Just a bit of commentary: I would not allow players to fall below the level they started with. It's just frustrating to see new players start out with more points than you have. Even though activity might be rewarded too much currently, it is still something you *want* to reward. At least for new players.
For a game we implemented it with points starting with 0, straight ELO except for linear "newbie" protection from 0 to 400. So at 0 points, you can't lose points, whereas with 200 points, you only lose half as much as you would normally. Now I don't claim that this would be the ideal model for Poker. It is geared much more towards having longer-lasting ladders. But I feel that it works quite well.
And another thing: Giving 10% of what you have into the pool makes the whole progression exponential, whereas pure ELO is linear. In case the whole system stays within close bounds, this will probably be no problem. But for longer-lasting ratings, it might produce some strange effects.
@Maytsh: I know, that it would be nice to have something like an elo system. But you most elo systems, also the ESL systems, are designed for matches which have 2 opponents. It would be a lot of calculations to find a system which considers all results of a 10 players match. And you must have a good furmula for this procedure.
At equity poker we gave all players points for their ranking in a match. With a system close to the formula 1 system. 1st = 12 pts, 2nd 10pts etc.... So every place in a match earns points. the points were added and divided by the number of matches. this gave the score to the player. the system worked fine, but it also has its problems. in my opinion we should think in seasons, in which we use a points or elo system which is based on the results of the players. a system like this could be supplemented by special achievments winning streaks or example...
It isn't that problematic. For each player, there are 4 outcomes of the form "better", "tied" or "worse" given by whether the other player left the table sooner or later (if at all). So one just has to average over those 4 ratings to arrive at the effective rating. You can even calculate it for players the very moment they are leaving the table.
The administrator has disabled public write access.
Would be possible this way, but a lot of calculation. And i know the problems the esl had with their elo system. players in the top 20 just wanted to play against themselfs (top20). at a long time period motivation sunk and many players left the ladder (including me and my team). i think a (time-unlimited) ladder isn´t the right format for poker. i think we need something like a season. with several motivational items we should place into.
we also added a credit system to our league. so everybody, additional to the ranking, is able to challenge others and play for these credits. credits can be earned during the season and with the endresult.
now we (at equity poker) discuss to have weekly seasons where the results will be added to a global ranking which ends to the end of the year. so everybody has a long term motivation.
we had 3 seasons so far. 160 players are registered. and long seasons automaticly reduce motivation from the time when players think they can´t win the season. depending on the frequency the players play, the point came for some quite early, for others later. i think the best way to give a long motivation to the players is to have several new chances to become 1st without forgetting the results of the past which would be added to the global anual ranking.
the, let´s call them "mini-seasons", could base on something like an elo system.
On the Ranking site over at Poker-heroes, wouldn't be better if there was a minimum amount of wins in the "Won in Percent" category (i.e. minimum 20 wins) which should be rising as the number of games increases?
Thx for the suggestion, we switched now to your idea.
Well, calculation is cheap. Where one exponential function won't kill the server, four won't either
I must reiterate that I have no idea about where the specific problems of Poker lie. Our ladder actually seems to have the opposite problem: Top players only playing against low-ranked players.
So concerning your mini-season (I would actually call it tournament at that scale): I quite like the idea of aggregating a score out of these. You could associate it with some Poker-related metaphor like a bracelet or something. Have themed tournaments, maybe with special rules or more worthwhile "prices".
But on the other hand, ELO isn't really made for such sort "spurts". Real tournaments work as well as they do (or don't?) because they have strict control over which player plays against which. ELO and friends just have to deal with whatever you throw at it. Players can then always abuse the system by constructing cases where the system doesn't have enough information to properly judge the situation.
To give concrete examples:
a) Say all players start with 1000 points, no "newbie protection". Now new players are most likely overrated. So players looking for a top score will to try to play against as many new players as possible. Frustrating new players in the process that most likely won't understand why they are preyed upon.
b) On the other hand, starting at 0 points, with "newbie protection" you might get a bias towards group that play a lot amongst themselves, because every time someone falls under the "newbie limit", everyone gets a bit of a boost. Might be hard for new players to get into matches with good players, even though they can at least earn good points by playing with similar-ranked players.
ELO is still a lot better than most simple systems in these cases, mind you, but there will always be problems. It might take a bit of tuning to get things right (Personally I'd favor start with 0, protection until 100).
Plus: The shorter the seasons, the more aggressive the style of play will be. With the infamous instant Pre-Flop-All-In being at the lowest end (= one-game-tournaments). Also something to consider.