30 Apr 08
Originally posted by ExumaFirst see how hard it is to get to 2000, and then find out how hard it is to get to 1000. Now do you see why?
That is number of pages of players below 1000 RHP rating vs 5 pages of over 2000 RHP rating. Yes, I have no moves to make, resisting the urge to add a bunch of games. Patience keeps the gameload manageable...
05 May 08
Originally posted by ExumaStill, I think his point is that the list does not follow a non-standard normal distribution. It would be interesting for someone to plot the frequency distribution to see what it actually looks like. Like the poster, I just asumed the distribution followed a normal curve.
That is number of pages of players below 1000 RHP rating vs 5 pages of over 2000 RHP rating. Yes, I have no moves to make, resisting the urge to add a bunch of games. Patience keeps the gameload manageable...
Of course, if you think about it the curve couldn't be normal since everyone starts at 1200, and it takes a long time to build a rating. I would like to know, however, if the frequency of play has anything to do with rating. Not that the most active are the highest, but maybe those who have very low scores quit playing thus skewing the scores.
This is all guess work. We can only tell if someone will go to the trouble to plot it.
05 May 08
Originally posted by petrovitchJust because there are more people under 1000 than there are above 2000 doesn't imply that the distribution is not normal...
Still, I think his point is that the list does not follow a non-standard normal distribution. It would be interesting for someone to plot the frequency distribution to see what it actually looks like. Like the poster, I just asumed the distribution followed a normal curve.
Of course, if you think about it the curve couldn't be normal since everyone s ...[text shortened]... res.
This is all guess work. We can only tell if someone will go to the trouble to plot it.
For example, if the distribution was normal with mean 1300 we would expect there to be more people under 1000 than above 2000...
But, yes, I think you are correct in sayin that the distribution here is probably not normal due to the effects of abandoned accounts etc.
I have a feeling the different ways that provisional and non-provisional ratings are calculated would probably have an effect too.
07 May 08
Originally posted by monteirofYes, but your rating can drop a long way in 100 days...
Rankings updated every hour.
Provisionally rated players are excluded from the rating table.
Provisional ratings (< 20 rated games complete) prefixed by a 'p'.
Non-movers in the last 100 days are not ranked.
And the provisional rating system will cause an affect because it means that not all players are starting with the same rating when they enter the table... And because provisional performance is capped at 400 above or below your opponents rating it is as much a function of the ratings of your opponents as it is an indication of your playing strength
11 May 08
1 edit
Originally posted by droflaceC:\htdocs\Personal Chess Training\statistics.html
Just because there are more people under 1000 than there are above 2000 doesn't imply that the distribution is not normal...
For example, if the distribution was normal with mean 1300 we would expect there to be more people under 1000 than above 2000...
But, yes, I think you are correct in sayin that the distribution here is probably not normal due ...[text shortened]... that provisional and non-provisional ratings are calculated would probably have an effect too.
I compiled a list of rhp site player statistics. You may find a current copy at the above we address. I'll try to keep it current. I can't put the script on-line because it takes several minutes to execute.
The arithematic mean is 1320 with a standard deviation of 236. 1303 is middle of the road while 1209 is the most common rating in frequency.
From the Z-Scores we can tell that most players range between 1000 and 1600. 2000 is 2.89 standard deviations from the mean and 2400 has a Z-Score of 4.58 Players above 2400 are 0.02 percent; that's not 2 percent, but 2 hundredths of one percent.
STATISTIC VALUE
N 19374
Mean 1,320
Standard Deviation 236
Median 1,303
Mode 1,209
Mean Absolute Deviation 184
RATING RANGE FREQUENCY PERCENT
2400 2600 3 0.02 %
2200 2400 30 0.15 %
2000 2200 139 0.72 %
1800 2000 439 2.27 %
1600 1800 1,592 8.22 %
1400 1600 4,214 21.75 %
1200 1400 6,776 34.97 %
1000 1200 4,744 24.49 %
800 1000 1,334 6.89 %
600 800 0 0.00 %
RATING Z-SCORE
2400 4.58
2200 3.73
2000 2.89
1800 2.04
1600 1.19
1400 0.34
1200 -0.51
1000 -1.36
800 -2.20
600 -3.05
11 May 08
Originally posted by Supermanlol! True, though I am terrible at blitz, I don't see the tactics fast enough. Its fun though.
You know you can play Blitz while waiting.
To Petrovitch - thanks for the work you put in - that is a really interesting project - I too was under the impression that the number of high ranked (over 2000) players would be unnaturally high - especially given the amount of computer accusation that goes on. But looking at it statistically, very few by percentage are in that category. An interesting question would be - how many games on average it takes for a player to reach 2000 here, ignoring actual time. (Not trying to make work for you :-)
11 May 08
Originally posted by ExumaThat question can be answered in three parts.
lol! True, though I am terrible at blitz, I don't see the tactics fast enough. Its fun though.
To Petrovitch - thanks for the work you put in - that is a really interesting project - I too was under the impression that the number of high ranked (over 2000) players would be unnaturally high - especially given the amount of computer accusation that goes on. ...[text shortened]... es for a player to reach 2000 here, ignoring actual time. (Not trying to make work for you :-)
First, the number of games played is not correlated very highly with ratings. This is because many weak players play thousands of games an never improve their game. The don't play to improve; they play for fun.
Second, a player who is serious about his game will continue to impove throughout his/her life, but the rate of acceleration is dependent on how much they study, how they study, their personal experience, and their ability to learn and comprehend. It takes several hours a day of intensive study, not playing games, but study, to get there.
Third, players already of that strength, about 300 games.
12 May 08
3 edits
Originally posted by petrovitchI stated there was no correlation between a player's rating and the number of games played. I guess I'd better back that up with evidence. Remember, these figures are real-time so while the mean, standard deviation, etc. should remain very close they may vary from one examination to another. So this mean may be slightly different from previous means examined in other threads. It changes with every game played on this site.
That question can be answered in three parts.
First, the number of games played is not correlated very highly with ratings. This is because many weak players play thousands of games an never improve their game. The don't play to improve; they play for fun.
Second, a player who is serious about his game will continue to impove throughout his/he ...[text shortened]... games, but study, to get there.
Third, players already of that strength, about 300 games.
N MEAN STD R R2 INTERCEPT SLOPE
19,392 1,320 236 0.0710 0.0050 1,311.4071 0.0362
This simple linear regression shows that the equation of the straight line is:
Rating := 1311 + 0.0362 * Number of Games
So the slope show that your rating will increase by 0.04 points for every game you play, but these figures are not reliable. The correlation between a player's rating and the number of games played is 0.07 It needs to be 10 times that much to be significant. And the coefficient of determination, r2, is only 0.0050 meaning that only 1/2 of one percent of the variation between mean and the realtionship of these two numbers can be explained. So, while the number of books read, number of hours studied, etc. may tell us something, the number of games played tells us absolutely nothing about a player's rating or the acceleration of his rating.