially if such a game supports ordered endresults. I just happen to agree with the new rules in more than one aspect. It does, after all, correspond more closely to the Object of the Game. > > >>The difference from today is that bad seating, bad luck, bad breath >>and bad mana in general should cost you a maximum of aprox 30 points >>during any one game. It also reflects the rules of the game better >>as there really (at a five player table) should be one winner and >>four loosers. >> > > That's handled properly by the current system (the one winner scores > a win vs. each of the non-winners). The current system merely has the > added value of rating the non-winners play properly amongst themselves > (a 1VP non-winner is rated better than a 0VP non-winner, unlike your > proposed system in which 1VP ties with 0VP - a situation not true in > tournaments, where VPs are used as tiebreakers after GWs). > > Actually I didn't suggest that 1VP ties with 0VP. They both lose to the one with a Game Win. No rating-comparision done between those who lose. The only time you use the algorithms for a tie is when the table doesn't generate a Game Win - which makes a 2-2-1-0-0 table tie, and I can agree that a lot of players will have a hard time accepting that 2VP ties with 0VP in this case, but conceptually you bring out the rules for a tie when 'noone won'. I guess it all falls back to how you look at the game. I really favour the 'no credit for a nice try' that's part of the game. I'm getting used to that getting 1 VP or 0 VP is more a matter of seating than skill (rush-decks going upstream in order to get their backs clean when going forward, combat-walls squeezing predator to death while using a forward motion comparabe to *ahem* a wall). I have yet to see a High Visibility Deck making it into the finals that actually isn't taken care of comparatively soon there. That hardly constitutes as it's predator playing better. However, and more important, the game seems to be designed as an all or nothing game. Tournament rules have changed in this direction, and that gave me the idea for this suggestion. Sten During

> Different, yes. Explicitly illogical, no. The new rules actually does > state that getting 3-0-0 VP during the prelims should be rated > higher than getting 2-2-2 (in case of VERY bad luck), and from some > of the reactions online a lot of people seem to think that this is > illogical as well. Sten, may I play at those 6 player tables next time I visit Gothenburg too? ;-) Roger

"Todd Banister" <vtesr...@white-wolf.com> wrote in message news:WgY28.567$oT3....@eagle.america.net... [ quoted text not captured ] I contacted him and he said he sent them in the same evening as the tournament. Roger

Sten During wrote: > LSJ wrote: > > Sten During wrote: > >>It happened because the rating-system is thoroughly broken - or more > >>correctly, it's not made for handling a multiplayer environment. > >> > > It works at least as well as the "solution" you propose, which largely > > ignores the multiplayer aspect. > > That was kind of the idea behind my suggestion ;) ? You complain that the current system is "broken" and equate that with the theory (however flawed) that it isn't made for handling a multiplayer environment, then claim that your idea of a solution is a system which you now claim is based on the idea of ignoring the multiplayer aspect? Sure. Alright. Can't argue with that. > To me it's equally illogical to apply a ratingsystem designed for games > with only two sides to a multiplayer game, especially if such a game > supports ordered endresults. Then why ignore that ordering? The current system does not. > I just happen to agree with the new rules in more than one aspect. It > does, after all, correspond more closely to the Object of the Game. How does a system which treats a non-win 2VP as equivalent to 0VP more closely correspond to the object of the game (accumulation of the most VPs)? > Actually I didn't suggest that 1VP ties with 0VP. They both lose to > the one with a Game Win. No rating-comparision done between those > who lose. Ah. The players are only playing against the winner, not each other. > The only time you use the algorithms for a tie is when the table > doesn't generate a Game Win - which makes a 2-2-1-0-0 table tie, > and I can agree that a lot of players will have a hard time accepting > that 2VP ties with 0VP in this case, but conceptually you bring out > the rules for a tie when 'noone won'. So a game between two grand masters and two newbies in which the two grandmasters end up tied at 2 (after ousting everyone else) means that the two grand masters' ratings plumet and the newbies all go up (based on the concept of tieing all contestants irrespective of performance). > I'm getting used to that getting 1 VP or 0 VP is more a matter of > seating than skill (rush-decks going upstream in order to get their > backs clean when going forward, combat-walls squeezing predator to > death while using a forward motion comparabe to *ahem* a wall). ? Either the game is skill-based (and ratings should be kept) or it is random (based, sya, on seating over skill). > I have yet to see a High Visibility Deck making it into the finals > that actually isn't taken care of comparatively soon there. That > hardly constitutes as it's predator playing better. ? > However, and more important, the game seems to be designed as an all > or nothing game. Tournament rules have changed in this direction, and > that gave me the idea for this suggestion. Tournament Rules still use VPs quite prominently. [ quoted text not captured ]

The Lasombra wrote: > > For the purposes of maintaining rank, it is better to qualify for > the final and not play than it is to qualify for the final and do > anything other than win. Really? I would think scoring 2 victory points but getting ousted would still generally be more beneficial to your rating than not playing. Other than that, yea, not playing is always better than getting ousted without scoring a point - whether it's the final or any other rated game. Fred

Roger Carhult wrote: [ quoted text not captured ] Heh ;) Seems I'm too unclear. 3-0-0 means 3VP one round - and one Game Win, zero VP the two other rounds. 2-2-2 is the typical (perfect) gathering of VP during three rounds for the 100% defensive wall deck. Sten During

"Todd Banister" <vtesr...@white-wolf.com> wrote in message news:<VgY28.566$oT3....@eagle.america.net>... [ quoted text not captured ] Thanks Todd, I understand now why I went down in the rankings, I just feel that it's a bitter pill having done consistently 'well' over the last 4 years or so. The problem I have with the system is the definition of 'well'. Getting in finals practically every time in my mind would constitute 'well', unfortunately for me the ranking system does not agree. It frustrates me immensly that players who have had many bad tournaments can have one decent one (getting 6th place for instance) and be above me in the rankings. To me this is not a fair reflection of a players skill or achievements. Pride hurts, (but hey f#ck pride). I don't have the ability to come up with new equations to propose a change and I don't have an answer (I would say solution but by others may not agree that there is a problem), I suppose I'll have to take my medicine, continue to get incredibly drunk at tourney's and be content with the fact that I'll probably come third again. While I'm here Hi to all the Frenchies, Francois and Reyda in particular.

LSJ wrote: > Sten During wrote: My english is apparently not what it's supposed to be. I'll try clearing up below ;) > >>LSJ wrote: >> >>>Sten During wrote: >>> >>>>It happened because the rating-system is thoroughly broken - or more >>>>correctly, it's not made for handling a multiplayer environment. >>>> >>>> >>>It works at least as well as the "solution" you propose, which largely >>>ignores the multiplayer aspect. >>> >>That was kind of the idea behind my suggestion ;) >> > > ? > > You complain that the current system is "broken" and equate that with > the theory (however flawed) that it isn't made for handling a multiplayer > environment, then claim that your idea of a solution is a system which > you now claim is based on the idea of ignoring the multiplayer aspect? > > Sure. Alright. Can't argue with that. My 'claim' is that as the ratingsystem is designed for games with only two sides then one way of applying it to a multiplayer game is to only award the winner. More on it below. > > >>To me it's equally illogical to apply a ratingsystem designed for games >>with only two sides to a multiplayer game, especially if such a game >>supports ordered endresults. >> > > Then why ignore that ordering? > The current system does not. My suggestion ignores the ordering for reasons described below. > > >>I just happen to agree with the new rules in more than one aspect. It >>does, after all, correspond more closely to the Object of the Game. >> > > How does a system which treats a non-win 2VP as equivalent to 0VP more > closely correspond to the object of the game (accumulation of the most VPs)? Again I must have been unclear. I was referring to the object of the game as written in the rules book. It says that the goal is to be the single winner, at the least that is how I interpret "the winner is the player with the most victory points at the end of the game". This interpretation may be incorrect, but taken together with the new Game Win rules I was strengthened in my belief that it was a correct interpretation. So we're down to if my interpretation is a valid one or not. If so, then I hope that my following reasoning is solid. As all players have to aim at getting most VP at a table, not sharing that amount with any other player, then, assuming that a typical round is played at a five player table, follows that gaining a Game Win in more than one fifth of the rounds you participate in means that you're a better than average player. By more than one fifth I'm excluding all tables that doesn't generate a Game Win. > > >>Actually I didn't suggest that 1VP ties with 0VP. They both lose to >>the one with a Game Win. No rating-comparision done between those >>who lose. >> > > Ah. The players are only playing against the winner, not each other. The players are competing against each other but only one can win. The winner takes it all. No reward for a good performance that still wasn't good enough. In a round that doesn't generate a Game Win then either 'noone won' or 'everyone lost'. It seems logical to handle rating differently compared to when you have a Game Win in this case, even within the scope of my proposition. One such way is to punish those who entered the round with a higher rating for not being able to force a Game Win. However, I did read your protest below. Addmittedly different but with a solid logic of it's own, and as stated earlier, based on my interpretation of the object of the game, an interpretation that may be incorrect. > > >>The only time you use the algorithms for a tie is when the table >>doesn't generate a Game Win - which makes a 2-2-1-0-0 table tie, >>and I can agree that a lot of players will have a hard time accepting >>that 2VP ties with 0VP in this case, but conceptually you bring out >>the rules for a tie when 'noone won'. >> > > So a game between two grand masters and two newbies in which the > two grandmasters end up tied at 2 (after ousting everyone else) > means that the two grand masters' ratings plumet and the newbies > all go up (based on the concept of tieing all contestants irrespective > of performance). Yes, for reasons stated above. If this seems too rash, then just throw away the changing of rating based on ties, but I kind of like the idea of punishing 'those who should have known better' ;) Anyway, I definitely see why players would become disappointed for reasons you describe above, so, to make players feel that the system (my proposition) seems more fair, just don't implement the handling of ties at all. > > >>I'm getting used to that getting 1 VP or 0 VP is more a matter of >>seating than skill (rush-decks going upstream in order to get their >>backs clean when going forward, combat-walls squeezing predator to >>death while using a forward motion comparabe to *ahem* a wall). >> > > ? > Either the game is skill-based (and ratings should be kept) or it > is random (based, sya, on seating over skill). > > >>I have yet to see a High Visibility Deck making it into the finals >>that actually isn't taken care of comparatively soon there. That >>hardly constitutes as it's predator playing better. >> > > ? The two examples above were an attempt to point at situations where a single players skill doesn't help much. I think that most of us have had the day ruined by a purely defensive predator, or in worst case a newcomer with an aggressive deck who is not used to the prey/predator style of this game and runs upstream in the hope of surviving longer, rather than hurting his/her prey in order to grab a VP. > > >>However, and more important, the game seems to be designed as an all >>or nothing game. Tournament rules have changed in this direction, and >>that gave me the idea for this suggestion. >> > > Tournament Rules still use VPs quite prominently. > > It does, in the same way as soccer and hockey uses difference in scored goals compared to goals let in (what is the correct english?). I'll try to be more clear with this example. In a hockeygame the team who scores (VP) most wins (gets on Game Win), and the game may end up as a tie (half a Game Win to each team (yes, I'm aware that most leagues use zero, one or two points for loss, tie and win)). These scores minus the amount of scores a team has allowed opposing teams to do is totalled in order to calculate tiebreakers. At any time during a league when two teams have the same amount of Game Wins then difference in scores (in this example effectively VP) is used as tiebreaker. If still tied then, depending on league, for example most scores made is used as yet another tiebreaker. This could with some imagination be compared to TP. All in all, it's my honest belief that a ratingsystem based only on Game Wins is to be preferred. As you made it rather obvious that players will not like being punished during a tie then I'd like to propose a slighly different system. Game Win(ner) only receives rating from each of the losing players. If no Game Win then no recalculating of rating at all. Sten During

Sten During wrote: > > I think that most of us have had the day ruined by a purely > defensive predator, or in worst case a newcomer with an aggressive Sorry, 'defensive predator' should be 'defensive prey'. Sten During

I am not so into this rating stuff, but i occasionally play a game of hearts (ie multiplayer) on www.playsite.com . They use a system with rankingpoints there that accounts for different startingratings and a ranking within the game from first to last...If anyone is interested, check it out... Wouter

In message <a9abcc59.02012...@posting.google.com>, Barny Baker <barny...@synergygroup.co.uk> writes: >The problem I have with the system is the definition of 'well'. >Getting in finals practically every time in my mind would constitute >'well', unfortunately for me the ranking system does not agree. The ranking system basically pits you into "one-on-one" fights with everyone on the table. If you always get most VPs, your rating will always go up. (Because you will win every "one-on-one" match up.) However, under the old system of VPs counting for the final, you could, say, sweep one table where you didn't get very much out of it, lose heavily on the following two tables against four newbies (e.g. a 2/1/1/1/0 split, with you being the zero), still get into the final with 5 VPs from that first table and lose again against everyone. -- James Coupe "Never give in to them," she whispered. "No matter what PGP 0x5D623D5D EB they do or how important you feel it is to get their accep- D690ECD7A1FB457CA21 tance. Never kill part of yourself for them. Because other 3D7E668C3695D623D5D people will notice that part is missing before you do."

