Friday, September 25, 2009

GOTW - Whackometer Week 4

Here are the results for the Week 4, Game-of-the-Week Whackometer, which attempts to see whether one (or more) of the GOTW judges are "whacko".

This week, Michael said , "I confess, I am a whacko judge". Do the statistics bear him out?
Name Week-4 || Average
Jeff 0.589 || 0.388
Greg 0.017 || 0.329
Arun 0.749 || 0.492
JimD 0.897 || 0.556
Mike 0.182 || 0.199
Remember that a number closer to 1.000 means that the judge agreed with the collective wisdom of the other judges, a number closer to -1.000 means that the judge disagreed with the collective wisdom of the other judges, and a number close to 0.000 means that the judge neither agreed nor disagreed with the collective wisdom of the other judges.

This week, three of the judges (Jeff, Arun, Jim) all were in pretty high agreement with the wisdom of the group. Michael was less so, and Greg had almost no relationship in his choices with everyone else.

Overall, Michael still is somewhat less correlated with the wisdom of the group than the other judges. So, he's edging closer to whackoville, but he's not there yet. There's still a ways to go.



On another note, I was interested in seeing whether any of the judges exhibited a bias toward one team or another in picking game of the week. For example, did "Storm McRainy" award most GOTW points to Seattle, or "Jack Daniels" give lots of GOTW points to Tennessee? This can also be done with correlations.

Fortunately, the results were rather uninteresting. All judges agreed with the other judges with correlations ranging from 0.743 to 0.849. That's pretty high, and suggests there's absolutely no bias from any judge. I'll keep track of this for a few more weeks, but I doubt anything will come of it.

USCL Ratings and Power Rankings: Week 4 (2009)

Another week in the USCL, and another set of power rankings and ratings.

As a note of clarification, I took a good long hard look at how I was creating the power rankings. In the previous system, I was using the following ratios to calculate power ranking -- 43% team record, 43% opponents' records, 14% opponents' opponents' records. The more I dwelled on it, and the more I studied the rankings it was giving me, the more I came to realize that there was too much stress on the strength-of-schedule, and too little on the actual record achieved; a win is a win, after all, regardless of who the opponent is.

Therefore, in this slightly altered formula of power rankings, the ratios are now 55% team record, 35% opponents' records, and 10% opponents' opponents' records. I don't plan to change it again this year... famous last words.

Given that, here are the rankings for post-week 4 in the USCL's 2009 season.

1 - San Francisco 1.000
2 - New Jersey 0.992
3 - Seattle 0.938
4 - Miami 0.850
5 - Boston 0.823
6 - New York 0.817
7 - Tennessee 0.736
8 - Dallas 0.729
9 - Arizona 0.721
10 - Baltimore 0.657
11 - Philadelphia 0.656
12 - Chicago 0.640
13 - Queens 0.629
14 - Carolina 0.493

Some interesting observations. New York is several places above Baltimore, even though New York's record is 1.5-2.5, and Baltimore's record is 2.5-1.5. This is highly due to the strength of schedule. New York has played New Jersey twice and Seattle once -- three weeks of tough opposition, while Baltimore has played lower scoring teams.



The ratings are presented below. Remember, to be included on the list, a player must have played three lifetime games in the USCL, as well as having played in the 2009 season.

