20 Jul '12 06:31Do you carry a field in the database that flags a loss due to a timeout?
It would be great to be able to see the calculated rating of an opponent including timeouts and a rating excluding timeouts. Or at least something like "Games Lost to Timeout".
It often happens that you play against someone with a rating of 900 or something like that, expecting that level of player, where in actual fact the player rating is only so low because of game timeouts due to a period of inactivity.
In today's busy lives, we go through phases where we simply cannot get to the chess site for a week, or a month, because of work pressure. This sometimes causes all active games to timeout which drops your rating a lot. This result is correct because you lost the game and therefore your rating should drop. But for your future opponents it would be useful to know this. So they know what to expect.
I also appreciate that this might be something that could be used to manipulate ratings. If you see you are going to loose, you just let the game timeout, which gives you a good rating when timeouts are ignored.
So to counter this I would suggest not just excluding all timeouts when calculating the "rating excluding timeouts", but rather looking for a pattern where during a contineous period of time, all games for the user timed-out, not just a random one or two. This would indicate a nonselective timeout, which is the actual scenario where the timeout skews the rating.
It would be easy to identify those "periods of timout" by looking the the time difference between one logon, and the next. If the time difference between one logon and the next was more than a week/month/2months (whatever you think is apprepriate) then those timeouts are eligible to be ignored.
Any takers for this suggestion?
It would be great to be able to see the calculated rating of an opponent including timeouts and a rating excluding timeouts. Or at least something like "Games Lost to Timeout".
It often happens that you play against someone with a rating of 900 or something like that, expecting that level of player, where in actual fact the player rating is only so low because of game timeouts due to a period of inactivity.
In today's busy lives, we go through phases where we simply cannot get to the chess site for a week, or a month, because of work pressure. This sometimes causes all active games to timeout which drops your rating a lot. This result is correct because you lost the game and therefore your rating should drop. But for your future opponents it would be useful to know this. So they know what to expect.
I also appreciate that this might be something that could be used to manipulate ratings. If you see you are going to loose, you just let the game timeout, which gives you a good rating when timeouts are ignored.
So to counter this I would suggest not just excluding all timeouts when calculating the "rating excluding timeouts", but rather looking for a pattern where during a contineous period of time, all games for the user timed-out, not just a random one or two. This would indicate a nonselective timeout, which is the actual scenario where the timeout skews the rating.
It would be easy to identify those "periods of timout" by looking the the time difference between one logon, and the next. If the time difference between one logon and the next was more than a week/month/2months (whatever you think is apprepriate) then those timeouts are eligible to be ignored.
Any takers for this suggestion?