Sten During wrote: > LSJ wrote: > I was referring to the object of the game as written in the rules book. > It says that the goal is to be the single winner, at the least that is > how I interpret "the winner is the player with the most victory points > at the end of the game". This interpretation may be incorrect, but > taken together with the new Game Win rules I was strengthened in my > belief that it was a correct interpretation. The current system rewards a Game Win exactly the same as your proposed system - a win vs. every other opponent. This matches the new tournament rules as well. The current system also ranks people who have tied on GWs (only possible for those tied with 0 GW when restricted to a single game) based on their VPs. This matches the current tournament rules as well. > So we're down to if my interpretation is a valid one or not. If so, > then I hope that my following reasoning is solid. > > As all players have to aim at getting most VP at a table, not sharing > that amount with any other player, then, assuming that a typical round > is played at a five player table, follows that gaining a Game Win > in more than one fifth of the rounds you participate in means that > you're a better than average player. By more than one fifth I'm > excluding all tables that doesn't generate a Game Win. By getting more VPs than a given player, it follows that you're better than that given other player. The ELO rating system doesn't compare you to the average player - it compares you to your opponent. > The players are competing against each other but only one can win. > The winner takes it all. No reward for a good performance that > still wasn't good enough. This is different than the tournament rules. VPs are still tracked - not just GWs (winner take all). > > So a game between two grand masters and two newbies in which the > > two grandmasters end up tied at 2 (after ousting everyone else) > > means that the two grand masters' ratings plumet and the newbies > > all go up (based on the concept of tieing all contestants irrespective > > of performance). > > Yes, for reasons stated above. > > If this seems too rash, then just throw away the changing of rating > > based on ties, but I kind of like the idea of punishing 'those who > should have known better' ;) > > Anyway, I definitely see why players would become disappointed > for reasons you describe above, so, to make players feel that > the system (my proposition) seems more fair, just don't > implement the handling of ties at all. Then you get the opposite problem - if the 2 newbies really shine and oust the two grand masters but time is called before they can duke it out for the "winner take all" GW, then they receive no recognition for their outstanding game play (1.5-1.5-0-0). The current system handles both situations appropriately (modest increase or at least no decline for the "winning" grand masters; appropriate increase for "winning" newbies). > The two examples above were an attempt to point at situations where > a single players skill doesn't help much. > I think that most of us have had the day ruined by a purely > defensive predator, or in worst case a newcomer with an aggressive > deck who is not used to the prey/predator style of this game and > runs upstream in the hope of surviving longer, rather than > hurting his/her prey in order to grab a VP. This is an argument for doing away with tournaments altogether (since according to the argument, the game's multi-player aspect removes its value as a test of player skill). It is not an argument for your proposed rating system as much as it is an argument against either of the rating systems. I think your arguments may support a different rating system, but one that is not likely to be implemented due to the increased bookkeeping required: NON-OFFICIAL ALTERNATE RATINGS SYSTEM Victory Points are tracked as a "win" for the recipient against a "loss" by the player being ousted. The standard Elo formula is applied only to those pairs. Example: A B C and D are seated in that order around a table. A ousts B then ousts C and then ousts D. A receives a "win" against each other player and each other player receives a "loss" against A. No other comparisons are made. (This matches your proposal above, but only for sweeps). Example: A ousts B then C. Then D ousts A. Score: A > B A > C D > A Where "X > Y" means X and Y are considered "paired" for Elo rating, with X receiving the win and Y receiving the loss. The system would be trivially extended for time-out conditions: All players left standing (not ousted) when time is called are rated pairwise against each other as tied. Example: A ousts B and then time is called. A > B A : C A : D C : D You could then track a withdrawal as a tie against every other player still standing at the time of withdrawal, or just ignore it. You would also have to decide whether to apply the Elo ratings changes instantly at each VP, or to track the "delta VP" based on game-start ratings at each VP and apply the deltas at the end (the way the current system does). The final piece of the puzzle would deal with VP-stealing (via, for example, Life Boon). You could track that only from the person being ousted to the recipient, or you could track it through the player who is the victim of the Life Boon. I think the latter is more appropriate, but more cumbersome as well. Anyway, this whole system adds complication to the rating system in that specific VP "movement" (as seen from the model where each player brings a VP to the table to be given away when she is ousted.) It more adequately achieves the result you're arguing toward, I think. [ quoted text not captured ]

"Barny Baker" <barny...@synergygroup.co.uk> wrote in message > While I'm here Hi to all the Frenchies, Francois and Reyda in > particular. Thanks for that, Barns....nah i don't feel left out...not at all...*mumbles, grumbles* :) TPOTH Koldun [ quoted text not captured ]

Barny Baker wrote: > > "Todd Banister" <vtesr...@white-wolf.com> wrote in message news:<VgY28.566$oT3....@eagle.america.net>... [ quoted text not captured ] Actually, I don't care about pride either. But you raise an excellent question about this system: Does it swing your ratings up and down far too wildly? It looks to me like it absolutely does. Here's a guy with 50+ rated games to his experience and ratings are going up and by 40-60 points IN A SINGLE GAME!!! Considering that the effective range of ratings is 3000 plus or minus 300 (maybe 350 on the outside), that's a hell of a swing. I always had the idea this system wasn't real good. But if a guy with a lot of history like that can run his rating from 3288 down to 3122 in one tournament - and he was in the finals at that (!!!) - these things are basically worthless. This is essentially a reflection of how you've done in the last two or three tournaments, not how you've done over time. I've never sat down and looked at the system but I think I'm safe in saying that it needs more stability to a players' ratings who have a lot of games under their belt. A *LOT* more. Fred

Frederick Scott wrote: > Actually, I don't care about pride either. But you raise an excellent > question about this system: Does it swing your ratings up and down far > too wildly? > > It looks to me like it absolutely does. Here's a guy with 50+ > rated games to his experience and ratings are going up and by 40-60 > points IN A SINGLE GAME!!! Considering that the effective range of > ratings is 3000 plus or minus 300 (maybe 350 on the outside), that's > a hell of a swing. Elo ratings do not use number of games in the equation. 1 game or 50+, your rating is your rating. Only your rating (and the amount by which it differs from your opponent's) is used in the Elo calculation. > I always had the idea this system wasn't real good. But if a guy with > a lot of history like that can run his rating from 3288 down to 3122 in > one tournament - and he was in the finals at that (!!!) - these things > are basically worthless. This is essentially a reflection of how you've > done in the last two or three tournaments, not how you've done over > time. The system works by assuming that a player with a high rating should "win" vs. a player with a low rating. When that doesn't occur (the lower-rated player wins), then the rating is adjusted accordingly (by a larger amount than if the player loses to a higher-rated player). > I've never sat down and looked at the system but I think I'm safe in > saying that it needs more stability to a players' ratings who have a > lot of games under their belt. A *LOT* more. Constructive suggestions are welcome, although the "stability" of long-time player's ratings isn't fundamental to the theory - its desirability isn't necessarily a given. The Elo system has withstood scrutiny for quite some time. [ quoted text not captured ]

LSJ wrote: > > Frederick Scott wrote: > > I always had the idea this system wasn't real good. But if a guy with > > a lot of history like that can run his rating from 3288 down to 3122 in > > one tournament - and he was in the finals at that (!!!) - these things > > are basically worthless. This is essentially a reflection of how you've > > done in the last two or three tournaments, not how you've done over > > time. > > The system works by assuming that a player with a high rating should > "win" vs. a player with a low rating. When that doesn't occur (the > lower-rated player wins), then the rating is adjusted accordingly (by > a larger amount than if the player loses to a higher-rated player). I assume there's also an adjustment just for winning. That is, if a game is held between five players with identical rankings, the uneven results will cause their ratings to change. It's not just a question of adjusting ratings when a higher rated player is defeated by a lower rated player. > > I've never sat down and looked at the system but I think I'm safe in > > saying that it needs more stability to a players' ratings who have a > > lot of games under their belt. A *LOT* more. > > Constructive suggestions are welcome, although the "stability" of > long-time player's ratings isn't fundamental to the theory - its > desirability isn't necessarily a given. The Elo system has withstood > scrutiny for quite some time. Well, the obvious constructive suggestion would be to tune down the effect that each single game has in comparison to the previous rating. You say there's no discernment by the ratings calculations based on how many previous games a player has played. That part doesn't alarm me so much. (The only reason I made reference to the number of games was to make it clear that the ratings swings weren't due to the player in question having only a few games in his history. In some ratings systems, that might be a justification for wild swings. I was just pointing out that it couldn't be the cause of this situation.) If that is so, however, that means that a player's current rating at any moment reflects his most recent performance to a much greater degree than his earlier history. That's fine. All we're arguing about is how *much* his most recent games affect his current rating. From the calculation results Todd Banister posted from Barney Baker's history, it appears to me that a player's ratings is so heavily influence by his last one or two tournaments as to make the whole system essentially worthless. I suppose it's interesting that a player has placed well in his last couple of tournaments, but performance can be so streaky that it hardly tells me much about how good he is. Is there some control on how much the last game is weighted vs. the player's prior rating in each calculation? I would expect there to be some kind of coefficient that allows those who use the system to specify this preference. If so, that's what I would monkey with to make it work better. Fred What

> The system works by assuming that a player with a high rating should > "win" vs. a player with a low rating. When that doesn't occur (the > lower-rated player wins), then the rating is adjusted accordingly (by > a larger amount than if the player loses to a higher-rated player). The things that it doesn't take into consideration are the meta game and the fact that it is a multi-player game. It's not just the person to the left they that are playing against, but also the whole table. If you sit down with a nasty deck, odds are the rest of the table is going to do what they can to shut you down. So your rating can go down and your prey, the one who you are assumed to win against, may not have had anything to do with your downfall... Sorrow --- "I am Jack's wasted life." - Narrator

Frederick Scott wrote: > > LSJ wrote: > > > > Frederick Scott wrote: > > > I always had the idea this system wasn't real good. But if a guy with > > > a lot of history like that can run his rating from 3288 down to 3122 in > > > one tournament - and he was in the finals at that (!!!) - these things > > > are basically worthless. This is essentially a reflection of how you've > > > done in the last two or three tournaments, not how you've done over > > > time. > > > > The system works by assuming that a player with a high rating should > > "win" vs. a player with a low rating. When that doesn't occur (the > > lower-rated player wins), then the rating is adjusted accordingly (by > > a larger amount than if the player loses to a higher-rated player). > > I assume there's also an adjustment just for winning. That is, if a game > is held between five players with identical rankings, the uneven results > will cause their ratings to change. It's not just a question of adjusting > ratings when a higher rated player is defeated by a lower rated player. Yes. The whole thing is explained in the VEKN rules. Appendix A. [ quoted text not captured ] Conversely, a player with a truly high rating can only achieve that result by a long winning streak. What you may be witnessing is a symptom that the game is more dependent on seating position than on player skill, and thus the player's ratings tend to oscillate around the baseline as a function of seating (as proposed by other posters recently). That the well-established Elo system fails to live up to some people's expectations is a good argument for that case. > Is there some control on how much the last game is weighted vs. the player's > prior rating in each calculation? I would expect there to be some kind of > coefficient that allows those who use the system to specify this preference. > If so, that's what I would monkey with to make it work better. No. [ quoted text not captured ]

[ quoted text not captured ] The current system has no particular emphasis on "prey". And yes, the multi-player property is (still) problematic. [ quoted text not captured ]

[ quoted text not captured ] ... unless you're referring to the K factor of the particular event, currently set to 32 for all VEKN events (a player's rating cannot change by more than 32 points in any given pairing). [ quoted text not captured ]

[ quoted text not captured ] I'm not sure what this means. If a "given pairing" corresponds to a single game, clearly that's not true. As a counter-example, according to Todd's post, Barney's rating changes from 3301 to 3240 (61 points) between rounds one and two in his December 9th event. Fred