Rank Rating Name
1 2655 Jaan Ehlvest
2 2638 Hikaru Nakamura
3 2621 Julio Becerra
4 2604 Sergey Erenburg
5 2600 Boris Gulko
6 2588 Larry Christiansen
7 2576 Dean Ippolito
8 2575 Joel Benjamin
9 2574 Marko Zivanic
10 2571 Jacek Stopa
11 2554 Gregory Serper
12 2546 Alex Shabalov
13 2545 Josh Friedel
14 2543 Giorgi Kacheishvili
15 2539 Yuri Lapshun
16 2539 Eugene Perelshteyn
17 2533 Alejandro Ramirez
18 2532 Pascal Charbonneau
19 2529 Alex Lenderman
20 2515 Marc Esserman
21 2505 Eli Vovsha
22 2505 Bruci Lopez
23 2492 Rogelio Barcenilla
24 2491 Thomas Bartell
25 2490 Levon Altounian
26 2478 Alex Stripunsky
27 2478 Jesse Kraai
28 2474 Sam Shankland
29 2459 Jorge Sammour-Hasbun
30 2459 Dmitry Schneider
31 2458 Daniel Ludwig
32 2449 Denis Shmelov
33 2449 Sergey Kudrin
34 2446 Slava Mikhailuk
35 2438 Jan van de Mortel
36 2435 Angelo Young
37 2435 Larry Kaufman
38 2434 David Pruess
39 2433 Oleg Zaikov
40 2431 Tegshsuren Enkhbat
41 2429 Mackenzie Molner
42 2419 Florin Felecan
43 2415 John Bartholomew
44 2414 Mehmed Pasalic
45 2414 Blas Lugo
46 2408 Mark Ginsburg
47 2407 Nikola Mitkov
48 2406 Jonathan Schroer
49 2405 John Donaldson
50 2402 Robby Adamson
51 2399 Keaton Kiewra
52 2391 Jay Bonin
53 2379 Shinsaku Uesugi
54 2375 Andrei Zaremba
55 2374 Vadim Martirosov
56 2374 Eric Rodriguez
57 2366 Yian Liou
58 2366 Ron Simpson
59 2364 Ron Burnett
60 2354 Victor Shen
61 2352 Bryan Smith
62 2351 Peter Bereolos
63 2349 Aviv Friedman
64 2344 Michael Lee
65 2344 Yaacov Norowitz
66 2340 Matthew Herman
67 2331 John Bick
68 2329 Bayaraa Zorigt
69 2315 Eric Tangborn
70 2311 Ilya Krasik
71 2306 Todd Andrews
72 2296 Jon Burgess
73 2294 Ray Kaufman
74 2269 Craig Jones
75 2261 Nelson Lopez
76 2257 Josh Sinanan
77 2256 Tsagaan Battsetseg
78 2250 Danny Rensch
79 2227 Alejandro Moreno Roman
80 2217 Leo Martinez
81 2207 Elizabeth Vicary

Friday, September 18, 2009

GOTW Whackometer - Week 3

Here are the results for the Week 3, Game-of-the-Week Whackometer, which attempts to see whether one (or more) of the GOTW judges are "whacko".
Name Week-1 Week-2 Week-3 Average
Jeff 0.792 0.319 -0.149 0.321
Greg 0.471 0.307 0.522 0.433
Arun 0.493 0.563 0.164 0.407
JimD 0.493 0.505 0.329 0.442
Mike 0.661 -0.058 0.009 0.204
Remember that a number closer to 1.000 means that the judge agreed with the collective wisdom of the other judges, a number closer to -1.000 means that the judge disagreed with the collective wisdom of the other judges, and a number close to 0.000 means that the judge neither agreed nor disagreed with the collective wisdom of the other judges.

This past week, the results were pretty varied. Greg agreed most with the other judges, while Jeff somewhat disagreed with them. Michael, once again, was independent of them.

For all three weeks, however, each judge still has a positive number, indicating that (overall) each judge agrees with the other judges to some degree.

Thus, so far, no whacko.

Thursday, September 17, 2009

USCL Ratings and Power Rankings: Week 3 (2009)

UPDATE. I forgot to increase the current week counter on the spreadsheet, and thus what I posted was wrong. Mea culpa. Thank you for HA81 for the constructive criticism which made me go back to look at this.

Here is the correct list.

