The Rating Game: Elo’s chess rankings prove their worth

Arpad Elo knew his stuff.

The Hungarian American physicist in the 1950s and 1960s adapted and perfected a system to rate the strength of chessplayers, first for the U.S. and then for the world. The Elo rankings proved both revelatory and useful, allowing for better pairings at tournaments and much fairer class competitions.

And to this day, the system is a remarkable predictor of results. A Class B player tends to beat a Class D player, a master tends to beat an expert, even a 2725-rated grandmaster will have a clear edge a 2695-rated grandmaster — not every time, of course, but you could make a lot of money betting just on Elo’s rankings.

One little hitch in the system, however, is calculating the strength of younger players surging up the wall chart. That could be how 15-year-old NM Arthur Guo (current rating: 2359) found himself the winner on tiebreaks of the recent National Open, the big Swiss blowout back at its traditional Las Vegas home after a one-year COVID-19 hiatus. Guo’s victory included a string of Elo-esque “upsets” of grandmasters in the middle rounds, including a nice Round 7 win over GM Hans Niemann that vaulted him into a tie for the lead.

A Closed Catalan develops into a wide-open middle game (one very interesting road not taken would have been 19… Bxe2!? 20. exf6 Bxd1 21. fxg7 Nxg7 22. Nh6+ Kh8 23. Rxd1, and White has good but nebulous compensation for the lost exchange), and it is the grandmaster who blinks first in the critical phase: 23. Bd5 (see diagram; the four bishops and the multiple open files give both players a bewildering range of options, but Guo manages to land the crucial first punch) Bxf2+!? 24. Kh1? (more forthright is accepting with 24. Kxh2 Rc2+ 25. Kg1 Rxb2, as White appears to keep things equal in lines like 26. Ne7+ Kh8 27. Rdb1 Rd2 28. Rd1 Rd3 29. Rxd3 Bxd3 30. Nc6) Rc2 25. Be5 Bc5 26. Rxc1 Rxc1 27. Rxc1 Rd8 28. Bg2 (Bxe6?? Bb7+ and mate next) f6!, and despite the wide-open board, White’s pieces find themselves short of safe squares.

It’s over surprisingly quickly on 29. Bf4 (Bd6?! [Ba1 Rd2 30. Bc6 Kf8 31. g4 Rxa2 is bleak for White] Bxd6 30. Bd5 Be5 [Kf7 31. Rc6] 31. Bxe6+ Kf8 32. Kg1 Rd2 33. Rc6 Bb7 34. Rc4 Rg2+ 35. Kf1 Rxh2 and wins) Bd3! 30. Ne3 (Nh4 g5; 30. Nd6 Bxd6 31. Rd1 Nxf4) g5, and every flight square for the attacked bishop is covered; Niemann resigned.


Perhaps the most improbable upset this columnist ever covered came 20 years ago this year, in the late, unlamented FIDE world championship knockout tournaments of that era. Someone thought it a good idea to have the world title decided by winnowing down a huge field a la Wimbledon through a series of capricious, anything-can-happen two-game mini-matches.

And that’s how little-known French IM Olivier Touzane, who qualified for the tournament by winning an internet event, nearly eliminated top-seeded Indian GM (and future world champ) Viswanathan Anand in the very first round with a sensational win with Black. Anand, with a nearly 400-point ratings advantage, builds up a nice positional edge in this Petroff Defense, but is clearly flummoxed when his opponent refuses to lie down and concede.

With 18. c4! Bxd2 (Be6 19. d5) 19. cxd5! Bxe1 20. dxc6 Ba5 21. Qd3 g6 22. cxb7, White sacrifices the exchange to obtain a strong bind, with the advanced b-pawn tying down Touzane’s forces. But White makes the mistake of waiting for a Black blunder just at the stage when Black decides to dig in and up his game.