LSJ wrote: > > Frederick Scott wrote: > > > From the calculation results Todd Banister posted from Barney Baker's history, > > it appears to me that a player's ratings is so heavily influence by his last > > one or two tournaments as to make the whole system essentially worthless. I > > suppose it's interesting that a player has placed well in his last couple of > > tournaments, but performance can be so streaky that it hardly tells me much > > about how good he is. > > Conversely, a player with a truly high rating can only achieve that result > by a long winning streak. > > What you may be witnessing is a symptom that the game is more dependent > on seating position than on player skill, and thus the player's ratings > tend to oscillate around the baseline as a function of seating (as > proposed by other posters recently). > > That the well-established Elo system fails to live up to some people's > expectations is a good argument for that case. If I understand this correctly, I don't think it necessarily points at seating position as the problem in rating players. Seating position may or may not be the culprit, it's hard to tell. In my experience, seating position can be a big factor in isolated games. It's just that there are other luck and coincidence factors at work as well and it's hard to tell which ones are more important and to what degree. Anyway, to get to the point, it seems like the core of the problem is that this system has difficulty tracking player quality in game which vary greatly from one contest to the next due to other factors besides skill. In chess, ability from one player to the next varies greatly and skill is by far the largest predictor of how a contest will turn out, arguably the only one (except perhaps the determination of who moves first). In Jyhad, any number of luck factors will make it difficult to predict a winner even when one player is significantly more skilled than another player. Clearly, the Elo system is too sensitive to random variations in results like this to make it very useful for our purposes. I can't suggest anything right off the bat, but it would be good to have a system that buffers recent winning and losing streaks better than this one does. Fred

Frederick Scott wrote: > LSJ wrote: > > ... unless you're referring to the K factor of the particular event, currently > > set to 32 for all VEKN events (a player's rating cannot change by more than > > 32 points in any given pairing). > > I'm not sure what this means. If a "given pairing" corresponds to a single game, > clearly that's not true. As a counter-example, according to Todd's post, > Barney's rating changes from 3301 to 3240 (61 points) between rounds one and two > in his December 9th event. A "pairing" involves exactly two players (not a "grouping"). At a five player table, a given player faces 4 pairings - one with each of the other players. See Appendix A of the VEKN rules. [ quoted text not captured ]

"Pierre Rebstock" <fo...@hotmail.com> wrote in message news:<2f590ad96064a24c255...@mygate.mailgate.org>... > "Barny Baker" <barny...@synergygroup.co.uk> wrote in message > > > While I'm here Hi to all the Frenchies, Francois and Reyda in > > particular. > > Thanks for that, Barns....nah i don't feel left out...not at > all...*mumbles, grumbles* > :) > > TPOTH Koldun Mr Rebstock, everyone knows you're British now.

>Constructive suggestions are welcome, although the "stability" of >long-time player's ratings isn't fundamental to the theory - its >desirability isn't necessarily a given. The Elo system has withstood >scrutiny for quite some time. Regarding a different system, what comes to mind is golf (or tennis or Magic or whatever). Players get ranked in terms of money. 'k, that won't work. Players get points for winning. Better the result, the more points. Bigger the event, the more points. This has been considered before, of course. Complaint that often comes up is that someone could just play a lot of events and rack up the points. Another problem is determining how much events should be worth. But, there is a reason those sports come to mind rather than any number of point accumulation systems in place for CCGs. Theirs have durations. Money leader *last year*? Points leader *last year*? ATP ranks based on results in the last year. Also, for the other issue, even Magic has figured out how to give out pro tour points, though they are relatively scarce - most Magic players have none. I could see some attempt, if people really thought the current system was pointless and they thought any system was somehow important, to come up with a formula for alloting points to a tournament, alloting points to the top players, and keeping track of when points were accumulated (which year, last 12 months, whatever).

LSJ <vte...@white-wolf.com> wrote in message news:<3C4D8AD3...@white-wolf.com>... [ quoted text not captured ] That doesn't make it infallible. By you rational then the ratings are not a definitive answer to who is the best player/most skilled player/whatever. It is a measure of current form, a very very short-term measure. As short as your last tournament and practically nothing before. When it is better to come 6th, or better to win one tournament having done nothing in your previous few than to consistantly reach finals and come 2nd/3rd or wherever surely there is something wrong.

Barny Baker wrote: > LSJ <vte...@white-wolf.com> wrote: > > Constructive suggestions are welcome, although the "stability" of > > long-time player's ratings isn't fundamental to the theory - its > > desirability isn't necessarily a given. The Elo system has withstood > > scrutiny for quite some time. > > That doesn't make it infallible. True. See the welcome given above. > By you rational then the ratings are not a definitive answer to who is > the best player/most skilled player/whatever. It is a measure of > current form, a very very short-term measure. As short as your last > tournament and practically nothing before. It's not my rationale. I'm just restaing other posters' positions, as I kept saying. At any rate, no one has ever claimed that the ratings of VTES are a definitive answer to who is the best [...]. > When it is better to come 6th, or better to win one tournament having > done nothing in your previous few than to consistantly reach finals > and come 2nd/3rd or wherever surely there is something wrong. In any zero-sum formulation it is better not to play than to play and lose. This is inherent/funadamental to the zero-sum aspect. Coming in 2nd in the finals is typically better than coming in 6th and not making it to the finals. Check it and see. Coming in 3rd should break even statistically half the time with not playing at all (although the fact that it is a finals should move that a bit toward the "should play" range, given the assumption that the other players' ratings will also be "temporarily inflated" due to their own recent wins). [ quoted text not captured ]

> The current system has no particular emphasis on "prey". Really? Then please explain the statement you made as follows: "The system works by assuming that a player with a high rating should "win" vs. a player with a low rating. When that doesn't occur (the lower-rated player wins), then the rating is adjusted accordingly (by a larger amount than if the player loses to a higher-rated player)." Perhaps not your prey, but it has some emphesis on the player who you are supposed to win against, be it your prey or predator. In any case, it's not just that person you have to "win against". > And yes, the multi-player property is (still) problematic. Ok, since this is a (seemingly) well known fact, are other methods of rating generation being looked into at all? Or are the powers that be content to continue on with a system that is known to be really unfair at worst and problematic at best? Sorrow --- "This is my costume. I'm a homicidal maniac. They look just like everyone else." -- Wednesday Addams

Sorrow wrote: > > > The current system has no particular emphasis on "prey". > > Really? Then please explain the statement you made as follows: > > "The system works by assuming that a player with a high rating should > "win" vs. a player with a low rating. When that doesn't occur (the > lower-rated player wins), then the rating is adjusted accordingly (by > a larger amount than if the player loses to a higher-rated player)." > > Perhaps not your prey, but it has some emphesis on the player who > you are supposed to win against, be it your prey or predator. In any > case, it's not just that person you have to "win against". As you say - no emphasis on prey. You "lose" to a given other player at your table if and only if that player (regardless of predator/prey relationships) scores more VPs than you do. Again, see Appendix A of the VEKN rules. This is all there. > > And yes, the multi-player property is (still) problematic. > > Ok, since this is a (seemingly) well known fact, are other methods > of rating generation being looked into at all? Or are the powers > that be content to continue on with a system that is known to be > really unfair at worst and problematic at best? I didn't say the system was problematic. The multi-player nature of the *game* is problematic. At any rate, I haven't seen any viable options presented so far. Again, constructive input is welcome. [ quoted text not captured ]

LSJ <vte...@white-wolf.com> wrote in message news:<3C4DC33F...@white-wolf.com>... > Coming in 3rd should break even statistically half the time with not > playing at all. Yikes. That's 3rd in the tournament, right? That seems off. If I show up at a tournament with, say 20, reasonably competitive players, it seems like 3rd (or 4th or 5th) should be considered a good performance for rating purposes and give you a bit of a boost. The current system really seems to reward come-from-nowhere players who beat a high ranking player or two at one or two tournaments. And seems to really punish solid players who go to a lot of tournaments, do consistently well, but don't necessarily win every tournament. In Chess a "Grand Master" is going to beat a good player 99% percent of the time. Its all skill and there are no random elements. In V:TES there are plenty of random elements. A "Grand Master" is still probably only going to win the table, maybe 60-65% percent of the time. 2 table wins in the prelims is impressive. 3 rare. And if Kasparov loses 4 staight games to unranked players you'd expect a big drop in his rating. A high ranked V:TES player who gets ousted with 10 minutes left in the round at a table of lesser-ranked players, who go to time, has just lost to 4 players, and his rating will plumet - as we see with Master Baker numbers from the 12/9 tournament. Yada, yada, yada, so what's would be a better system? Don't know. -Ben Swainbank

bswai...@hotmail.com (Ben Swainbank) wrote in message news:<b20d2553.0201...@posting.google.com>... > LSJ <vte...@white-wolf.com> wrote in message news:<3C4DC33F...@white-wolf.com>... > > Coming in 3rd should break even statistically half the time with not > > playing at all. > Yikes. That's 3rd in the tournament, right? That seems off. If I show > up at a tournament with, say 20, reasonably competitive players, it > seems like 3rd (or 4th or 5th) should be considered a good performance > for rating purposes and give you a bit of a boost. > The current system really seems to reward come-from-nowhere players > who beat a high ranking player or two at one or two tournaments. And > seems to really punish solid players who go to a lot of tournaments, > do consistently well, but don't necessarily win every tournament. (There were some discussion about the random and multi-player nature of this game, not quoted here) Well. How about a system that does a zero-sum calculation on all of the participants in a tournament - it would make those getting more VPs and GW better in terms of ranking. This is of course is now hard because we have "Game Wins" and "Victory Points" so how do you make a ranking of a players in a tournament. But this is what I am saying: Suppose we CAN make a decent ranking of tournament participant in terms of their play in the tournament (using VP's GW's or combination of both): 1. A.A (winner) 2. J.D 3. 4. 5. 6. 7. ... ... 31. 32. 33. 34. G.A A.A was the best player in the tournament and J.D came second etc. The participants in the tourney would have their ratings updated according to their position in a given tournament and compared to some metric of skill in the tournament and the number of participants. (for instance the average of rating in the tournament suppose that is like 3023). So you could say A.A won a tournament with average raiting 3023 J.D with 34 participants. Then suppose we have player that had 3023 rating before the tournament and he came to "shared" 17. position. He is just on the better half of the tournament and so his rating should go up just a bit like +1 etc. If this person had won the tournament his rating should have gone up like +100. Well anyways I hope you understand, what I am aiming at - A scoring system in you check how well you managed a tournament, not how unlucky you were in some situation like: 1st round you get 1VP a fellow wins the table with elo 2000 (lots of - for you), but lots of other good players 2nd round you get 2VP a fellow wins the table with elo 2000 (lots of - for you), but lots of other good playrs 3rd round you get 3VP, GW in a table with avg elo 2800 (just a small + for you) and then you get to finals and get spanked there you come up third and some fellow wins the tournament. What you get ? You get LOSE a lot of points even as you played quite good because you lose tables where a person with poor rating won. If the change in rating was tied to the tournament it self you would gain points not lose them by coming up 3rd. So would these poor players (they would gain even more) so it's not a case in where their "rights" to gain points is stomped on. So even good players would gain points (until their points get very high, when it is expected - not rewarded) by getting in the first fifth of the tournament by some ranking method. Jukka

LSJ wrote: > At any rate, no one has ever claimed that the ratings of VTES are > a definitive answer to who is the best [...]. I don't think anyone was ever laboring under that illusion, no. I guess that little rundown of Barney's rating changes really opened my eyes about just how bad they really are, though. Fred

On 22 Jan 2002 17:43:26 -0800, Ben Swainbank wrote: >A high ranked V:TES player who gets ousted with 10 minutes left in the >round at a table of lesser-ranked players, who go to time, has just >lost to 4 players, and his rating will plumet - as we see with Master >Baker numbers from the 12/9 tournament. > >Yada, yada, yada, so what's would be a better system? It's seems wrong to rate a player against all other players in the game. If you have a lower-rated player sitting across the table from you then the system expects you to beat him. But his weakness makes your predator and prey stronger and so your chances of doing well are decreased rather than increased. If one wanted to stick with the ELO system then it might be better if a player were considered to have beaten another only when he ousts him. But I'd prefer a different paradigm - baseball stats. In those you have a series of numbers and averages which give you a rounded picture of a player's performance - home runs, on-base percentage, earned run average, etc. And these are shown both for the current season and as lifetime totals. They don't factor in the quality of the opposition because that would be too complex - you're not facing a single player but a whole team of them. Andrew

Ben Swainbank wrote: > > LSJ <vte...@white-wolf.com> wrote in message news:<3C4DC33F...@white-wolf.com>... > > > Coming in 3rd should break even statistically half the time with not > > playing at all. > > Yikes. That's 3rd in the tournament, right? That seems off. If I show No. That's third in a game. [ quoted text not captured ]

"Andrew S. Davidson" wrote: > If one wanted to stick with the ELO system then it might be better if > a player were considered to have beaten another only when he ousts > him. ... with the increase in bookkeeping. See my earlier post which suggested exactly this. [ quoted text not captured ]

