|
Statistics Help | |||||||||||
This page has questions and answers about how Pocket-Monkey's statistics work. Statistics QuestionsStatistics AnswersQ. What statistics does Pocket-Monkey track? For each kind of game, we track the number of those games a player has played, how many of those the player won, how many he/she lost, and how many ended in a draw (for games where draws are possible). Additionally, we maintain an Elo rating for each player for each kind of game. [Back to Question List] [Back to Top] Q. Why can you cancel games with only a few turns? For the first few turns (the number varies depending on the kind of game), we allow cancelling the game so that if two players start playing and then decide they don't want to play together, they can leave the game without having it count against them. This is particularly important with open invites because the players are unlikely to know each other, but we allow it regardless of how the game started. [Back to Question List] [Back to Top] Q. What happens with expired games? If a player fails to make a move within the time limit of the game, that player loses the game, and it's counted just like any other loss: That player's number of losses goes up, the other player's number of wins goes up, and their Elo ratings are adjusted appropriately. [Back to Question List] [Back to Top] An Elo rating is a numeric rating -- 1654, 1712, 1802, etc. -- which indicates a player's rated strength in a given game relative to the other players in the community. The higher the number, the stronger that player is believed to be. Typically, ratings are between 1400 and 2200, where 1400 is a quite low rating and 2200 is a very high rating indeed. Elo ratings were invented by Professor Arpad Elo in the 1970s for use in Chess, and are still used today by the World Chess Federation. But they're for much more than Chess -- Elo's rating system works for lots of different games, from Chess to Checkers to Croquet (the World Croquet Federation -- yes, there really is one -- uses Elo ratings). Elo's system is so successful because at its heart it recognizes that simple win ratios and such don't tell the whole story of a player's strength in a game -- a moderately good player can have a high win ratio simply by choosing opponents who are new to the game or who aren't very good. Similarly, someone who constantly tries to improve by playing opponents who are better than he/she is will tend to have a lower win ratio, which doesn't accurately represent his/her ability. To deal with that, when rating the result of a game, the Elo system adjusts each player's rating by a larger or smaller amount, based on the relative strength of the players. A strong player who wins over a weaker opponent doesn't get as much credit for the win as a weaker player who wins over a stronger one. The other success of the system is that it recognizes the impossibility -- or at least extreme difficulty -- of having an "absolute" rating for a player. Elo ratings are not absolute, they are relative, and they're only applicable within a body of players. Your rating within one group of people has no meaning within another group of people, because of the differences in the other players. (There are some exceptions to that -- for instance, where both groups use identical rating systems and there is a large degree of overlap between the players in the two groups. But in that case, you can think of them as one big group, so the statement stands.) [Back to Question List] [Back to Top] Q. How much will my rating change if I win a game? The short answer is: It depends. It depends on the relative strengths of you and your opponent, and it depends on how well the rating system knows you (e.g., how many games it's seen you play). Early on, you'll see your rating change by larger amounts (low double-digit figures) as the system gets to know you and tries to move your rating quickly to the general area where it should be. Later on, when the system knows you fairly well, you'll see your rating move less (in the single digit range). That's all assuming you're playing people with ratings similar to yours; you'll get more credit for wins if you play people with much higher ratings, and less credit when playing people with lower ratings. [Back to Question List] [Back to Top] Q. Do tournament games count in my stats and rating? Yes, tournament games do count in your stats and your rating. Right now it's all one rating (both tourney and non-tourney games); at some point we may break that down into more detail. [Back to Question List] [Back to Top] Q. I have an Elo rating from another site, can I use it here? No, because Elo ratings are not valid when you move from group to group, they are relative ratings within a group of players. See the discussion of Elo ratings above for details. [Back to Question List] [Back to Top] Q. I have an official Elo rating for Chess from FIDE, can I use it here? No, because your FIDE rating is only valid amongst FIDE players. See the answer to the "I have an Elo rating from another site, can I use it here?" question above and also the discussion of Elo ratings for details. [Back to Question List] [Back to Top] Q. What rating does a player start with on the site? 1600 [Back to Question List] [Back to Top] Q. My rating in a game type just changed without my winning/losing that type of game, why? Your rating may change during a maintenance interval. Periodically we have to go back and make a game (or a number of games) "unrated" as part of dealing with an account involved in cheating or ratings manipulation. When we've done that a few times, we recalculate the ratings from scratch during a maintenance interval, running a process that walks through all of the rated games ever played on Pocket-Monkey, in order, and building up the ratings information. This can have a knock-on effect on your rating even if you've never played a game with the account(s) involved in the affected games, or even if you've never played a game with someone who's played a game with the account(s) involved. Suppose Player A and Player B have a series of games, and then Player B plays with Player C, who later plays with Player D. At some point it turns out that those games between Player A and Player B should be made unrated (retroactively). That changes Player B's rating going into the game with Player C, which changes the amount of the adjustments to their ratings for the outcome of that game -- which of course changes Player C's rating going into the game with Player D, etc., etc. Usually these effects are quite small, a point or two, but sometimes when several games are involved the effect can be larger. The effect on your rating may also go either way -- your rating may go up or down depending on which players you've played with, and when. At the end of a recalculation, the ratings are accurate for the (updated) set of rated games the community as a whole has played. [Back to Question List] [Back to Top] Q. What's the difference between "rated" and "unrated" games? A rated game is counted in the statistics; an unrated game isn't. [Back to Question List] [Back to Top] Unrated games are handy in several situations. For instance, suppose Alicia is teaching Marcello how to play Chess. It's probably a bit unfair to have Marcello get losses on his stats when he's just learning the game, and so Alicia would probably invite Marcello to unrated games instead until he had the hang of it. [Back to Question List] [Back to Top]                                                   |