PDA

View Full Version : Ratings Central is back



SofaChamp
05-08-2020, 02:00 PM
Just saw that RC has processed a whole load of old tournament results and thus should probably be able to give us a fairly decent list of pro players' actual playing strength.

Here is the current top ten:
21368

Seems about right?

zeio
05-09-2020, 09:48 AM
Had a good laugh.

Ovtcharov should be way down. Harimoto behind LYJ? Come on.

WCQ and LJK. How long did these 2 spend in the qualification stage?

Honestly, the ITTF's one is more like it, except for Calderano and Falck.

Same for the women's. Ishikawa at 16, above CIC at 18 and FTW at 20? Seriously?

SofaChamp
05-09-2020, 04:15 PM
If you look at the points you'll see that Lin Jun-Yu is only 10 ahead of Harimoto = in a match between the two the form of the day will decide.

And of course there are limitations to this system but as far as I know it's the closest we have to a list that tries to give players' current strength. In particular the fact that club matches are also used here makes this very interesting since players not active on the world tour but playing professionally at club level are included.

zeio
05-11-2020, 08:22 AM
Because of the way Ratingscentral work (https://mytabletennis.net/forum/forum_posts.asp?TID=86547&PID=1076616&title=rating-central#1076616), especially the assumptions on matchup which is a major factor in pro table tennis, and also the fact that CNT players break the curve, making direct comparisons between players are inherently difficult, where the mean and the SD have to be considered together (https://www.ratingscentral.com/PointChangeExamples.php). So instead of the point difference, I'm more concerned with their relative positions because a ranking system is supposed to properly reflect the clues drawn from what a rating system is designed to do - predict the outcome of a match based on past results.

LYJ has never beaten Harimoto in their 3 formal encounters(disregarding the T League, where Harimoto was losing left and right, and Ratingscentral doesn't take that into account either). Putting LYJ above Harimoto suggests those 3 losses were upsets, which is counter-intuitive for a system based on belief(probability). OTOH, under the ITTF system, LYJ was only 50 points behind in February and would've knocked Harimoto off the 5th spot had he closed out the match against Mizutani at the German Open. Hence, the ITTF system also suggests they're close in strength, given their performance in the past 12 months.

Same deal for Ishikawa, CIC and FTW. Ishikawa has always had trouble playing these two. She still has a winning record against them, but her recent encounters were losses, so the ITTF's merit-based system better reflects their recent performance. In CIC's case, her most recent 2 wins came after she started working with Zheng Jiaqi, which also helped her against Ito, despite losing 3 out of the last 4. In FTW's case, she also beat CM last year just like Ishikawa did.