LSJ wrote: > >The current system rewards a Game Win exactly the same as your proposed >system - a win vs. every other opponent. This matches the new tournament >rules as well. > >The current system also ranks people who have tied on GWs (only possible >for those tied with 0 GW when restricted to a single game) based on >their VPs. This matches the current tournament rules as well. > Agreed, but for reasons described below I believe that swinging ratinghandling in this direction as well (more Game Win centered) would be beneficial. > > > >>So we're down to if my interpretation is a valid one or not. If so, >>then I hope that my following reasoning is solid. >> >>As all players have to aim at getting most VP at a table, not sharing >>that amount with any other player, then, assuming that a typical round >>is played at a five player table, follows that gaining a Game Win >>in more than one fifth of the rounds you participate in means that >>you're a better than average player. By more than one fifth I'm >>excluding all tables that doesn't generate a Game Win. >> > >By getting more VPs than a given player, it follows that you're better >than that given other player. > Here I actually differ in opinion. VTES is a multiplayer game, including the room for randomness (in order of seating, a players mood etc) that is, and should be, a part of the game. There is also the aspect of trying to optimize ones gaming with the new Game Wins rules in mind, ie if I build and play in such a way that I tend to either get exactly 3 VP or exactly 0 VP, and assume that I actually manage to get those 3 VP during 22% of the games I participate in (in this theory all games are five player tables). Compare this with another player who manages to get exactly 1.5 VP at 100% of the games he participates in. With your reasoning above, then the latter player would have aproximately twice the skill as the former (1.5 * 1.0 compared to 3 * 0.22). However, assuming that a normal (future) tournament attracts 25 players (yes this is more that what is currently the average), then we can safely assume that the solid 1.5 VP player will never be seated at the finals table, but the former player ought to make it every tenth or so tournament (applying standard deviation isn't my strongest side) during which he actually grabs two Game Wins at the prelims. Now, as rating is a tournament only thing, I'll be so bold as to state that more or less the entire tournamentenvironment would hold the player who went to the finals every tenth tournament as the better player. I believe that ratings ought to mirror this, and with the two hypothetical players I described, the current system will clearly show that the player who by definition will never get to a final is still the better player. > > >The ELO rating system doesn't compare you to the average player - it >compares you to your opponent. > Agreed, and this might be the core of the problem with applying it to any myltiplayer game. > > >>The players are competing against each other but only one can win. >>The winner takes it all. No reward for a good performance that >>still wasn't good enough. >> > >This is different than the tournament rules. VPs are still tracked - >not just GWs (winner take all). > I know, my suggestion is after all only a suggestion - and different at it as well ;) <snipping pros and cons of handling ties> > > > >>The two examples above were an attempt to point at situations where >>a single players skill doesn't help much. >>I think that most of us have had the day ruined by a purely >>defensive predator, or in worst case a newcomer with an aggressive >>deck who is not used to the prey/predator style of this game and >>runs upstream in the hope of surviving longer, rather than >>hurting his/her prey in order to grab a VP. >> > >This is an argument for doing away with tournaments altogether (since >according to the argument, the game's multi-player aspect removes its >value as a test of player skill). It is not an argument for your >proposed rating system as much as it is an argument against either of >the rating systems. > Maybe, or maybe not. If I say that people in general says that those who frequently make it to the finals are good players, then I don't fear too much opposition. Memorizing all your end-positions at all tournaments when you end up among the five first, is definitely something that could be done. It would give you an immediate absolute value, and most players would agree upon such a list numbering good players. It would not be a rating system as such, but it would nontheless be considered as a valid way of valuing the skill of the best players (most of us would never even end up on the list, which is why this would make a very poor ratingsystem). [ quoted text not captured ] Both yes and no. I hope that I have adequately enough described my reasoning around why I don't believe that difference in VP between players who don't grab a Game Win actually holds much value. There is also another point, with my own doubts taken in mind. The rating system, ELO, usually is based on a number in the thirty-region, so I assume (probably wrongly) that it's exactly thirty. Two players with equal (plus minus fifteen) rating meet. The winner gains 15 in rating and the loser lose the same amount. In case of a tie nothing happens. Two players with very large difference (600 +) in rating meet. The better player wins. He gains 1 and loser loses 1. The better player loses. He loses 29 and opponent gains 29. In case of tie better player loses 15 and opponent gains 15. This is the solid base for the ELO system, but it's solid only within the scope of a two-sided game. With the current, or any ratingsystem that allows for internal comparision between all participants, numbers will swing too widly. At a five-player table where all player start with equal rating I'm ousted after which the game times out. I drop 15 * 4 in rating as all players get more VP than I did. Mr Grandmaster brings his feared cheeze (TM) deck to the Big Important Tournament. With his 4500 rating he's a walking legend which will cost him dearly today. During the three prelims the tables gang up on him, with both the judge as well as Mr Grandmaster agreeing to that the only way of being able to gain at least 0.5 VP at any table he's participating at is to oust him first. The deck in question also happens to be an argument for some kind or later rules changing. At each table his rating drop 29 * 4 for a grand rating loss of 29 * 12 during one tournament. During the next five consequtive tournaments he promptly sweeps each table he participates in making it a grand 20 tables in one row at which he grabs 5VP each. He gains 1 * 4 in rating from each table in order to enable him to regain 1 * 4 * 20 rating of what he lost earlier. My example is absurd, so I'll satisfy with the less absurd example of Mr Grandmaster running into bad mana at one table, dropping 116 in rating from that table. The tournamentsweep later earning him a place in history, etc, ect, will give him back 80 of the rating lost. In my book this is a bad thing, and a Game Win orientated rating system, with all the obvious problems that you have pointed out, will to a rather large extent counter this, and I think that countering the wild swings in rating is actually more important (if rating is important at all) than making sure that each player is rewarded for his performace compared to all other participants during one game. Sten During > >

bswai...@hotmail.com (Ben Swainbank) wrote in message news:<b20d2553.0201...@posting.google.com>... [ quoted text not captured ] My REAL view about this is that ratings DON'T MATTER AT ALL, and that when we play to improve our ratings we are unwise. i've dropped from 13th to 300th or so without, i think, any REAL change in my abilites or in the competition, AND I DON'T CARE!!! However, i think it's a fairly simple matter to fix the ratings so that they reward you for getting to the final and for winning tournaments [which i think are desirable aims]. i suggest we do it this way: Do the calculation in the normal way. After all the calculations are done, add 5 [or 4, for a 4-player final] points to the winner of the tournament 4 [or 3] to the person who comes second and so on. If only one person gets VPs in a 5-player final, add 5 points to that person's calculated rating, and 1 point to everyone else.

LSJ <vte...@white-wolf.com> wrote in message news:<3C4EA276...@white-wolf.com>... > Ben Swainbank wrote: > > > > LSJ <vte...@white-wolf.com> wrote in message news:<3C4DC33F...@white-wolf.com>... > > > > > Coming in 3rd should break even statistically half the time with not > > > playing at all. > > > > Yikes. That's 3rd in the tournament, right? That seems off. If I show > > No. That's third in a game. Sorry but it's third in a tournament, not only that, third or better in consecutive tournaments over a a long time, my worst positions over the last 2 years (an educated guess) have been 6th in the U.K. Nationals and 5th at The European qualifyer. If it was third in a game I wouldn't be labouring the point, but it is not and I am.

legbiter wrote: > My REAL view about this is that ratings DON'T MATTER AT ALL, and that > when we play to improve our ratings we are unwise. i've dropped from > 13th to 300th or so without, i think, any REAL change in my abilites > or in the competition, AND I DON'T CARE!!! Well, fine. But if we want to make that the "official" view of ratings then the logical thing to do would be to get rid of ratings. I'm not sure I see the point of keeping a sham ratings system around. > However, i think it's a fairly simple matter to fix the ratings so > that they reward you for getting to the final and for winning > tournaments [which i think are desirable aims]. > > i suggest we do it this way: > > Do the calculation in the normal way. > > After all the calculations are done, add > > 5 [or 4, for a 4-player final] points to the winner of the tournament > 4 [or 3] to the person who comes second > and so on. > > If only one person gets VPs in a 5-player final, add 5 points to that > person's calculated rating, and 1 point to everyone else. "Add?" Just "add"? I think you're missing the point about the ratings system being zero sum. If you just start adding points without going out and robbing them from someone else, then it starts to reward people merely for participating (i.e. you get the opportunity to be in the final). This opens up the system to a whole slew of implications that a closed zero sum game doesn't have: mainly, that being able to be in as many tournaments as possible becomes a ratings boon. (Even the players who don't get to and finals at all get rewarded, since they frequently get to play players who've been in the final, which is a ratings boon, too.) I say, "Well, sure he's got a good rating, but he comes from Los Angeles where they have a million tournaments." This is a problem for ratings because it means they don't actually rate people. There would also be a gradual ratings inflation as the average over time slowly increased from the 3000 that people start with. So now people start with the average number of points but next year, it will be somewhat less than average. 10 years from now it will likely be much less than average. Strange stuff. If you're going to go to a "rewards" system for rating people and give them their little gold stars for participating and winning things, let's do it blatantly like the original DCI rating system for Magic before they started using the current one: 10 points for showing up, double the points for advancing to each subsequent round. That was for a single elimination tournament. Our round-robin preliminaries would probably be more like 10 points for playing each round and 10 points for each victory point scored therein. 50 for going to the finals and 50 more for each victory point scored there. And at the end of the year, you total them, send the point leaders a plaque (and an invitation to the Continental Tournament of their choice) and wipe the points clean and start over. Fred

Frederick Scott wrote: >legbiter wrote: > >>My REAL view about this is that ratings DON'T MATTER AT ALL, and that >>when we play to improve our ratings we are unwise. i've dropped from >>13th to 300th or so without, i think, any REAL change in my abilites >>or in the competition, AND I DON'T CARE!!! >> > >Well, fine. But if we want to make that the "official" view of ratings >then the logical thing to do would be to get rid of ratings. I'm not >sure I see the point of keeping a sham ratings system around. > As long as (valid or not) there is some kind of emotional consensus that a ratingsystem somehow doesn't function well, then the obvious reaction will be that rating doesn't matter. Basically, if I don't feel that the rating mirrors a players achievements during a prolonged amount of time then I'll be less inclined to respect, and consequently, care about said rating. This statement is true no matter if the rating system used actually does mirror the reality to 100% or if the system actually is flawed. The only important thing will be what I believe to be the truth, especially if my belief (wrong or not) is echoed by others often enough. Chess uses the very same ELO rating. Active chess players have the utmost respect for this rating and the feeling when seated against an opponent with some 300+ higher rating is not an especially welcome one. You know that the coming game will be hell. It's worth noticing that ELO is officially considered as flawed when used for players with extremely high rating (2200+ at the late eighties, I don't know the limit today), and those players have an internal rating only used at grandmaster level. The thing that is broken with ELO in chess is that if you, newly introduced into the rating system, meet a player with an 1600 rating, will probably lose, but forcing a tie is definitely doable and winning is only considered highly unusual. If a player with 1900 rating beats a player with 2200 rating you're watching a first class international sensation. ELO is still lineear though. In chess you start at 1300 rating, and a player with an 1900 or so rating is most probably a local/regional champion. Anyway, ELO as a means of calculating ratings is not in itself broken. It's proven itself for several decades. I believe that ELO is probably the best system we can use at the moment. It's only a matter of what you choose to measure. >> >>After all the calculations are done, add >> >> > >"Add?" Just "add"? I think you're missing the point about the ratings >system being zero sum. If you just start adding points without going out > <snip> Breaks the system totally or at least forces a manual control of the total amount of allocated/ added rating in order to raise the starting rating for a new player - which will force a manual... It will no longer be rating, but it could be used as a means of calculating the achievements of a player during his/her entire history of play. This would of course also be an interesting number, but rating is supposed to give an idea about a players current strength - a little bit like announcing a sprinters five best times during the last year in advance of a big championship. Sten During

