Chess.com cannot do arithmetic

???

(135,658)/(685,069)  = 0.198 approximately.

i.e Percentile: 20%

54.5% IS a percentile.  They never specified what it is a percentile of.

That would mean 54.5% of users have a lower rating that you. I think it is correct.

Maybe it is correct, but then the Global number (which in other time limits seems to be the number of active players in that category ) would be wrong.  What DO these numbers mean?

Do people actually look at that and care about it?

Click on the link for global and look at the rating distribution. 1100 falls really close to the middle of the curve. That is what the percentage is referring to. Lots of people fall on the same percentage band on the curve.

Okay. Do you also know what the numbers 135,658 and 685,069 refer to?  "Global" and  a hash symbol are not completely self -xplanatory.

Well, the larger number would be the number of players in that pool. I would assume that is active players, though I don't know that for sure, especially since I have numbers for bullet in that screen and I don't play it.

The smaller is your place on the list of players for that pool. Basically, the leaderboard location.

If that is true Martin, then the calculated percentile is wrong.  Either iot ]has been calculated wrong, or the numbers do not mean what they seem to mean.

I am not really too fussed. I'll just accept that I cannot make sense of these numbers in the case of 1 minute bullet. They make sense for my Blitz and 3-day ratings.

I'm pretty sure the percentile is your postion in the rating distribution. 

https://en.wikipedia.org/wiki/Percentile_rank

You are probably right, but that does not explain the figures. I don't care about my position in the bullet rankings  In any case it will be a lot higher in a few months time, when I get used to playing at that speed. 

I would like to understand the information that is presented in our statistics.  At the moment I do not. There are unexplained anomalies.

From what I can determine, Chess.com no longer calculates individual percentiles.  I have noticed that my percentile only changes when my rating crosses a plateau (1500, 1600, etc.)  When I was below 1500, my percentile was about 55%* regardless of the actual rating.  When I crossed 1500, it went to 85%.  It remained there until I crossed 1600, when it changed to 91%*.  When my rating dropped below 1600, it went back to 85%.

It would be nice if they would explain what the percentile means.

*I'm not sure that these are the exact numbers.  My memory isn't that good these days. 

I'm fairly confident the percentile is basically what is in my post 13.

The ranking numbers are most likely based on the leaderboard listing, though they may not update in realtime (which makes a lot of sense computationally). I also think they are based on recent activity. My rank and percentage is currently N/A but I haven't finished a Daily rated game in over 90 days .

The reason I believe there is some lag is that the person in 20th place on the Daily leaderboard show 38th on the profile.

Martin

According to my stat sheet, my current ranking is 64K out of 933K.  That's a percentile of 93%.  My stat sheet shows, 85.5%.  Not only that, for the last month or so, my rating has ranged from the low to the high 1500's.  My percentile has remained at 85.5% regardless.  Doesn't make a lot of sense.

Did you look at the distribution on the Global Leaderboard page? Percentage is based on what part of that graphic your rating falls into, not a direct numerical calculation of 100 - [ (Rank/Active)*100]

Take a read through the article I linked. It explains percentile as it pertains to a distribution fairly well. It also explains how the 1500-1599 rating band could fall into the 85.5 percentile.

Martin

My rating just dropped from 1507, where I was in the 85.5th percentile to 1499 where I am in the 76.6th percentile.  I can't believe that an eight point drop in rating would change the percentile that much (note that I crossed the 1500 plateau when the percentile changed).  Before the drop 85.5% of the pool had equal or lower ratings.  After the drop, 76.6% of the pool had equal of lower ratings?  That's a big change in such a large pool.

Also, as I noted previously, as long as I remained in the 1500's, the percentile remained at 85.5%.  Does that make sense?  If my rating is 1507, 85.5% of the pool has an equal or lower rating.  If my rating is 1599, 85.5% of the pool has an equal or lower rating.

Did you look at the article or how the distribution graph gets sliced/banded to calulate percentiles? It is pretty easy to see from that how what you are asking is possible.

I'm not an expert at distributions and all that goes along with it, but the explanation of how it works seems decent enough.

 I think that it goes back to what I said earlier.  Chess.com does not calculate an individual's percentile.  It groups the players, and the percentile is the group's, not the individual's.  When I dropped below 1500, I switched groups.

I think that the following link gives a better definition of percentile, and explains why some of us are confused.

http://statistics.about.com/od/Descriptive-Statistics/a/What-Is-A-Percentile.htm

Yeah, it isn't super clear, especially for people not familiar with the concept (that's why I looked it up), and probably could use a help article.

No offense, but it's just division.

I think the answer is simply chess.com isn't using that global number in the calculation. The global number is probably much larger, and used for advertisement, than e.g. the number of active players. That's my guess.