2014 US Chess Championship Round 11

Kamsky, Lenderman, and Akobian tied for 1st place. In the playoff, Lenderman and Akobian will play an Armageddon game with the winner facing Kamsky in a 2 game rapid match. Clearly Kamsky has the best chances, since he only has to win 1 match to win the championship, while the Lenderman and Akobian have to win both matches. But beyond that I can't make a forecast, since there is so little data on Armageddon matches. Predicting the rapid match with Kamsky is also difficult since Akobian has no rapid FIDE rating. It would be possible to plug in his USCF quick rating (2612). However, I don't have a database of rapid games, so what draw percentage should the model assume?

The only forecast I can make is that Kamsky's chances are at least 50%. May the best player win.

2014 US Chess Championship Round 10

Lenderman and Molner drew, and Shankland reduced Friedel's chances to 0.

The last round of a tournament is often difficult to predict since the players frequently have a strong incentive to play for a particular result. For instance, one player may have must-win game while another player might just want a quick draw. This can distort the draw percentages, which are important for the forecast. In most tournaments, two leaders playing against each other in the final round is a recipe for a short draw, but here Akobian and Lenderman might not want to risk entering a playoff with Kamsky. In their post-game interview, I didn't detect any signs that they were planning on a peaceful last game. Thus, the probability of a draw in Akobian - Lenderman may be less that 0.55, which the model assumes based on grandmaster games in my database. A lower draw rate will diminish Akobian's and Kamsky's chances while increasing Lenderman's. On the other hand, Kamsky may be inclined to take more risks in his last round game with Friedel, since Kamsky has to win in order to have a shot at 1st place. Therefore the draw percentage in Kamsky - Friedel may also be lower than the model assumes, and Kamsky's chances may need to be revised upwards.

The standings are incomplete because some games are still being played this round, but none of the other players can win the tournament.

2014 US Chess Championship update

Kamsky and Akobian just drew, so Kamsky's chances of winning the tournament plummeted. But they may rebound somewhat if Lenderman draws Molner, which seems likely.

2014 US Championship Round 9

Akobian's chances fell greatly today after he lost to Shankland. Kamsky was unable to convert against Molner, while Lenderman won and is now tied for 1st with Akobian. Lenderman now has excellent chances since he faces an easier schedule than Akobian; the next round features Akobian - Kamsky and Molner - Lenderman. The most important games in the last round are Kamsky - Friedel and Lenderman - Akobian.

Friedel is in contention for 1st, but the model is very pessimistic that he will win; his rivals are much higher rated.

***

A chess blog is rather incomplete without any chess games, so here is one of my losses to Akobian in a blitz tournament. I was able to reconstruct most of this game afterwords but not the other game I lost to him. He makes it look easy...

Wilson, Matthew (1944) - Akobian, Varuzhan (2683) [B07]
Washington Speed Chess Championship Seattle (2), 31.07.2010

1.e4 d6 2.d4 Nf6 3.Nc3 c6 4.g3 Qc7 5.Bg2 e5 6.Nge2 Nbd7 7.0–0 Be7 8.h3 0–0 9.Be3 b5 10.a3 a5 11.f4 Bb7 12.dxe5 dxe5 13.Qe1 b4 14.axb4 axb4 15.Rxa8 Rxa8 16.Nd1 Ra1 17.c3 c5 18.cxb4 cxb4 19.fxe5 Nxe5 20.Bf4 Nxe4 21.Bxe4 Bxe4 22.Nd4 f5 23.Nxf5 Qc5+? [23...Bxf5! 24.Bxe5 (24.Qxe5 Qxe5 25.Bxe5 Bxh3 26.Re1 Bg4 wins) 24...Qc2! and the threat of 25...Bxh3 is decisive.] 24.Be3 Qc7 25.Nxe7+ Qxe7 After 26.Bd4, White, though much worse, is still alive. 26.Qf2?? Nf3+ 27.Kh1 and Black went on to win. 0–1

US Chess Championship Round 8

This was a very significant day for the forecasts. Akobian defeated Erenburg while Kamsky only managed a draw against Naroditsky, so now Akobian leads by a full point with 3 rounds to go. He is the clear favorite, but Kamsky and Lenderman still have chances. One factor working in Kamsky's favor is that he has an easier schedule; he plays bottom seeds Molner and Friedel in rounds 9 and 11. Akobian will face Shankland and Lenderman in those rounds. The crucial game will be the round 10 duel between Kamsky and Akobian.

2014 US Chess Championship Round 7

Perhaps the model was right to have so little confidence in Lenderman; today he lost to Kamsky and is now one point behind the leader. Akobian won a nice game against Robson and is currently in clear first place.

The model thinks that it is mostly a race between Akobian and Kamsky with some chances for Onischuk.

Unfortunately, tomorrow is a rest day for the players, so there will be no games to watch. Our productivity at work will soar...

2014 US Chess Championship Round 6

Lenderman and Gareev lost today, allowing Akobian to tie for 1st. Perhaps the forecast was right to discount Lenderman's chances? Onischuk and Kamsky drew their game and are now just half a point behind the leaders.

The latest forecast:

2014 US Chess Championship Round 5

Gareev stumbled against Akobian, so now Lenderman leads by a full point. The forecast is slowly starting to increase its confidence in Lenderman's chances.

2014 US Chess Championship Round 4

Lenderman continues to impress. After defeating Robson, he leads with 3.5/4. Kamsky scored his first win, but his rivals Onischuk and Gareev also prevailed today.

The forecast is surprisingly pessimistic about Lenderman's chances. Last week I started a more thorough investigation of the draw percentage assumptions, and I suspect that this will influence the results. The first step was to build a database that purged all the rapid, blitz, exhibition, and computer chess games. Next I'll construct a statistical model that will estimate the draw percentage given the players' ratings. This could take a while; a similar project studying whether 1.e4 was better than 1.d4 took several weeks. Along the way we will also find out if 1.d4 really is safer than 1.e4 - all else equal, is there a higher draw rate for games beginning with 1.d4? Much work remains to be done, but there are a few preliminary results. The new draw rate assumption will probably cause the forecast to give Lenderman higher chances.

2014 US Championship Round 3

Lenderman and Gareev scored today, placing Lenderman in clear first place with 2.5/3. But the forecast is not too optimistic about his chances due to his relatively low rating. Kamsky remains the favorite, but his chances are diminishing as he has yet to win a game.

Standings:

2014 US Championship Round 2

All 6 games were drawn today, so there is not much change in the forecasts.

Lenderman and Robson remain in the lead

2014 US Championship

Round 1 of the US Chess Championship began today. Kamsky is the clear favorite due to his 2713 Elo rating, nearly 50 points above his closest rival (Nakamura declined to participate). The format is a 12-player round robin featuring America's top chess talent.

Your blogger once had the honor of playing a two game match against Akobian in a blitz tournament. The result was predictable.

After 40,000 simulations, we have our first forecast for the tournament. Kamsky has a better chance than anyone else, but don't be shocked if he doesn't win. As usual, the first column sums to more than 100% because multiple players can tie for first.

What would happen if the tournament was a double-round robin instead? Kamsky's chances of at least tying for 1st would rise to around 56%; he would win the tournament outright 45% of the time. A big problem with many tournaments and the World Championship is that they are far too short. Even if one player is definitely better than the rest, there are so few games that an upset could easily occur and a weaker player is crowned victor. As more games are played, there is less scope for luck and the stronger player will more likely prevail.

Round 1 update: Robson and Lenderman emerge as the early leaders.

In the new forecast, Robson's chances climbed the most.