USCL 2009 Week 3 Power Rankings (CORRECTED)

1 - San Francisco 1.000
2 - Seattle 0.931
3 - New Jersey 0.922
4 - Miami 0.883
5 - New York 0.843
6 - Dallas 0.813
7 - Arizona 0.801
8 - Tennessee 0.737
9 - Philadelphia 0.718
10 - Chicago 0.699
11 - Boston 0.691
12 - Queens 0.562
13 - Carolina 0.543
14 - Baltimore 0.527


I had been asked by United States Chess League Commissioner, Greg Shahade, to also publish power rankings for the 14 USCL teams. Note that these Power Rankings are entirely based on mathematical formulas; there is no subjectivity at all. They are calculated using the same formula that used previously on this website. Top team always get a 1.000, the remainder of the teams get numbers less than that.



Below are the unofficial player ratings for the United States Chess League for 2009, post-week 3.

USCL 2009 Week 3 Player Ratings

Rank Rating Name
1 2641 Jaan Ehlvest
2 2632 Hikaru Nakamura
3 2631 Julio Becerra
4 2604 Sergey Erenburg
5 2588 Larry Christiansen
6 2584 Joel Benjamin
7 2576 Dean Ippolito
8 2574 Marko Zivanic
9 2571 Jacek Stopa
10 2566 Pascal Charbonneau
11 2554 Gregory Serper
12 2546 Alex Shabalov
13 2542 Josh Friedel
14 2539 Eugene Perelshteyn
15 2533 Alejandro Ramirez
16 2529 Alex Lenderman
17 2529 Giorgi Kacheishvili
18 2525 Boris Gulko
19 2515 Dmitry Schneider
20 2506 Sam Shankland
21 2505 Bruci Lopez
22 2505 Thomas Bartell
23 2501 Levon Altounian
24 2500 Marc Esserman
25 2500 Eli Vovsha
26 2492 Rogelio Barcenilla
27 2486 Daniel Ludwig
28 2478 Alex Stripunsky
29 2464 Sergey Kudrin
30 2448 Mehmed Pasalic
31 2444 Jan van de Mortel
32 2442 Mark Ginsburg
33 2441 Denis Shmelov
34 2439 Jorge Sammour-Hasbun
35 2434 David Pruess
36 2430 Florin Felecan
37 2424 Robby Adamson
38 2424 John Bartholomew
39 2423 Tegshsuren Enkhbat
40 2415 Slava Mikhailuk
41 2414 Ron Burnett
42 2411 Jonathan Schroer
43 2407 Nikola Mitkov
44 2405 John Donaldson
45 2401 Angelo Young
46 2398 Larry Kaufman
47 2391 Jay Bonin
48 2388 Oleg Zaikov
49 2388 Keaton Kiewra
50 2380 Ron Simpson
51 2377 Eric Rodriguez
52 2375 Andrei Zaremba
53 2360 Bryan Smith
54 2355 Blas Lugo
55 2354 Victor Shen
56 2352 Michael Lee
57 2351 Peter Bereolos
58 2349 Aviv Friedman
59 2347 Matthew Herman
60 2332 Vadim Martirosov
61 2323 Yian Liou
62 2315 Eric Tangborn
63 2311 Ilya Krasik
64 2306 Todd Andrews
65 2296 Jon Burgess
66 2295 John Bick
67 2281 Craig Jones
68 2261 Nelson Lopez
69 2256 Tsagaan Battsetseg
70 2250 Danny Rensch
71 2241 Leo Martinez
72 2216 Josh Sinanan
73 2207 Elizabeth Vicary

Sunday, September 13, 2009

GOTW Whackometer, Part 2

Well, I woke up yesterday morning, with a sick feeling that the co-linearity problem was actually a real problem, given the population size. So, I redid the correlations, but this time correlated a judges score with the sum of the points of the other judges. In this way, a judge's own points are not considered in the correlation.

The results are rather interesting.

Name Week-1 Week-2 Average
Jeff 0.792 0.319 0.556
Greg 0.471 0.307 0.389
Arun 0.493 0.563 0.528
JimD 0.493 0.505 0.499
Mike 0.661 -0.058 0.302

Mike's correlation in week 2 is pretty close to 0, which suggests that he had almost no correlation with the other judges. However, in both weeks, all the judges are reasonably similar in their correlations. Two weeks is not a very good sample.

