I was thinking about this the other day, "What is the win loss record of a 1300 vs. a 1200?" If a player rated X plays another player rated X-100, how often could the former be expected to win? I plan on randomly looking at games where players play opponents rated 100 points below them, and see the win-loss ratio.
I haven't looked yet, but my hypothesis is that a player rated 100 points above his opponent is twice as strong; he should win 2 out 3 times.
If this was true, then I(1200) should be able to beat a 1300 1 out of 3 times. However, different people have different circumstances, so my results may not be accurate.
What do you think? For every hundred points your ratings increase, do you believe you are twice as strong?
If you were ranked 1200 and you played someone ranked 1300 and drew the game you were expected to achieve 0.362 so you would gain a few points.
Initial rating: 1200
Progress coefficient: K=36
Game #1: opponent=1300, result=0.5, expected=0.362, ELO gained=+4.97
New rating: 1205
Performance: 1300
Originally posted by AmauroteA rating differance of 100 points at any rating should produce statistically the same result (i.e. on 100 games the higher rated player should score 67.5 points from 100), irrespective of whether the ratings are 1000 and 1100 or 2200 and 2300, shouldn't they?
The difference is immaterial between ratings lower than 1700 - the gulf between 1700 and 1800 is considerable, but the difference between 1700 and 1600 is minimal.
Originally posted by Dragon FireIf you beat someone 3-2, what would be the rating difference? I was curious. I had a match like this the other day, in Chinatown, Honolulu. At least, I won the match though. But, this guy was really good. I had to do my best.
A rating differance of 100 points at any rating should produce statistically the same result (i.e. on 100 games the higher rated player should score 67.5 points from 100), irrespective of whether the ratings are 1000 and 1100 or 2200 and 2300, shouldn't they?
Originally posted by Dragon FireI can only speak subjectively here, frankly, but I personally feel that the rating gap between 1600 and 1700 is much less material than between 1800 and 1700 - there's a theoretical gap between the latter, whereas earlier ratings seem to be determined more by concentration, gameload and psychological factors. I'm sure many people will beg to differ, but I can only vouch for my own personal experience on the site.
A rating differance of 100 points at any rating should produce statistically the same result (i.e. on 100 games the higher rated player should score 67.5 points from 100), irrespective of whether the ratings are 1000 and 1100 or 2200 and 2300, shouldn't they?
Originally posted by powershakerScoring 60% longterm against someone would suggest you are about 50 Elo points higher than they are.
If you beat someone 3-2, what would be the rating difference? I was curious. I had a match like this the other day, in Chinatown, Honolulu. At least, I won the match though. But, this guy was really good. I had to do my best.
As for where 100 point gaps really make a difference...well, from a standpoint of "how often will player x beat player y," it doesn't matter (much) what the exact numbers are, just the difference in ratings. But rating points are definitely harder to get as you move out of the meaty part of the bell curve and into the extremes (particularly 2000+). It's a lot easier to go from the top 70% of all players to the top 50% than to go from the top 5% to the top 1%.
Originally posted by AmauroteA 1600-player against a 1700-players will have the same win/loss ratio than 1000-player against a 1100-player, or a 2000-player against a 2100-player. Statistically speaking.
I can only speak subjectively here, frankly, but I personally feel that the rating gap between 1600 and 1700 is much less material than between 1800 and 1700 - there's a theoretical gap between the latter, whereas earlier ratings seem to be determined more by concentration, gameload and psychological factors. I'm sure many people will beg to differ, but I can only vouch for my own personal experience on the site.
But it is harder for a 1600-player to get to 1700 than it is for a 1000-player to get to 1100.
Originally posted by FabianFnas.... and it it well nigh impossible for a player who has been between 1800 and 2000 for decades to get to 2100.
A 1600-player against a 1700-players will have the same win/loss ratio than 1000-player against a 1100-player, or a 2000-player against a 2100-player. Statistically speaking.
But it is harder for a 1600-player to get to 1700 than it is for a 1000-player to get to 1100.