VOTE CHESS RATING INSTEAD OF FIXED SCORING SYSTEM

Say What? One month?

I suggest using The Glicko System for finding ratings after finding the current ratings by following these steps:

1. Take the numbers of games your group has done, and place them into groups of 35. (Group 1 is games 1-35, Group 2 is games 36-70, Group 3 is games 71-105, etc.)
2. Each group gets a score multiplier. (Group 1's multiplier is 1, Group 2's is .80, Group 3's is .65, Group 4's is .50, Group 5's is .40, Group 6's is .30, Group 7's is .25, Groupo 8's is .20, Group 9's is .15, Group 10's is .10, Group 11's is .08, Group 12's is .05, Group 13's .03, Group 14's is .02, Group 15's is .01)
3. Take the winning percentage, and find that percent of the number of games in group 1.
4. Take the drawing/losing percentage, and find that percent of the number of games in group 1.
5. Subtract the result of step 4 from the result of step 3, then multiply the difference by the multiplier.
6. Repeat for the rest of the groups.
7. Add up all the resulting products from step 5 and multiply it by 15, then add 1200. If the number is above 3500, take it down to 3500. If the number is below 0, take it to 0. Round the number to the nearest whole number.
8. The new number is your current rating!

linksspringer has downloaded all the vote chess results from the site, and done some initial calculations using Elo's system.  He found the top ten teams to be 1.Funtastic Chessmates 2.BLACK OR WHITE 3.Chess Group! 4.Team Italia 5.The Dream Team 6.Team Canada 7.Fast Players Group 8.European Union Team 9.Team Romania 10.We Chat Global.  Obviously, this is just a preliminary list, and things are likely to change rapidly, but anyway, it seems like a fairly good first approximation.  I'm sure he'll post more details once he's had more time to work on it.

Riga, I dont want to dampen anyone's spirit, but I think there is something seriously wrong with linksspringer's calculations! Our group has already played with two of those "top ten teams" you have mentioned - they belong in the top 20 all right, but as far as their playing strength is considered, they are far inferior to Narz's group, Indian Marauders, The Power of Chess, Cheers Vote Chess Cafe, etc. (no, unlike others, I am NOT mentioning my group here to push it).

So basically you want to replace one crappy system with another.

To be fair to all, it is definitely better than the existing one.

babs has a good point

i concur

I'm sure we all have our favourite vote chess groups that we would like to see at the top of the list.  Indian Marauders and Cheers! Vote Chess Cafe are obviously two very strong teams that over time will probably rise to the top of any Elo or Glicko based list.  Ideally, each rated group would have at least 30 games available for rating, but most teams are far below that right now, so any rating has to be seen as provisional.  Another factor is that while competitive groups like Indian Marauders and Cheers! Vote Chess Cafe have been actively seeking out strong groups to play against, large "friendly" groups like Funtastic Chessmates! and The Dream Team have yet to face their main rivals.

I strongly believe that either Elo or Glicko would be a vast improvement over the current Vote Leaderboard because these systems at least make an effort to take into account the strength of opponents each group faces in its games.  In the long run, that will work in favour of the stronger teams much more than the current points system.  I think linksspringer definitely has the right idea, but it is important that everyone understand that any ratings produced at this early stage are going to reflect the few results available against other groups whose real strength is also not yet known.

I very much like the idea of having group ratings (preferably for both vote chess and for matches).  Erik mentioned the concern that not all groups will play enough to have representative ratings, but thats the same for individual players as well.  So, as long as you can view the game (or match) count alongside the rating, then it can be evaluated accordingly.  In anycase, it tells me more of the groups potential than the points system which is weighted heavily based on the quantity of games/matches.  Also, for those they have played lots of games/matches, then it becomes pretty meaningful.  I also think this can be started at any point, regardless of past games/matches (if there was a way to calculate those to get starting rating then great), again the game/match count will reflect what its based on.  Its just a metric, you take it for what its worth...and for me, more valuable than the points.  The points are meaningful to those trying to climb a circuit ladder of some sort.

A year later.....

And I still agree

Yes, obviously some elo rating system would be more interesting than the current "look how many games I can play" list...

While starting over from scratch would be frustrating for everybody, and, as has been stated by chess.com, it would take a very long time to establish ratings, I think this is not necessary. With available data, it's okay to give teams initial elo ratings without having to do some very complicated iterative calculations.

Here is a simple proposal :

• first, you set two groups of VC teams : those having played more than 10 games (established - group A) and those having played less than that (group B)
• For Group A, you calculate an initial rating based on %won as follow : 0% gives a 1000 rating, 50% gives a 2000 rating and 100% gives a 3000 rating.
• Then you calculate the average # of games played by teams in group A, as well as the Standard Deviation (SD)
• You then adjust the initial rating by +/-200pts. for each SD in the # of games played, with a mamixum of +/- 3SD (so +/- 600 pts at most). This way, teams which have played significantly more than others still get a bonus and should be among the top teams, but so will the teams with a good %won over a decent # of games
• Now for group B, you can simply assign an initial rating on a different scale, for example 700 (0% won) - 1700 (100% won), and you don't need to adjust according to the # of games played
• New teams could get a 1200 rating

I don't know how it would work out, and obviously the # can be tweaked a little, but I think that would be a fair and convenient way to get fair rating estimates without unduly erasing former results

Chess.com staff - could you make a simulation (should be quite easy with raw data available)

Almost a year an a half later, any chance of saying yes to this now?

Pretty please???

With sugar on top?

Yeah, some groups have played 900 games, so their rating would probably be pretty accurate by now.

Just so this discusion dosn't go away I think the idea of having an actual leaderboard or ladder is a great idea. I think the system in place now is rather useless. Could you imagine playing one on one with the vote chess system in place now? The idea of having a rating system for vote chess is not new but I hope that it happens at some point.

I like the rating idea a lot, but even better would be some kind of standing or ladder or something which is also mentioned in this 3 page thread. 

Vote chess can be a lot of fun and a great learning experience and seems to be one of the best ways of merging the best aspects of chess and "social networking" which seems to be a big part of chess.com.  I like that I've gotten to know-in-chess some people from like everywhere, either that they're tactical geniuses or just to see how hard the better players have worked to get there, or that they are insane pirate/canibals that I'd be afraid of meeting in even a well lit alley. 

Anything that improves vote chess and makes the team feel a slightly higher imaginary stake in it would be good for the site (IMHO) and for us. 

Yup, this thread is going for 2 years now, maybe we should march to the square :)

PLEASE HELP -- Does anyone know where on this site to find the listing of the perameters of how games here are scored?...e.g. how many points lost by timing-out vs. being checkmated?  TY!!

Your new rating is calculated based on the relative ratings of you and your opponent and the result of the game (1 for a win, 1/2 a point for a draw and 0 for loss).  The method in which you lose be it on time, by checkmate or through resignation has no effect on the calculation.  In general, if you lose to players who are rated much higher than you, it will pull your rating down less than if you lose to someone who is lower rated than you.

Good idea. In fact it seems some groups are already mentioning some ELO points in matches they play among themselves - not very clear.

It's a shame this thread is dead.