HOWEVER, note that this is for the games that were selected. Remember that there are about 20 other games, that ALL judges marked as 0. Including these games would lead to a much higher correlation for everyone.

Friday, September 11, 2009

Game of the Week Controversy

Only week 2 in the United States Chess League, and there's already controversy about the "Game-of-the-Week" judging. It appears that several members of the Boston Blitz are upset that their team member, Marc Esserman, only came in second place in the voting.

There have been claims of a "whacko judge". But how can we define "whacko"? Quantitatively, I mean...

One method is by performing standard Pearson correlations on an individual judge's scores with the total scores of all the judges. Yes, there's some co-linearity issues, but I'm not really going to worry about that.

Correlations produce a score called "r", which ranges from -1 to 1. When r=1, it means that one set of data lines up exactly with another set of data. For example, the ordered data set {1, 2, 3} and {2, 4, 6} have an r=1. On the flip side, an r =-1 means that one data set is exactly opposite (in terms of lining up) with another. Take the data set {1, 2, 3} again; the data set {6, 4, 2} has an r=-1, because 1 and 6, 2 and 4, 3 and 2, line up in the opposite order.

So what does an r=0 mean? It means that the two data sets have absolutely no statistical correlation at all, either negative or positive; it is like there are two random data sets.

In terms of USCL GOTW judging, what we would like to see is some sort of positive correlation between each judge's scores and the scores given to the games as a whole. If there is some sort of "whacko" judge, then they should have a near zero or a negative correlation. Indeed, even if one judge has a noticeably lower positive correlation than the other judges, it something of which to keep track.

For Week 2, GOTW judging, I calculated the correlations between a judge's score and the tally of all the judges. Before I go into the numbers, let me first give an example.

Winner / Total Points / Score of "Greg"
Friedel / 18 / 2
Esserman / 14 / 5
Ehlvest / 9 / 0
Perelshteyn / 9 / 3
Charbonneau / 6 / 0
Krasik / 6 / 4
Lopez / 4 / 0
Zaremba / 3 / 0
Becerra / 3 / 0
Matlin / 1 / 0
Recio / 1 / 1
Altounian-Burnett / 1 / 0

The Pearson correlation ("r") between Greg's scores and the scores of all the judges was 0.594, which is a high positive number, which indicates that his choices matched reasonably well with the choices of the group as a whole.

Let's look at the correlations of all the judges of week 2, from highest (i.e., agreed closest with the group) to lowest.

Week 2 Correlations
Arun: 0.770
Jim: 0.733
Jeff: 0.604
Greg: 0.594
Michael: 0.280

Michael's score is well-below the other judges, but still positive, which means that he did agree somewhat with all the other judges.

So, let's go back and see what happened for the Week 1 voting, where the results were apparently less controversial.

Week 1 Correlations
Jeff: 0.882
Michael: 0.799
Arun: 0.683
Jim: 0.683
Greg: 0.667

These are all reasonably close, when Jeff and Michael having a better correlation.

So to summarize, we can determine the average correlation for both weeks.

Average Correlation for both weeks
Jeff: 0.743
Arun: 0.727
Jim: 0.708
Greg: 0.631
Michael: 0.540

All are reasonably high, so I think it may be too early to refer to a "whacko" judge.

Thursday, September 10, 2009

2009 USCL Ratings Week 2

The ratings have been posted for week 2 of the USCL in 2009. Again, only players who have played in 2009 are listed, and to be listed, players have to have played a minimum of three lifetime USCL games.