Sten During wrote: (Gad, I wish you'd adjust your news posting software so it wouldn't insert random line breaks around your posts. It's very irritating to read.) > > Anyway, ELO as a means of calculating ratings is not in itself broken. > It's proven itself for several decades. I believe that ELO is > probably the best system we can use at the moment. It's only a matter > of what you choose to measure. My issue with the current system is that it's far too weighted towards recent results in comparison to previous rankings. Jyhad frequently produces results which buck the trend in terms of skill - due to seating order, bad vs. good luck in the draw of the cards, metagame choices and so forth. As a result, skillful players can't run their ratings above a certain level before an unlucky defeat (or several) by someone ranked much lower chops them back down towards average. It may be that this can still be arranged through the Elo system. Having looked at the system, I would say that the problem - and what therefore likely needs changing - is is the calculation of how many points are won and lost by each pair of players based on the difference in their current rating. Right now, a player gets 3300 rating and he just gets kicked far too hard if he doesn't beat everyone else he plays, which he's never going to do. So no one can hold a high rating for long. The solution is to reduce the effect of the difference in ratings points has on the calculation of win probability, and therefore how many points get added if the high-rated player wins vs. how many get subtracted if he loses. Apparently, the ratio of 32 points as the coefficient to 400 as the divisor of the difference in ratings of the exponent of 10 in the overall divisor (just go look at appendix A if you really want to know(!)) is too high. I'd set the latter number to 1000 or so (a pure intuitive guess here; it could probably stand someone doing some calculations to see how it worked out) so that players' ratings could range a lot higher and lower before differences in ratings made it a lot harder for them to gain points (if high rated) or lose point (if low rated). It would still take time to run your rating very high (since the 32 doesn't change) or very low, but Jyhad being the type of game it is, I would argue that this is justified and how things have to be. Comments? Fred

>"Add?" Just "add"? I think you're missing the point about the ratings >system being zero sum. If you just start adding points without going out >and robbing them from someone else, then it starts to reward people merely >for participating (i.e. you get the opportunity to be in the final). This >opens up the system to a whole slew of implications that a closed zero >sum game doesn't have: mainly, that being able to be in as many tournaments >as possible becomes a ratings boon. (Even the players who don't get to >and finals at all get rewarded, since they frequently get to play players >who've been in the final, which is a ratings boon, too.) I say, "Well, >sure he's got a good rating, but he comes from Los Angeles where they have >a million tournaments." This is a problem for ratings because it means >they don't actually rate people. If the numbers are on the right scale for a cumulative point system, just divide the person's rating by their number of tournaments to get their EPA (earned points average or ego-boosting player accomplishments). I don't see anything relevant to a zero sum system. Is the system in place even zero sum in practice, anyway? If I wanted to jack up my rating, I could encourage newbies to play in tournaments, beat them, encourage different newbies, and beat them, and so on. Newbies essentially have ratings higher than their ability. Or, have low ranked players retire, so your competition is only worth a decent number of points. Or, have high ranked players eliminate their registry, so they aren't above you anymore. Actually, if eliminating your registry doesn't change everyone else's numbers, it changes the average. Another thing, while currently pointless and never likely to be of any value except psychologically, it's very easy to sandbag your rating to distort your (perceived) puissance. Win all the big tournaments and screw around at all the smaller, more common locals. Actually, I've seen this happen unintentionally. One of the best B5 players in the world is local. While he consistently did well at major tournaments, at locals he screwed around a lot because that's more fun for him. Anyway, until high ratings are rewarded (something I don't recommend), they only seem to exist for fun, i.e. discussions like this. I'm just amazed that Todd puts up with it all. Doing the ratings is a bitch, I'm sure. I do our Archons for local tournaments now, and I find doing those a bitch, nevermind the number of times I ask "shouldn't that *older* tournament he played in have changed what'shisname's rating which is still 3000?"

Curevei wrote: > I don't see anything relevant to a zero sum system. Is the system in place > even zero sum in practice, anyway? If I wanted to jack up my rating, I could > encourage newbies to play in tournaments, beat them, encourage different > newbies, and beat them, and so on. Newbies essentially have ratings higher > than their ability. Or, have low ranked players retire, so your competition is > only worth a decent number of points. Or, have high ranked players eliminate > their registry, so they aren't above you anymore. Actually, if eliminating > your registry doesn't change everyone else's numbers, it changes the average. All right, if you want to play tricks like that, you can "generate" points from thin air, I suppose. I kind of don't count stuff like that because it's still zero sum. The newbies lose what you win and their rankings will always exist, whether they continue playing or not. Likewise, a "retired" high ranked player still has a rating whether he looks at it any more or not. If people do stuff like this, there's no ratings system on earth that will work so it really seems moot to me. If it's a problem, I guess you start issuing and requiring "tournament cards" or something. > Another thing, while currently pointless and never likely to be of any value > except psychologically, it's very easy to sandbag your rating to distort your > (perceived) puissance. Win all the big tournaments and screw around at all the > smaller, more common locals. Actually, I've seen this happen unintentionally. Sure. You can do that in chess, also for that matter. Depends on your attitude about the game, I guess. At least, it would help a lot if the ratings actually reflected consistently good play. As it is, they don't - which is one more reason for players to not give a shit. As far as rewarding good ratings, I'm not sure that's such a bad thing. I wouldn't send prizes or anything but I might put the top ten rankings list somewhere prominent on the WW web site. I might even give free ins to the Continental tournaments or something. Only if rankings were actually somewhat reasonable and reliable, though. As it is now, you're just looking at the player's with the most recent winning streaks. Fred

Frederick Scott wrote: >Sten During wrote: >(Gad, I wish you'd adjust your news posting software so it wouldn't >insert random line breaks around your posts. It's very irritating >to read.) > Using Netscape at home, with an admittedly awful composer. You're actually having only half the irritation I have trying to compensate for what it actually tries to do :) > >>Anyway, ELO as a means of calculating ratings is not in itself broken. >>It's proven itself for several decades. I believe that ELO is >>probably the best system we can use at the moment. It's only a matter >>of what you choose to measure. >> > >My issue with the current system is that it's far too weighted towards >recent results in comparison to previous rankings. Jyhad frequently >produces results which buck the trend in terms of skill - due to seating >order, bad vs. good luck in the draw of the cards, metagame choices and >so forth. As a result, skillful players can't run their ratings above >a certain level before an unlucky defeat (or several) by someone ranked >much lower chops them back down towards average. > >It may be that this can still be arranged through the Elo system. >Having looked at the system, I would say that the problem - and what >therefore likely needs changing - is is the calculation of how many >points are won and lost by each pair of players based on the difference >in their current rating. Right now, a player gets 3300 rating and he > As you have seen from my previous postings on this topic I agree that the adjustments to a players rating during the course of one single round are just too large. That it's difficult to get above 3300 unfourtenately makes sense within the scope of the current implementation. Chess for example more or less tops out at the 2200 - 2300 range, which is 900 - 1000 above what you start with. As chess and luck have very little in common your rating is pretty much skillbased. Add to this that you're competing against four other players during one game and you can try dividing that 1000 with four and you're down to what's currently pretty much the situation with VTES. The reason that we don't top out at some 3250 is mainly due to that the numbers we use only starts punishing highrated players at maximun if the difference is actually higher than our topranked players compared to the starting amount. The main problem is probably not that you shift up to 20 or so in rating with a lower rated player performing better that you during one game. The main problem is probably that if you flunk one round that times out you're down 80 in rating in one go whereas a grandmaster in chess that suffers from stroke during a game will go down maximally 31 in one go. The proposition I made (no matter how faulty in other aspects) is based on the fact that you'll never be able to drop more than K-1, which, if I have read this thread correctly, should be 31 points. Remains other problems with a fair handling of ties and the added problem that a player is still able to jump an absurd amount in rating upwards. Another potential problem is that with the new Game Win rules then the system ought to award players that attempts to grab one, preferably two, Game Win during the prelims even if such gaming also calculates with embarrasing defeats at the tables where they fail to get a Game Win. If there is the slightest tendency that rating obviously doesn't follow the finalists but rather awards those with a lot of VP but no Game Wins, then the rating will not mirror the ranking lists of the tournaments, and by ranking list I mean the ordered list starting with the tournament winner and going downwards. Sten During >

Hmm are my posts AGAIN ignored by everyone or is google acting up again? (sorry that I have the whole post here quoted, but if the previous post didn't get to news groups then maybe this will?) Jukka prime_at_...@hotmail.com (Jukka Lindstrom) wrote in message news:<74b21954.02012...@posting.google.com>... [ quoted text not captured ]

"LSJ" <vte...@white-wolf.com> wrote in message news:3C4D8AD3...@white-wolf.com... > Frederick Scott wrote: > > I've never sat down and looked at the system but I think I'm safe in > > saying that it needs more stability to a players' ratings who have a > > lot of games under their belt. A *LOT* more. > > Constructive suggestions are welcome, although the "stability" of > long-time player's ratings isn't fundamental to the theory - its > desirability isn't necessarily a given. The Elo system has withstood > scrutiny for quite some time. But if we do care about having some stability of ratings (and it sounds like we do, at least for some values of "we"), wouldn't it be feasible/reasonable to decrease the weighting factor that determines how many points can be exchanged per game? The "change in rating" equation is currently 32*(rating- based probability of winning), right? If it were 16*(prob.) then (obviously) you wouldn't get people's scores increasing or decreasing by as much per game (or per tournament); this would make the rating system more "forgiving" in the sense of not decreasing any player's rating as fast for losing (and not increasing it as fast for winning, either). Of course, someone may have brought this up already, what with this message I'm replying to having been written yesterday and all. :-) Josh her majesty's a pretty nice girl someday i'm gonna make her mine

"Andrew S. Davidson" <a...@csi.com> wrote in message news:CB26146595064F10.F52917DA...@lp.airnews.net... > But I'd prefer a different paradigm - baseball stats. In those you > have a series of numbers and averages which give you a rounded picture > of a player's performance - home runs, on-base percentage, earned run > average, etc. And these are shown both for the current season and as > lifetime totals. They don't factor in the quality of the opposition > because that would be too complex - you're not facing a single player > but a whole team of them. Oooh. Now there's an idea. Average VPs per game, average game wins per game, average finishing place per tournament... That could be quite interesting. Supportable by current reporting systems too. Any other interesting stats people can think of to see in VTES? Josh because you're sweet and lovely girl i love you

"Joshua Duffin" <jtdu...@yahoo.com> wrote in message news:a2nd9n$12l8ut$1...@ID-121616.news.dfncis.de... > Oooh. Now there's an idea. Average VPs per game, average > game wins per game, average finishing place per tournament... > That could be quite interesting. Supportable by current > reporting systems too. Any other interesting stats people > can think of to see in VTES? I started doing this locally last year. I just built a database yesterday so that I can keep track of it easily into the future. Here's a sample: http://www.vtesinla.org/Gateway2001Results.htm > because you're sweet and lovely girl You're a sweet and lovely girl too. I think the ratings are so manual at this point that more detailed stats might be too difficult for now. Its on my wish list though. -Robert Robert Goudie rob...@vtesinla.org

Sten During wrote: > Frederick Scott wrote: > > >(Gad, I wish you'd adjust your news posting software so it wouldn't > >insert random line breaks around your posts. It's very irritating > >to read.) > > Using Netscape at home, with an admittedly awful composer. You're > actually having only half the irritation I have trying to compensate > for what it actually tries to do :) That's odd, I'm using Netscape (old Netscape, not the version 6+ rewarmed AOL thing) to compose posts as well. > >My issue with the current system is that it's far too weighted towards > >recent results in comparison to previous rankings. Jyhad frequently > >produces results which buck the trend in terms of skill - due to seating > >order, bad vs. good luck in the draw of the cards, metagame choices and > >so forth. As a result, skillful players can't run their ratings above > >a certain level before an unlucky defeat (or several) by someone ranked > >much lower chops them back down towards average. ... > As you have seen from my previous postings on this topic I agree that > the adjustments to a players rating during the course of one single round > are just too large. > > That it's difficult to get above 3300 unfourtenately makes sense within > the scope of the current implementation. This statement is either circular or wrong: if you mean, the current system is having this effect because that's the effect the current system - given the various constants they're choosing to use - has, that would be circular logic, what Mathematicians term a "tautology". If you mean this is the only result that can be achieved using ELO, I disagree. Change the constants, as I suggested before. Players will get then be able to get above a ways beyond 3300 before the lopsided effect of playing opponents with lesser ratings starts to kick in. > The main problem is probably not that you shift up to 20 or so in rating > with a lower rated player performing better that you during one game. > The main problem is probably that if you flunk one round that times out > you're down 80 in rating in one go whereas a grandmaster in chess that > suffers from stroke during a game will go down maximally 31 in one go. I disagree. That's not really a problem. You just change the constants to account for the fact that any one recorded result is subject to large amount of random luck. The system works fine, it just requires different constants to tune down the effect of any single result. Since each game turns in 6 (for a four-player game) or 10 (for a five-player game) results, this is not a problem. It just means it will take the highest and lowest rated players more games to get to their natural ratings then the current system, which is completely fair and appropriate. > Another potential problem is that with the new Game Win rules then the > system ought to award players that attempts to grab one, preferably two, > Game Win during the prelims even if such gaming also calculates with > embarrasing defeats at the tables where they fail to get a Game Win. > If there is the slightest tendency that rating obviously doesn't follow > the finalists but rather awards those with a lot of VP but no Game Wins, > then the rating will not mirror the ranking lists of the tournaments, > and by ranking list I mean the ordered list starting with the tournament >winner and going downwards. 1) It's not really a problem. 2) There's no way to be perfect in this respect. I think you're trying to claim that no system will ever work right unless the ratings system perfectly mirrors the tournament incentives. On some super-anal-retentive level, that's slightly true but I don't see a real substantial issue here. If there was, the situation before the Game Wins rule was even worse: players had an incentive to limit how many victory points they tried to get in a single game in order to assure themselves of a minimum number of victory points. This was why the Game Wins rule was created: players weren't attempting to do the best they possibly could in a single game. They were sacrificing their chances of doing fantastically well in favor of more security about getting the minimal number of victory points to qualify to be in the final - a far sight from what was good in terms of their rankings. Now, they have pretty much complete incentive to do the best they can in a single game, which is what they're being ranked by. You can't perfectly mimic the incentive to do the best overall in a tournament with the realm of a single game. And it's insane to try to that within the realm of the entire tournament, as you once suggested. (Dropouts and such will make the whole exercise arbitrary and silly.) And in the process of trying either, you lose a lot of very valid results within the context of a single game: when one player scores more VPs than another player but does not win the game. That would be bad. Fred

