There has been some discussion about applying a rating system like chess has, the Elo rating, to badminton. Now, I'm not an expert on rating systems, but for the fun of it, I calculated similar ratings based on Fargo Rating system. It's pretty close to Elo rating (and Go's rating system) and it works on a similar principle. Such a rating system gives every player a numerical rating and based on these ratings, you can calculate the probability of the most likely outcome of any two players' match. If the rating difference is exactly 100 points, then the better player is twice as likely to win any given point. In other words, given two players with a 100 point rating difference in a badminton game, the most likely outcome is 21-10 points (or, if you want to be exact: 21-10.5). The probability of a player winning a single point can be calculated from the equation = 1 / (1 + 2**(rating difference/100)). Obviously, extrapolating this to a whole match is problematic, but I hope these numbers can give some interesting insights anyhow. So I added most of the major tournaments since the last World Championships (including the WC) and calculated the Fargo Ratings of the professional male players. This is how the top looks like: 593 (700) Chong Wei Lee 578 (700) Dan Lin 577 (700) Long Chen 573 (700) Jin Chen 550 (700) Simon Santoso 549 (700) Ajay Jayaram 548 (700) Sho Sasaki 548 (700) Peter Hoeg Gade 543 (700) Kenichi Tago 542 (700) Jan O Jorgensen The first is the rating, second is robustness, ie. the number of points played. (The robustness is cut at 700. The bigger the robustness, the slower the rating changes.) Now, first thing to note is that if you take the rating difference of LCW and LD/CL, the expected outcome of a single game is around 21-19 in LCW's favor. So that's a pretty tight in expected outcome. A fifty point difference in this scheme would predict 21-15 as the most likely outcome. Chen Jin's rating is pretty high, considering his record against the top three players. I would assume most wouldn't put their money on him against LCW or LD, even with favorable odds. But if you look at his results, he's still pretty consistent against most players and plays pretty tight matches even against the top three. The odd one out here is Ajay Jayaram as his current ranking is around 27th, but his rating suggests that he "should" be ranked higher. Again, if you look at his results, the rating should not come as such a big surprise anymore. In the major tournaments, he has usually lost in pretty tight matches against the top players. Let me know what you think about the ratings. And ask for more details about the system if I didn't explain clearly enough. Below is a longer rating list. 593 (700) Chong Wei Lee 578 (700) Dan Lin 577 (700) Long Chen 573 (700) Jin Chen 550 (700) Simon Santoso 549 (700) Ajay Jayaram 548 (700) Sho Sasaki 548 (700) Peter Hoeg Gade 543 (700) Kenichi Tago 542 (700) Jan O Jorgensen 539 (700) Yun Hu 538 (556) Tommy Sugiarto 538 (700) Hans-Kristian Vittinghus 536 (700) Kashyap Parupalli 536 (419) Gurusaidutt R. M. V. 536 (700) Wan Ho Shon 535 (700) Tien Minh Nguyen 535 (700) Hyun Il Lee 534 (211) Anand Pawar 531 (240) P Kashyap 530 (642) Viktor Axelsen 529 (700) Wing Ki Wong 528 (700) Zhengming Wang 526 (700) Pengyu Du 525 (700) Kazushi Yamada 524 (499) Dionysius Hayom Rumbaka 524 (374) Takuma Ueda 521 (308) Daren Liew 520 (401) Tanongsak Saensomboonsuk 519 (485) Muhammad Hafiz Hashim 517 (700) Marc Zwiebler 516 (700) Taufik Hidayat 509 (261) Zi Liang Derek Wong 509 (700) Pablo Abian 507 (216) Choong Hann Wong 504 (700) Rajiv Ouseph 503 (426) Brice Leverdez 501 (209) Joachim Persson 501 (369) Boonsak Ponsana
Now we know the feats of concentration and diligence that Finns are capable of during the short dark days of winter. We knew Finland could produce more than Sibelius, reindeer and Molotov cocktails. I admire the concentration and effort you put into this system. But to assign numbers to the infinite variables of the human experience seems slightly 'mechanical' to those who love the art and unpredictability of the game. I mean absolutely no offense, but if this scale is accurate, the only people who would be really pleased would be those who know little about the game but love betting money. Look forward to other points of view.
Well, to be honest, that's a pretty accurate assessment. That said, my own hope that such numbers can reveal something non-obvious or give confirmation to hunches. As an example, I think the tight rivarly between LCW, LD and CL is mostly because they are pretty evenly matched. These numbers suggest that if you look all the results included in this sample, LCW is slightly above LD/CL, but it's so close that it could go either way. In other words, it's not that LCW is way better than, say, CL and happens to lose against him because of the mental game. The numbers suggest that CL is so close to LCW's skill level that it is no surprise that the matches go either way.
Many of the points raised by my thread on Elo apply here. I'm particularly interested by the guy who is 27th, but possibly should be much higher. The current system used in many sports does bias things towards those who are already in the higher rankings. For example, 9th to 16th in the world avoid the rest of the top 16 for 1 round longer that 17th to 32nd. Therefore, they tend to go 1 round longer, get more points & retain their position. Elo doesn't have a ratings uncertainty/robustness, but I've heard of a Glicko system which does apply one. Essentially, the more games you play, the tighter your rating becomes. Conversely, the more rating opportunities that you miss (e.g. through injury), the wider your rating becomes. Whether this form or rating is better or worse than what is currently used is down to opinion. Elo et al. favour consistent players who do well against most of their peers. Round-based points favour the guy who is occasionally brilliant.