Rank Rating Name
1 2640 Jaan Ehlvest
2 2626 Julio Becerra
3 2614 Larry Christiansen
4 2604 Hikaru Nakamura
5 2604 Sergey Erenburg
6 2574 Marko Zivanic
7 2572 Joel Benjamin
8 2568 Dean Ippolito
9 2558 Pascal Charbonneau
10 2554 Gregory Serper
11 2547 Josh Friedel
12 2546 Alex Shabalov
13 2540 Bruci Lopez
14 2539 Eugene Perelshteyn
15 2533 Alejandro Ramirez
16 2527 Alex Lenderman
17 2525 Dmitry Schneider
18 2513 Thomas Bartell
19 2501 Levon Altounian
20 2500 Marc Esserman
21 2500 Eli Vovsha
22 2500 Alex Stripunsky
23 2482 Sam Shankland
24 2468 Sergey Kudrin
25 2453 Andrei Zaremba
26 2444 Jan van de Mortel
27 2441 Denis Shmelov
28 2434 David Pruess
29 2431 Florin Felecan
30 2426 Jonathan Schroer
31 2424 Robby Adamson
32 2424 John Bartholomew
33 2423 Tegshsuren Enkhbat
34 2414 Ron Burnett
35 2407 Mehmed Pasalic
36 2407 Nikola Mitkov
37 2405 John Donaldson
38 2403 Larry Kaufman
39 2396 Peter Bereolos
40 2395 Angelo Young
41 2392 Ron Simpson
42 2391 Jay Bonin
43 2390 Oleg Zaikov
44 2377 Eric Rodriguez
45 2373 Keaton Kiewra
46 2360 Bryan Smith
47 2354 Victor Shen
48 2354 Ilya Krasik
49 2349 Aviv Friedman
50 2315 Eric Tangborn
51 2314 Michael Lee
52 2311 Todd Andrews
53 2288 John Bick
54 2280 Matthew Herman
55 2278 Tsagaan Battsetseg
56 2262 Craig Jones
57 2261 Nelson Lopez
58 2215 Josh Sinanan

Friday, September 04, 2009

2009 USCL Inaugural Raring List

Welcome to the (unofficial) USCL Ratings List of 2009! I am pleased to provide this service again to all the USCL fans out there.

The details of the calculations of this rating system have been described elsewhere, but to summarize... (a) it uses a modified Glicko system; (b) there are different scores for a draw if your team won or lost; (c) it accounts for color in your score expectancy; (d) uncertainty scores are automatically raised at the beginning of a season; (e) initial rating assignment depends on the board that you played; (f) all games are given equal weight (regular season, playoffs, tiebreakers).

That being said, the same conditions apply to each published list. First, the player must have played at least three lifetime games in the USCL, and second, the player must be active in the year the list is published.

Here is the initial list for 2009 - there are so few names on the list, because there are fewer players active in 2009 (so far).

Rank Rating Name
1 2618 Julio Becerra
2 2614 Larry Christiansen
3 2576 Gregory Serper
4 2568 Joel Benjamin
5 2552 Alex Lenderman
6 2546 Alex Shabalov
7 2544 Dean Ippolito
8 2544 Alejandro Ramirez
9 2539 Pascal Charbonneau
10 2535 Levon Altounian
11 2535 Bruci Lopez
12 2534 Eugene Perelshteyn
13 2512 Josh Friedel
14 2500 Eli Vovsha
15 2500 Alex Stripunsky
16 2495 Marc Esserman
17 2473 Sergey Kudrin
18 2447 Andrei Zaremba
19 2434 David Pruess
20 2432 Jonathan Schroer
21 2431 Florin Felecan
22 2424 John Bartholomew
23 2417 Nikola Mitkov
24 2416 Tegshsuren Enkhbat
25 2409 Oleg Zaikov
26 2407 Mehmed Pasalic
27 2405 John Donaldson
28 2403 Larry Kaufman
29 2398 Ron Simpson
30 2396 Peter Bereolos
31 2394 Eric Rodriguez
32 2392 Robby Adamson
33 2385 Ron Burnett
34 2373 Keaton Kiewra
35 2360 Bryan Smith
36 2354 Victor Shen
37 2315 Eric Tangborn
38 2311 Michael Lee
39 2298 Craig Jones
40 2288 John Bick
41 2280 Matthew Herman
42 2261 Nelson Lopez
43 2215 Josh Sinanan