Frederick Scott <freds64_at_...@removethis.com> wrote in message news:<3C4ECBDF...@removethis.com>... > legbiter wrote: > > My REAL view about this is that ratings DON'T MATTER AT ALL, and that > > when we play to improve our ratings we are unwise. i've dropped from > > 13th to 300th or so without, i think, any REAL change in my abilites > > or in the competition, AND I DON'T CARE!!! > > Well, fine. But if we want to make that the "official" view of ratings > then the logical thing to do would be to get rid of ratings. I'm not > sure I see the point of keeping a sham ratings system around. O dear, sorry if i implied that my views have any official status [they don't], that they were logical [they aren't, only personal], or that i object to a sham ratings systme [i don't, it doesn't matter to me, and for those to whom it does matter, good luck to them!]. > > > However, i think it's a fairly simple matter to fix the ratings so > > that they reward you for getting to the final and for winning > > tournaments [which i think are desirable aims]. > > > > i suggest we do it this way: > > > > Do the calculation in the normal way. > > > > After all the calculations are done, add > > > > 5 [or 4, for a 4-player final] points to the winner of the tournament > > 4 [or 3] to the person who comes second > > and so on. > > > > If only one person gets VPs in a 5-player final, add 5 points to that > > person's calculated rating, and 1 point to everyone else. > > "Add?" Just "add"? I think you're missing the point about the ratings > system being zero sum. Hmmm, without going through the maths [and i haven't, recently] i'm not sure that the current ratings systme IS zero sum. But if it were, why would this be important? If you just start adding points without going out > and robbing them from someone else, then it starts to reward people merely > for participating (i.e. you get the opportunity to be in the final). No, if they get to the final they get the extra points, otherwise not. Why is this bad? This > opens up the system to a whole slew of implications that a closed zero > sum game [if it is that] doesn't have: mainly, that being able to be in as many tournaments > as possible becomes a ratings boon. (Even the players who don't get to > and finals at all get rewarded, since they frequently get to play players > who've been in the final, which is a ratings boon, too.) I say, "Well, > sure he's got a good rating, but he comes from Los Angeles where they have > a million tournaments." Sure, and that's good. Ratings systems, like tournaments, are a way to encourage people to play, surely? This is a problem for ratings because it means > they don't actually rate people. Ah. HERE'S where we disagree. i don't think you CAN rate people's playing ability like this. > > There would also be a gradual ratings inflation as the average over time > slowly increased from the 3000 that people start with. So now people > start with the average number of points but next year, it will be somewhat > less than average. 10 years from now it will likely be much less than > average. Strange stuff. Quite right, and a good thing for those whose egos NEED a "rating". > > If you're going to go to a "rewards" system for rating people and give them > their little gold stars for participating and winning things, let's do it > blatantly like the original DCI rating system for Magic before they > started using the current one: 10 points for showing up, double the points > for advancing to each subsequent round. That was for a single elimination > tournament. Our round-robin preliminaries would probably be more like > 10 points for playing each round and 10 points for each victory point > scored therein. 50 for going to the finals and 50 more for each victory > point scored there. And at the end of the year, you total them, send the > point leaders a plaque (and an invitation to the Continental Tournament of > their choice) and wipe the points clean and start over. > > Fred Sure Fred, we could do that and so did Pokemon, essentially, except you got an enamel badge too. But we have the system we have and Todd has the needs of the Brides of Bannister, as well as our own, to attend to. i've suggested a tiny tweak which rewards good performance in tourneys more, and doesn't throw the current system completely out of the coffin. i think my idea is OK, and i think what YOU seem to be gunning for is unrealistic and WORSE THAN THE CURRENT SYSTEM, because it would make people play MORE for their ratings, to the detriment of the game.

Joshua Duffin wrote: > But if we do care about having some stability of ratings > (and it sounds like we do, at least for some values of "we"), > wouldn't it be feasible/reasonable to decrease the weighting > factor that determines how many points can be exchanged per > game? The "change in rating" equation is currently 32*(rating- > based probability of winning), right? If it were 16*(prob.) > then (obviously) you wouldn't get people's scores increasing > or decreasing by as much per game (or per tournament); this > would make the rating system more "forgiving" in the sense > of not decreasing any player's rating as fast for losing (and > not increasing it as fast for winning, either). I think this, by the way, is essentially the same thing I'm suggesting. Increase the 400 or decrease the 32 and you'll have more or less the same effect. Just the exact points will differ. My way involves the points jumping up and down as much as they currently do but all the ranking levels will spread out a lot more than 3000 plus or minus 300. Your way, the effective range stays the same but the jumps will decrease in size. I like mine a little better - more arithmetic differentiation between good and bad players. But either thing should work just fine. Fred

legbiter wrote: > > Frederick Scott <freds64_at_...@removethis.com> wrote in > message news:<3C4ECBDF...@removethis.com>... > > legbiter wrote: > > > My REAL view about this is that ratings DON'T MATTER AT ALL, and that > > > when we play to improve our ratings we are unwise. i've dropped from > > > 13th to 300th or so without, i think, any REAL change in my abilites > > > or in the competition, AND I DON'T CARE!!! > > > > Well, fine. But if we want to make that the "official" view of ratings > > then the logical thing to do would be to get rid of ratings. I'm not > > sure I see the point of keeping a sham ratings system around. > > O dear, sorry if i implied that my views have any official status > [they don't], that they were logical [they aren't, only personal], or > that i object to a sham ratings systme [i don't, it doesn't matter to > me, and for those to whom it does matter, good luck to them!]. I didn't take any such implication. I was just saying that if your views _did_ represent the majority, then it makes a lot more sense to get rid of the rankings, not bother diddling with them. Besides making no sense, it's just a huge waste of time to have a number that exists for no other reason then to complain about how bad it is and justify why everyone should ignore it. > Hmmm, without going through the maths [and i haven't, recently] i'm > not sure that the current ratings systme IS zero sum. It is, unless you take into account some of Ian's things. At least it's zero sum in the sense that what one rating gains in points, another one has got to lose the exact same number of points and visa verse. What happens to those various individual ratings in terms of going away or being ignore or whatever is another thing. For all intents and purposes, I think that pretty much makes it a zero sum thing. > But if it were, why would this be important? I spent much of my previous post explaining that. To sum it up, it changes the value of a rating over time. 3150 this year won't mean the same thing as 3150 next year is the most succinct way I know how to put it. IMO, that would be bad. > > This is a problem for ratings because it means > > they don't actually rate people. > > Ah. HERE'S where we disagree. i don't think you CAN rate people's > playing ability like this. If you can not, then it's pointless to have them. To the extent you can, this will likely cause a problem. > i think my idea is OK, Sure. It's an opinion thing, as long you accurately comprehend the flaw you'd be introducing into the system. > and i think what YOU seem to be gunning for is unrealistic I'm not sure why you would term it "unrealistic". Neither of these proposals is likely to fly in reality, I suppose. (And, by the way, don't confuse the idea for a cumulative point system with my other proposal, to modify the ELO constants in use. That, I think, is a very realistic and worthwhile proposal.) I don't perceive people really want a system that increases ratings over time but if they did, I like mine a lot better than yours because, 1) it's a whole lot more honest about what's going on; and 2) resetting points on an annual basis would be quite appropriate. > and WORSE THAN THE CURRENT SYSTEM, because it would make people play > MORE for their ratings, to the detriment of the game. Both cumulative points proposals aren't worth very much, I agree. OTOH it would be hard to come up with anything worse than the current system, which seems to spit out more or less randomish numbers on a periodic basis. But, however you feel about it, arguing that anything that would make people care about their ratings more is bad for the system is one of the sillier arguments I've heard anyone make on Usenet. Not that it's silly to say people shouldn't care about their ratings. That part is certainly reasonable enough, even if I disagree with it. It's just that your logical conclusion should be to get rid of ratings. Nothing more, nothing less. Fred

"Joshua Duffin" <jtdu...@yahoo.com> wrote in news:a2nd9n$12l8ut$1@ID- 121616.news.dfncis.de: > > "Andrew S. Davidson" <a...@csi.com> wrote in message > >> But I'd prefer a different paradigm - baseball stats. In those you >> have a series of numbers and averages which give you a rounded picture >> of a player's performance - home runs, on-base percentage, earned run >> average, etc. And these are shown both for the current season and as >> lifetime totals. They don't factor in the quality of the opposition >> because that would be too complex - you're not facing a single player >> but a whole team of them. > > Oooh. Now there's an idea. Average VPs per game, average > game wins per game, average finishing place per tournament... > That could be quite interesting. Supportable by current > reporting systems too. Any other interesting stats people > can think of to see in VTES? Minor nitpick: average finishing place per tournament would be useless since the number of players at each tournament varies wildly. So someone who scores a couple VPs at the EC would get a 50th place for his average, while someone scoring 0 VPs at a local tournament could get a 10th place. Otherwise, I certainly like the idea, it's a much more complete and accurate measure of a player's skill to know those stats than the current ranking system, IMHO. Flux

In article <Xns91A05B532...@130.133.1.4>, Flux <fl...@netc.pt> wrote: >Minor nitpick: average finishing place per tournament would be useless >since the number of players at each tournament varies wildly. So someone >who scores a couple VPs at the EC would get a 50th place for his average, >while someone scoring 0 VPs at a local tournament could get a 10th place. > >Otherwise, I certainly like the idea, it's a much more complete and >accurate measure of a player's skill to know those stats than the current >ranking system, IMHO. Easily enough fixed: Replace it with a 'scoring ratio' type number, like 1/50 (first in a 50-person tournament) or 4/20 (fourth in a 20-person tournament). Just to satisfy my aesthetic of 'higher number = better performance', mangle it by doing 1-(place/number of attendants), so the guy who places 50th in a field of 50 has (1-(50/50))=0 averaged into his lifetime place stat, and the guy who comes in third has (1-(3/50))=0.94. So someone who wins every tournament they enter has a place stat of 1.00, someone who loses every time has a stat of 0.00, and the rest of us vary somewhere in between. This stat can still be sandbagged by playing as many low-attendance tournaments as possible, and doesn't reward winning a really big event, but I don't feel these are huge vulnerabilities. Winning a big event like a Gen*Con or French Championship tends to garner you a reputation all by itself, anyway. gomi -- Blood, guts, guns, cuts Knives, lives, wives, nuns, sluts

"Barny Baker" <barny...@synergygroup.co.uk> wrote in message news:a9abcc59.02012...@posting.google.com... (snip) > The problem I have with the system is the definition of 'well'. > Getting in finals practically every time in my mind would constitute > 'well', unfortunately for me the ranking system does not agree. the ranking system is stupid, i agree, but anyway there is really no reward in being well ranked. So as there is nothin to win or lose, we can live with it. If White wolf starts to give prizes to, say, the 3 top ranked players, then it will be time to claim it's really unfair. > It frustrates me immensly that players who have had many bad > tournaments can have one decent one (getting 6th place for instance) > and be above me in the rankings. To me this is not a fair reflection > of a players skill or achievements. Pride hurts, (but hey f#ck pride). some players in france are very awful badasses, they went to a dozen tournaments only and are very high ranked. Only because of the ranking system's failure. On the opposite very skilled players who attended 25+ tournaments are low because the good / bad games ratio is more difficult to maintain on the long run than in a dozen games. > I don't have the ability to come up with new equations to propose a > change and I don't have an answer (I would say solution but by others > may not agree that there is a problem), I suppose I'll have to take my > medicine, continue to get incredibly drunk at tourney's and be content > with the fact that I'll probably come third again. Hey, you can believe this guy : IT WORKS !! =) > While I'm here Hi to all the Frenchies, Francois and Reyda in > particular. hi dude ! =) come again ! reyda