Thus: 23. Bxh6!? (more in the spirit of the position was 23. Bf4 Bb6 24. Qe4, retaining control of the position) Re8 24. Qf3 Re6! (Qxd4?! 25. Bb3 Re7 26. Bg5 Rd7 27. h4 is still fine for White; Black’s rook will play an excellent offensive and defensive role on the sixth rank) 25. Bb3?! (h3 was a better option; White is starting to drift) Rf6 26. Qg4 Rb6!, and suddenly the prized White passed pawn is toast and Black is generating some threats of his own.

It’s all downhill for White after that, as Touzane sacs an exchange of his own to simplify to a very favorable pawn-up ending: 30. g4 Rxb3! 31. Rxb3 Rxb3 32. Qxb3 Qxg4+ 33. Kf1 Qxd4. Queen-and-pawn endings can be notoriously tricky, but unfortunately for Anand, Black’s king is perfectly safe, his passed pawn can advance without hindrance, and White’s own king is on shaky ground.

All of that produces one of the shocking upsets of the era: After 36. Bxb6 axb6 37. Qg3 Qc1, White has no good checks and no answer for the push of the c-pawn. It’s over on 38. h5 (only making things worse) Qc6+ 39. Kf1 Qh1+, and Anand resigned facing 40. Ke2 (Qg1 Qxg1+ 41. Kxg1 gxh5 and the pawn ending is an easy win) Qxh5+ 41. Kd2 Qd5+ 42. Ke1 Qe6+ 43. Kf1 (Kd2 Qd6+ and the queens come off) c5, with a massive edge.

Fortunately for Mr. Elo, Anand would recover to win the second game of the mini-match and defeat the Frenchman in the rapid playoff.

Niemann-Guo, National Open, Las Vegas, June 2021

1. d4 Nf6 2. c4 e6 3. Nf3 d5 4. g3 Be7 5. Bg2 O-O 6. O-O c6 7. Nc3 Nbd7 8. Nd2 b6 9. e4 dxc4 10. Nxc4 Ba6 11. b3 b5 12. Ne3 b4 13. Ne2 e5 14. Bb2 exd4 15. Bxd4 Nc5 16. Nf5 Ne6 17. Bb2 Bc5 18. e5 Qxd1 19. Rfxd1 Ng4 20. Ned4 Nxe5 21. Nxc6 Nxc6 22. Bxc6 Rac8 23. Bd5 Bxf2+ 24. Kh1 Rc2 25. Be5 Bc5 26. Rac1 Rxc1 27. Rxc1 Rd8 28. Bg2 f6 29. Bf4 Bd3 30. Ne3 g5 White resigns.

Anand-Touzane, FIDE World Championship Knockout Tournament, Moscow, November 2001

1. e4 e5 2. Nf3 Nf6 3. Nxe5 d6 4. Nf3 Nxe4 5. d4 d5 6. Bd3 Nc6 7. O-O Be7 8. c4 Nf6 9. Nc3 O-O 10. Re1 Be6 11. cxd5 Nxd5 12. a3 Bf6 13. Be4 h6 14. Bc2 Nxc3 15. bxc3 Bc4 16. Nd2 Bd5 17. Rb1 Bg5 18. c4 Bxd2 19. cxd5 Bxe1 20. dxc6 Ba5 21. Qd3 g6 22. cxb7 Rb8 23. Bxh6 Re8 24. Qf3 Re6 25. Bb3 Rf6 26. Qg4 Rb6 27. Bg5 Qe8 28. Qf3 R8xb7 29. h4 Qd7 30. g4 Rxb3 31. Rxb3 Rxb3 32. Qxb3 Qxg4+ 33. Kf1 Qxd4 34. Be3 Qa1+ 35. Kg2 Bb6 36. Bxb6 axb6 37. Qg3 Qc1 38. h5 Qc6+ 39. Kf1 Qh1+ White resigns.

• David R. Sands can be reached at 202/636-3178 or by email at

Sign up for Daily Newsletters

You Might Also Like