Recently in a club I used to play in before I moved I saw that a provisonial rated player rated in the 200s beat a 1700, a 1900, and draw a 2100. Should the uscf come up with a rating test to adjust players ratings so that they do not have to wait weeks or even months to have the proper rating.
What kind of a ratings test do you have in mind, something other than playing in tournaments?
Did you follow that player’s rating? Normally, in that situation, a player will not need to wait weeks or months.
Bill Smythe
So far in 2017, we’ve had 16 players, all with provisional ratings under 1200 based on at least 4 games, earn 500 or more points in a single event, the largest gain being 737 points. Last year there was someone who saw a ratings increase of 879 points in a single event.
Someone who is very underated (usually due to having a stale rating accompanied by serious study or play outside of the US Chess ratings system) can catch up fairly quickly. However, it’s not clear to me that the original poster was referring to someone playing in a US Chess rated event.
Nice stat to know. The test should test a players opening ability, tactics and endgame skills to determine a rating level. Then based on their score their rating can be adjusted properly. Provisional K factor could also be raised a little so players have a more accurate rating once the player the right amount of required games.
But how often would a player have to take this test?
The player you mentioned has a rating. At the point they started playing tournament chess, they were terrible. They earned a 200 rating.
I have a rating. If I skip a year of chess, should I have to take the test to get a new rating? Maybe I should do that on purpose, to get a higher rating faster, instead of plodding through, beating my local players?
My guess is you’d have to ask a lot of questions, probably at least a hundred, in order to get an accurate assessment of one’s chess skills via a quiz. And you’d have to test those questions on thousands of chess players in order to validate the test instrument.
Who’s going to fund this research? Who’s qualified to develop the test?
elometer.net already exists. I’ve never been able to sit through the whole thing to see how accurate it is compared to my US Chess rating. 76 problems.
So, who created elometer.net and how was it validated?
If you don’t have the patience to take a 76 question test, and I probably don’t have it either, how useful would it be? My guess is I could play a few blitz games with someone and give you an idea of his rating strength. (Certainly enough to tell if someone plays at a 200 rating or a 1700 rating level.)
i’ve wondered about that also, starting “fresh”. beat a few 1800s and 1900s and come out with a master rating. is that possible?
…scot…
No answers. Just saying that it’s a thing.
If the best answer is always a strong and winning move then merely posing the question skews the answer because the person taking the test knows there is something there (in a tournament an expert or master who simply stops to look at a position between C-players can trigger the C-player on move to look for, and find, the normally hard-to-see tactical shot).
You’d need a number of questions with comparatively boring answers, which unfortunately triggers a person to not want to bother finishing it.
Ok. I took the elometer.net test.
I took it mostly seriously, but not quite as seriously as I would a tournament position. There were 76 problems, after all.
It scored me at 2178 with a 95% confidence interval. Range would be 2047-2308.
My current US Chess rating is 2054.
If someone wants a low rating for sandbagging purposes, what stops them from deliberately tanking the test?? If someone suffers from tournament nerves but is a good test-taker, that person will be over-rated.
I think we just have to accept that someone who builds up skills outside any of the major rating systems will play a much stronger game than his rating will indicate. Such players will win some stuff until their ratings catch up. Big deal.
The idea that we should come up with something to do what ratings do without doing it is a bit silly.
I hear its by a guy named Arpad Elometer. 
LOL!!
Why not play 50 or 75 games against one of the Fritz products in friend mode at a consistent time of say G-90 (or whatever time you want an approximate rating for) with Fritz playing fast. See what your handicap comes out as and compute your rating based on handicap and the rating of the particular Fritz engine that you’re using? Repeat with a different engine and compare results. I’m not certain that the friend mode will work with a non-Fritz engine.
The last time I tried this sort of thing with an engine, it felt like I was playing a strong master who every now and then let an idiot move for him.
Is there evidence demonstrating that such ratings correlate well with actual OTB ratings?
I don’t know of anyone who has looked for a correlation. I’d be curious to see if there is one and how strong the correlation is.
As to earlier computer chess engines, I do recall that Chessmaster 2000, or one of the CM iterations, would do what you mention, but it would also do the reverse. You’d be playing against it at a certain level and have an advantage coming out of the middle game only to have the program put on the afterburners and totally smoke you. I’ve not experienced that with Fritz 12. I’ve played a hundred or so games against it in Friend mode and my handicap has come down; very slowly at first, but more quickly since beginning lessons several months ago with a coach.
This entire thread is ludicrous. The rating system presently measures results. What better way is there to estimate a person’s likelihood of winning than by looking at results?
Bill Smythe