It seems that, in a game which usually has 4+ players, and such a large amount of random chance involved, that individual games should be worth very little. The way most other ladders work, it is relatively difficult to drop in position. If its a ranking system, 1st 2nd etc, at the higher levels, for someone to rise above you, they would have to beat you themselves. If you decide to use a point system, there is no reason for it to suddenly become a lot easier to lose your position. Also, if people are having trouble procesing words in a webpage, why not try using a word processor? You'd have the advantage of spell check, saving your work, etc. Then just copy and paste the results. It's what I do for any long post. Wisefool Mercuriel

Frederick Scott <freds64_at_...@removethis.com> wrote in message news:<3C4F4E60...@removethis.com>... > legbiter wrote: > > > > Frederick Scott <freds64_at_...@removethis.com> wrote in > > message news:<3C4ECBDF...@removethis.com>... > > > legbiter wrote: > > > > My REAL view about this is that ratings DON'T MATTER AT ALL, and that > > > > when we play to improve our ratings we are unwise. i've dropped from > > > > 13th to 300th or so without, i think, any REAL change in my abilites > > > > or in the competition, AND I DON'T CARE!!! > > > > > > Well, fine. But if we want to make that the "official" view of ratings > > > then the logical thing to do would be to get rid of ratings. I'm not > > > sure I see the point of keeping a sham ratings system around. > > > > O dear, sorry if i implied that my views have any official status > > [they don't], that they were logical [they aren't, only personal], or > > that i object to a sham ratings systme [i don't, it doesn't matter to > > me, and for those to whom it does matter, good luck to them!]. > > I didn't take any such implication. I was just saying that if your > views _did_ represent the majority, then it makes a lot more sense to > get rid of the rankings, not bother diddling with them. Besides making > no sense, it's just a huge waste of time to have a number that exists > for no other reason then to complain about how bad it is and justify > why everyone should ignore it. Well, i don't accept that's the only reason for their existence. i suggest they exist because people like a ratings system. > > > Hmmm, without going through the maths [and i haven't, recently] i'm > > not sure that the current ratings systme IS zero sum. > > It is, unless you take into account some of Ian's things. At least > it's zero sum in the sense that what one rating gains in points, another > one has got to lose the exact same number of points and visa verse. > What happens to those various individual ratings in terms of going away > or being ignore or whatever is another thing. For all intents and > purposes, I think that pretty much makes it a zero sum thing. Again, without the equation in front of me [and even then, i would have to work out some examples] i seem to recall that there's a scalar, such that if Rob on 3500 or whatever sweeps a table with the munchkin triplets each on 2800 they will each lose maybe 5 points and he will gain about 3. > > > But if it were, why would this be important? > > I spent much of my previous post explaining that. To sum it up, it > changes the value of a rating over time. 3150 this year won't mean > the same thing as 3150 next year is the most succinct way I know how > to put it. IMO, that would be bad. Right, understood. Where we disagree is that i think it wouldn't matter. > > > > This is a problem for ratings because it means > > > they don't actually rate people. > > > > Ah. HERE'S where we disagree. i don't think you CAN rate people's > > playing ability like this. > > If you can not, then it's pointless to have them. Not pointless, because people like them, and they encourage people to play the game. To the extent you > can, this will likely cause a problem. > > > i think my idea is OK, > > Sure. It's an opinion thing, as long you accurately comprehend the flaw > you'd be introducing into the system. > > > and i think what YOU seem to be gunning for is unrealistic > > I'm not sure why you would term it "unrealistic". Because you want a better system of ratings, and i think that this is unattainable for a multi-player game. Neither of these > proposals is likely to fly in reality, I suppose. (And, by the way, > don't confuse the idea for a cumulative point system with my other > proposal, to modify the ELO constants in use. That, I think, is a > very realistic and worthwhile proposal.) Yes, i read that after i posted and i agree that that's worth thinking about. I don't perceive people > really want a system that increases ratings over time but if they did, > I like mine a lot better than yours because, 1) it's a whole lot more > honest about what's going on; and 2) resetting points on an annual basis > would be quite appropriate. Sure, i think that IS a better idea, and it's also easier to implement. > > > and WORSE THAN THE CURRENT SYSTEM, because it would make people play > > MORE for their ratings, to the detriment of the game. > > Both cumulative points proposals aren't worth very much, I agree. > OTOH it would be hard to come up with anything worse than the current > system, which seems to spit out more or less randomish numbers on a > periodic basis. But, however you feel about it, arguing that anything > that would make people care about their ratings more is bad for the > system is one of the sillier arguments I've heard anyone make on Usenet. > Not that it's silly to say people shouldn't care about their ratings. > That part is certainly reasonable enough, even if I disagree with it. > It's just that your logical conclusion should be to get rid of ratings. > Nothing more, nothing less. i'm not following you here. My position is that i personally don't care about the ratings, and i don't play so's to improve my rating. If i did, i think i would have less fun. But i've got no problem with other people caring about their ratings and trying to improve them, because i'm tolerant of human diversity. Similarly, i think that "democracy" is a really silly system but it kinda works as implemented in the so-called western so-called democracies - but heaven forbid that we ever get MORE democracy!!! In the same way, i'm against what we might term an ideological fix to "perfect" the ratings because i think it won't work, and i'm against focussing on ratings because i don't think they're important. i can't see what's illogical or silly about any of the above, sorry! > > Fred

Barny Baker wrote: > > LSJ <vte...@white-wolf.com> wrote in message news:<3C4EA276...@white-wolf.com>... > > Ben Swainbank wrote: > > > > > > LSJ <vte...@white-wolf.com> wrote in message news:<3C4DC33F...@white-wolf.com>... > > > > > > > Coming in 3rd should break even statistically half the time with not > > > > playing at all. > > > > > > Yikes. That's 3rd in the tournament, right? That seems off. If I show > > > > No. That's third in a game. > > Sorry but it's third in a tournament, not only that, third or better > in consecutive tournaments over a a long time, my worst positions over > the last 2 years (an educated guess) have been 6th in the U.K. > Nationals and 5th at The European qualifyer. Sorry. It's third in a game. Third in a tournament tends to improve your rating. > If it was third in a game I wouldn't be labouring the point, but it is > not and I am. It's game. Coming in 3 in a 5-player table means 2 wins and 2 losses. Net: break even (statistically). [ quoted text not captured ]

In message <3C4FFEDE...@white-wolf.com>, LSJ <vtesrep@white- wolf.com> writes: >> Sorry but it's third in a tournament, not only that, third or better >> in consecutive tournaments over a a long time, my worst positions over >> the last 2 years (an educated guess) have been 6th in the U.K. >> Nationals and 5th at The European qualifyer. > >Sorry. It's third in a game. >Third in a tournament tends to improve your rating. Would there be any mileage in the earlier suggestion of ELO-ing everyone *once* for the tournament, based on their position in the tournament? (So the winner scores a win against everyone, the people in (joint?) bottom score a loss against everyone else and so on.) This would tend to polarise the scores a little more I guess. Are there any decent references on zero-sum ranking systems anywhere? My maths is rusty, but it'd be interesting to have a look. Perhaps alternatively, in an advanced database system, a player could have their results calculated long-term, and then over, say, the last ten games (that is, each round, so a finalist in a 3+F tournament would play 4 games, say) calculated in some fashion, so that a good ELO rating accompanied with a poor ten-game rating would suggest certain things, and so on. -- James Coupe "Never give in to them," she whispered. "No matter what PGP 0x5D623D5D EB they do or how important you feel it is to get their accep- D690ECD7A1FB457CA21 tance. Never kill part of yourself for them. Because other 3D7E668C3695D623D5D people will notice that part is missing before you do."

Frederick Scott wrote: > Sten During wrote: > >>Frederick Scott wrote: > > > That's odd, I'm using Netscape (old Netscape, not the version 6+ > rewarmed AOL thing) to compose posts as well. > You just guessed what I'm running... >>That it's difficult to get above 3300 unfourtenately makes sense within >>the scope of the current implementation. >> > > This statement is either circular or wrong: if you mean, the current system > is having this effect because that's the effect the current system - given > the various constants they're choosing to use - has, that would be circular > logic, what Mathematicians term a "tautology". If you mean this is the only > result that can be achieved using ELO, I disagree. Change the constants, > as I suggested before. Players will get then be able to get above a ways > beyond 3300 before the lopsided effect of playing opponents with lesser > ratings starts to kick in. It's the only result that, given the currently used values, can be had. Changing the constants as you suggest will of course change the maximum point any one player will in reality ever achieve. > > >>The main problem is probably not that you shift up to 20 or so in rating >>with a lower rated player performing better that you during one game. >>The main problem is probably that if you flunk one round that times out >>you're down 80 in rating in one go whereas a grandmaster in chess that >>suffers from stroke during a game will go down maximally 31 in one go. >> > > I disagree. That's not really a problem. You just change the constants > to account for the fact that any one recorded result is subject to large > amount of random luck. The system works fine, it just requires different > constants to tune down the effect of any single result. Since each game > turns in 6 (for a four-player game) or 10 (for a five-player game) results, > this is not a problem. It just means it will take the highest and lowest > rated players more games to get to their natural ratings then the current > system, which is completely fair and appropriate. And I disagree again. Tuning the constants will not change that a top- rated player (using todays numbers that ought to be around 3300) will drop 80/300 during one single round if faced with bad luck. What's mathematically correct is of no concern - as long as the players don't feel comfortable with the ratingsystem, or even worsem feel that it doesn't represent what's happening in a fair way, then the system is flawed no matter how correct the designer of the system can prove it to be. Why is it bad that a top-ranked player can lose more than 25% of their rating-total (3300 will be seen as a grand total of 300 as everyone starts at 3000) during the course of two hours? Mathematically it is, as you correctly state, no problem at all. From the players point of view it's not only a problem, but unacceptable that during the course of one tournament day you'll start as number 12 of all rated players, drop to number 150 and rise again to number 24. > > >>Another potential problem is that with the new Game Win rules then the >>system ought to award players that attempts to grab one, preferably two, >>Game Win during the prelims even if such gaming also calculates with >>embarrasing defeats at the tables where they fail to get a Game Win. >>If there is the slightest tendency that rating obviously doesn't follow >>the finalists but rather awards those with a lot of VP but no Game Wins, >>then the rating will not mirror the ranking lists of the tournaments, >>and by ranking list I mean the ordered list starting with the tournament >>winner and going downwards. >> > > 1) It's not really a problem. This is a problem. > > 2) There's no way to be perfect in this respect. You are 100% correct here. > > I think you're trying to claim that no system will ever work right unless > the ratings system perfectly mirrors the tournament incentives. On some > super-anal-retentive level, that's slightly true but I don't see a real > substantial issue here. If there was, the situation before the Game Wins > rule was even worse: players had an incentive to limit how many victory > points they tried to get in a single game in order to assure themselves > of a minimum number of victory points. This was why the Game Wins rule > Rating and earlier rules did correspond with each other. The only thing that mattered was the total amount of VP that you gained during the prelims. Ratingsystem effectively follows this very closely. Ratingsystem still follows this, but now you're to gain as many Game Wins as possible. There is a major risk that ratings and tournament- placings will diverge. What I'm saying is of course totally untrue for all the players that regulary grab 3 Game Wins during the prelims. Sten During

