This should definitely be incorporated into the system:
Instead of deciding how many points are won or lost after the game is over, it shoudl be decided at the start of the game calculated by the two players rating at that time. Then the points that are gained or lost should stay constant through the game, regardless of rating changes that are bound to happen during the game. This would solve the problem of those that carefully chose when to resign games and things like that.. Also, if a player decides to drop all his games and has a game with you but now has a rating of 900.. the points that you should have gained will still be intact, and you won't end up losing more points than you should if you do lose. Just a though, makes a lot more sense, afterall. a new rating is not calculated mid tournament in otb or anything like that.
Originally posted by ZumdahlNO NO NO NO!!!!!!!
This should definitely be incorporated into the system:
Instead of deciding how many points are won or lost after the game is over, it shoudl be decided at the start of the game calculated by the two players rating at that time. Then the points that are gained or lost should stay constant through the game, regardless of rating changes that are bound to happ ...[text shortened]... ore sense, afterall. a new rating is not calculated mid tournament in otb or anything like that.
people would choose when to start a game to manipulate their rating.
imagine starting 50 or so games against dustin when he is rated 1970. Then time him out in all 50 games when he later plummets to a 800 rating - ironman31 would no longer be our best 😲
the way it is now is not perfect, but what you suggest is much worse.
i used to play at another site and the system was the following.. the rating used is higher of the two ratings -- when the first move was made and when the last move was made in the game -- is used when calculating the rating difference for both opponents in the game. For example, if player A plays against player B and their ratings are 1300 for player A and 1400 from player B when the game is started, but just before the game ends their ratings are 1250 for player A and 1450 for player B. The system will use 1300 for player A (1300>1250) and 1450 for player B (1400<1450) when calculating rating adjustments for this game, which will be applied to their current ratings.
Originally posted by bazik and skipslotthis method would lead to us all increase our ratings!!!!!
......rating at the beginning or the one at the end of the game, whichever is highest..............
eventually we would have players rated 4000 ... whaat a messssssSSSS?!
slightly better would be to average the two.
but our present setup is much better still:
if we win against a player who is just being timed out hundreds of times then who cares.
if we beat dustin when he is keeping up with his games then give us good credit!
Originally posted by flexmoreNot true... one player would go up by the same amount that the other player goes down (as it is now).
this method would lead to us all increase our ratings!!!!!
eventually we would have players rated 4000 ... whaat a messssssSSSS?!
slightly better would be to average the two.
but our present setup is much better still:
if we win against a player who is just being timed out hundreds of times then who cares.
if we beat dustin when he is keeping up with his games then give us good credit!
Just the number of points that the change would be would be based on the higher of the ratings at the start and end, instead of just what they are at the end.
Richard.
I agree that taking max(start rating, end rating) independently for each player as the input to the adjustment calculation, then applying the adjustment to each player's end rating would probably be better than what we have now. I am sure it would be better than using the start rating - that would be even more unfair on people who play improving players than the present system can ever be to those who play people whose ratings fall.
This issue has often been discussed, but this is the first time I have seen this scheme proposed. Russ/Chrismo - if you're reading this, what do you think? Some thought would need to be given to players who had a provisional rating at the start of the game. I think that in that case only the end rating should be used.
It has indeed been discussed many times: the clue is in the thread title:
Manipulation noun Shrewd or devious management, especially for one's own advantage
Using anything other than the end rating allows for gross advancement of one's own rating: e.g. An 1800 player delibertely looses until they are down at 1000, then starts 40 games vs 1200 players, wins all games at 40 points a game: new rating 2600.
Easy isn't it? Using no false accounts and entirely within the rules. And for that reason, an absolute non-starter.
Originally posted by ToeUsing the max between start and end rating would never alow that!!
It has indeed been discussed many times: the clue is in the thread title:
[b]Manipulation noun Shrewd or devious management, especially for one's own advantage
Using anything other than the end rating allows for gross advancement ...[text shortened]... ly within the rules. And for that reason, an absolute non-starter.[/b]
cause there wouldn't be 40 points to each game he would won against those 1200 players.. because each time he wons a game.. the ranking goes up.. meaning that the more he win... against the 1200's the less his ranking goes up.. eventually going above the 1600 and he would start losing points for the wins!
Originally posted by bazikIndeed true: I was on about the initial proposal of "assign points at the start". The thread has modified over its history, but there were an unsafe number of recs on post #1.
Using the max between start and end rating would never alow that!!
cause there wouldn't be 40 points to each game he would won against those 1200 players.. because each time he wons a game.. the ranking goes up.. meaning that the more he win... against the 1200's the less his ranking goes up.. eventually going above the 1600 and he would start losing points for the wins!
Using the maximum player ratings over the period of the game could be a valid system: As you can only win one game at a time then, as you say, a "cheat approach" would have diminishing returns, quickly falling to very few points/win.
Even the max of start/end ratings would provide some protection against big rating swings.
Originally posted by gumbieI'm pretty sure that to avoid manipulation, some kind of maximum is needed, either max of start and end, or max during period.
I quite like the system as it is now.
But I think using the average between start and finish would certainly be better than using the max between start and finish.
Max of start and end would be better than the present system, max overall might be a bit costly to compute.