Sten During wrote: > > Frederick Scott wrote: > >> > >>The main problem is probably not that you shift up to 20 or so in rating > >>with a lower rated player performing better that you during one game. > >>The main problem is probably that if you flunk one round that times out > >>you're down 80 in rating in one go whereas a grandmaster in chess that > >>suffers from stroke during a game will go down maximally 31 in one go. > >> > > > > I disagree. That's not really a problem. You just change the constants > > to account for the fact that any one recorded result is subject to large > > amount of random luck. The system works fine, it just requires different > > constants to tune down the effect of any single result. Since each game > > turns in 6 (for a four-player game) or 10 (for a five-player game) results, > > this is not a problem. It just means it will take the highest and lowest > > rated players more games to get to their natural ratings then the current > > system, which is completely fair and appropriate. > > And I disagree again. Tuning the constants will not change that a top- > rated player (using todays numbers that ought to be around 3300) will > drop 80/300 during one single round if faced with bad luck. (I'm not sure what you mean by "80/300". At most you can drop 128 points during a single round, if your rating was infinitely greater than your opponents and you lost to them all. I'll use 80 points as an example of a very bad round for discussion purposes.) If you change the "32", as Josh suggested, it will. If you increase the "400", as I suggested, it will not. However, the 80 point drop would then be in the context of a much greater effective range of scores, which is just as good. We don't care about the actual numbers, only that they behave correctly. If Barney drops 80 points in a single round but is still 500 points better than most mediocre players, that's much better than dropping 80 points and suddenly having about the same rating as all the mediocre players. > What's mathematically correct is of no concern - as long as the players > don't feel comfortable with the ratingsystem, ...and having it behave in a way that seems to indicate a player's true skill will cause players to feel more comfortable with the system. I don't know what you mean by "mathematically correct". The system will behave much better in the sense at any given moment, more skilled players will be much more likely to have a higher rating than less skilled players - excepting the time period between when a new player starts his rating at 3000 and the rating moves toward the range that reflects his true skill. > > I think you're trying to claim that no system will ever work right unless > > the ratings system perfectly mirrors the tournament incentives. On some > > super-anal-retentive level, that's slightly true but I don't see a real > > substantial issue here. If there was, the situation before the Game Wins > > rule was even worse: players had an incentive to limit how many victory > > points they tried to get in a single game in order to assure themselves > > of a minimum number of victory points. This was why the Game Wins rule > > Rating and earlier rules did correspond with each other. The only thing > that mattered was the total amount of VP that you gained during the > prelims. Ratingsystem effectively follows this very closely. I explained why this was not true in my previous post. (The ratings system rated points PER GAME, not PER TOURNAMENT. The difference was significant, as can be seen by the fact that it was influencing the behavior of players to the extend that the Game Wins rule was thought to be a necessary change.) If you don't want to address this point, we have nothing to discuss as your arguments are based on an incorrect view of reality. Fred

"Flux" <fl...@netc.pt> wrote in message news:Xns91A05B532...@130.133.1.4... > Minor nitpick: average finishing place per tournament would be useless > since the number of players at each tournament varies wildly. So someone > who scores a couple VPs at the EC would get a 50th place for his average, > while someone scoring 0 VPs at a local tournament could get a 10th place. As Gomi pointed out, you could calculate a "place-in-tournament ratio" that factored out the number of players. Good idea in my book. > Otherwise, I certainly like the idea, it's a much more complete and > accurate measure of a player's skill to know those stats than the current > ranking system, IMHO. I think it'd be cool. C'mon people, there must be more stats we could calculate. What should they be? :-) Josh freaky me freaky you

In article <a2pmgb$132l02$1...@id-121616.news.dfncis.de>, Joshua Duffin <jtdu...@yahoo.com> wrote: >I think it'd be cool. C'mon people, there must be more stats >we could calculate. What should they be? :-) Well, here's what I'm tracking for January (non-tournament): Games Played: 6 Total VPs: 8 Game Wins: 2 Sweeps: 2 Sweepratio (Sweeps/Game Wins): 1 Shutouts (no VPs scored): 3 Personal Suckfactor (Shutouts/Games Played): 0.5 Oustratio*: 0.65 Average table size: 3.67 Game Wins per Game: 0.33 VPs per Game: 1.33 VPs scored per Opponent faced**: 0.36 * Oustratio: {sum[i->n](xi/yi)}/n, where x=oust order (1=first ousted, etc) and y = size of table. averaged over games played (n). Sweeps are 1 (5/5 or 4/4, whatever), being ousted first is the reciprocal of the table size (usually 1/5 or 1/4). I don't know if this number means anything necessarily; it might be a vague indicator of your 'endurance' or how 'defensively' you play, or just be an indicator of how often you get seating-position-screwed, I dunno. If everyone tracked all their stats for like a year someone could do a regression on VPs or Game Wins or Tourney Wins vs. Oustratio and see if there's a correlation. Or not. ** total vps/(average table size * number of games) gomi soon it'll be like baseball! [ quoted text not captured ]

"Robert Goudie" <rrgo...@earthlink.net> wrote in message news:OcH38.7223$Fh4.6...@newsread1.prod.itd.earthlink.net... [ quoted text not captured ] I really like the stat chart Robert made. I think it really shows a lot of information about the player. However I am at a loss on how you would determine a 1st, 2nd, 3rd, etc. rating. Which category would be most important? VP's per game? Steve Bucy

"Robert Goudie" <rrgo...@earthlink.net> wrote in message news:OcH38.7223$Fh4.6...@newsread1.prod.itd.earthlink.net... > "Joshua Duffin" <jtdu...@yahoo.com> wrote in message > news:a2nd9n$12l8ut$1...@ID-121616.news.dfncis.de... > > > Oooh. Now there's an idea. Average VPs per game, average > > game wins per game, average finishing place per tournament... > > That could be quite interesting. Supportable by current > > reporting systems too. Any other interesting stats people > > can think of to see in VTES? > > I started doing this locally last year. I just built a database yesterday so > that I can keep track of it easily into the future. Here's a sample: > > http://www.vtesinla.org/Gateway2001Results.htm That's super-sweet. And lovely. I like. > I think the ratings are so manual at this point that more detailed stats might > be too difficult for now. Its on my wish list though. Well, if you could build a database, surely the same thing could be done with the overall ratings system? Though if it is excessively manual right now, that might be an awful lot of work to get it all into the database. We could just start from now, if that'd make it much easier. Josh

Frederick Scott wrote: > Sten During wrote: > >> >>drop 80/300 during one single round if faced with bad luck. >> > > (I'm not sure what you mean by "80/300". At most you can drop 128 points > during a single round, if your rating was infinitely greater than your > opponents and you lost to them all. I'll use 80 points as an example of > a very bad round for discussion purposes.) > What I was actually referring to was that dropping 80 points out of 300 (300 being around max for anyone to rise above starting point) means that a top-ranked player drops over 25% of all gained points during one single game. This is very bad ;) > >>What's mathematically correct is of no concern - as long as the players >>don't feel comfortable with the ratingsystem, >> > > ...and having it behave in a way that seems to indicate a player's true > skill will cause players to feel more comfortable with the system. I > don't know what you mean by "mathematically correct". The system will > behave much better in the sense at any given moment, more skilled players > will be much more likely to have a higher rating than less skilled > players - excepting the time period between when a new player starts his > rating at 3000 and the rating moves toward the range that reflects his > true skill. Absolutely agreed. What the current system lacks (and by current system I'm not referring to some kind of mystical inherent flaw in ELO) is stability that makes sense. In a way it's too stable. I can play ten very succesful games as a new player and end up as a worldclass rated player, which most probably is not corresponding to my actual skill. If we play with the number 80 as what you lose during a bad game, and let's assume that a highranked player has a very bad day instead. That player will most likely plummet over 200 points during one tournament. Sure, such a display of bad performance should be duly noted, but really! Two horrid tournament in a row and that player is below starting point, which compared to chess is something a grandmaster won't be able to do during two horrid seasons of league-playing. > > >>>I think you're trying to claim that no system will ever work right unless >>>the ratings system perfectly mirrors the tournament incentives. On some >>>super-anal-retentive level, that's slightly true but I don't see a real >>>substantial issue here. If there was, the situation before the Game Wins >>>rule was even worse: players had an incentive to limit how many victory >>>points they tried to get in a single game in order to assure themselves >>>of a minimum number of victory points. This was why the Game Wins rule >>> >>Rating and earlier rules did correspond with each other. The only thing >>that mattered was the total amount of VP that you gained during the >>prelims. Ratingsystem effectively follows this very closely. >> > > I explained why this was not true in my previous post. (The ratings system > rated points PER GAME, not PER TOURNAMENT. The difference was significant, > as can be seen by the fact that it was influencing the behavior of players > to the extend that the Game Wins rule was thought to be a necessary change.) > If you don't want to address this point, we have nothing to discuss as your > arguments are based on an incorrect view of reality. > I'm trying to point out that the ratingssystem in a much larger degree follows the old tournament system compared to the new tournament system. Agreed the ratings change due to one single game, but I think we can agree to the rating following the distribution of VP during that single game. Using old tournament rules you simply summed up all VP and those with most went to the finals. That corresponds better with the ratings system compared to todays tournamentsystem that awards a more agressive chancetaking style of play as you today can afford to be embarrasingly manhandled during a game or two as log as you get one Game Win. We may have different opinions about who displays most skill - the one with a Game Win from the only round where the player wasn't ousted first, or the one with three rounds of a solid 2VP each, but the new rules are absolute - if those 6 VP gained doesn't include one Game Win then the former player will be better placed in order to get to the finals. I'm saying that the ratingssystem awards the latter player in both the old and the new tournamentsystem, in which case the ratingssystem follows the old rules more closely compared to what is the situation today. Sten During

go...@panix.com (Gomi no Sensei) wrote in message news:<a2pq0o$j3u$1...@panix2.panix.com>... > Well, here's what I'm tracking for January (non-tournament): Interesting. I'm curious about a couple of your computations (bear in mind that it's been ages since my crappy Stats class, and even longer since I had any real math -- I'm about 8 years removed from Diff Eq at this point). My numbers from last night follow (and are not really atypical, though maybe I'll keep tracking for a bit just for fun): Games Played: 3 Total VPs: 8.0 Game Wins: 2 Sweeps: 1 Sweepratio: 0.5 Shutouts: 0 Personal Suckfactor: 0.0 Oustratio: 0.83 Okay, so I'm not sure about this "x=oust order" thing. My games went: 3 of 4 VP, my initial prey got one VP, and then I got the rest. Is this 4/4, or something else? Seems like, from your 'endurance/defensively' comments. Later games were 1 of 6, but 3rd player ousted (my grandprey, then my prey, then me, then my grandpredator...it was almost roulette...). Final game was my sweep. So, that looks to me like (4/4 + 3/6 + 4/4)/3. Is that correct? This calculation gives the ratio I listed above. > * Oustratio: {sum[i->n](xi/yi)}/n, where x=oust order (1=first ousted, etc) > and y = size of table. averaged over games played (n). Sweeps are 1 (5/5 > or 4/4, whatever), being ousted first is the reciprocal of the table size > (usually 1/5 or 1/4). I don't know if this number means anything necessarily; > it might be a vague indicator of your 'endurance' or how 'defensively' you > play, or just be an indicator of how often you get seating-position-screwed, > I dunno. If everyone tracked all their stats for like a year someone could do > a regression on VPs or Game Wins or Tourney Wins vs. Oustratio and see if > there's a correlation. Or not. Average table size: 3.33 (one 6, 2 4s) Game Wins per Game: 0.66 VPs per Game: 2.66 VPs scored per Opponent faced: 0.8 > soon it'll be like baseball! Is that good? Xian is surprised the national pastime isn't pro wrestling (or tractor pulling)

go...@panix.com (Gomi no Sensei) wrote in message news:<a2pq0o$j3u$1...@panix2.panix.com>... Stupid deja. And stupid me for messing up two measurements. Average table size: 4.66 = (4+4+6)/3, not 3.33 or whatever I said VPs scored per Opponent faced: 0.57, not 0.80. Xian

"Steve Bucy" <tb...@lainet.com> wrote in message news:a2pue...@enews1.newsguy.com... > > "Robert Goudie" <rrgo...@earthlink.net> wrote in message > news:OcH38.7223$Fh4.6...@newsread1.prod.itd.earthlink.net... > > I think the ratings are so manual at this point that more detailed stats > might > > be too difficult for now. Its on my wish list though. > I really like the stat chart Robert made. I think it really shows a lot of > information about the player. However I am at a loss on how you would > determine a 1st, 2nd, 3rd, etc. rating. Which category would be most > important? VP's per game? Right, that's the issue with any "multiple personal stats" approach to ratings. My offhand preference would be "game wins per game" for the first ranking, though it might be equally good or better to have multiple ranking lists: one for "game wins per game", one for "most tournaments won", that sort of thing. ("VPs per game" has the slight disadvantage that playing a lot of four-player games may decrease your stat without reflecting a lower skill level.) Josh cool as the deep blue ocean

"Gomi no Sensei" <go...@panix.com> wrote in message news:a2pq0o$j3u$1...@panix2.panix.com... > Well, here's what I'm tracking for January (non-tournament): > > Games Played: 6 > Total VPs: 8 > Game Wins: 2 > Sweeps: 2 > Sweepratio (Sweeps/Game Wins): 1 > Shutouts (no VPs scored): 3 > Personal Suckfactor (Shutouts/Games Played): 0.5 > Oustratio*: 0.65 > Average table size: 3.67 > Game Wins per Game: 0.33 > VPs per Game: 1.33 > VPs scored per Opponent faced**: 0.36 Cool cool. Maybe I'll try to track that too. Of course that'd require remembering correctly how many VPs I (and others, if they wanted to be tracked too) got each game... which I definitely have trouble with sometimes. > soon it'll be like baseball! Can "bases stolen" become "cheats attempted"? :-) Josh fabricate!