For improvers4 min read
Theory is a safety net
In theory, a move at 1400–1599 Elo is a mistake only 4% of the time. The move that leaves it is, 15%. The jump is there at every level, from 3.6 to 7.7 times.
- in theory
- leaves theory
- later in the game
What the numbers show
- In theory, moves are rarely mistakes, 4.0% at 1400–1599 Elo. On the move that leaves it, it is 15.2%, 3.8 times as much.
- The jump is largest for the strongest players: from 0.7% to 5.4%, 7.7 times.
- Against later moves of the same stretch of the game, the exit is 1.36 times as risky up to 999 Elo, 1.16 at 1400–1599 and 0.89 from 2200. So for strong players the exit is no riskier than other moves outside theory: they lose the net, not the single move.
- The exception is a very early exit by strong players: from 2200 Elo, an exit at half-move 6 is a mistake in 8.9% of cases, at half-move 14 in 5.1%.
- What many players choose is usually tested. The numbers do not say that leaving theory causes the mistake, only that this is where the risk starts.
3.6×
3.6×
3.7×
3.8×
4.0×
4.7×
5.8×
7.7×
Mistake rate of the move that leaves theory divided by the mistake rate of a move in theory, in the same stretch of the game.
| Rating | in theory | leaves it | first move after | later |
|---|---|---|---|---|
| up to 999 | 7.9% | 28.6% | 19.3% | 21.0% |
| 1000–1199 | 6.2% | 22.2% | 14.8% | 17.3% |
| 1200–1399 | 5.0% | 18.3% | 12.2% | 15.0% |
| 1400–1599 | 4.0% | 15.2% | 10.2% | 13.1% |
| 1600–1799 | 3.1% | 12.5% | 8.5% | 11.3% |
| 1800–1999 | 2.2% | 10.3% | 7.0% | 9.6% |
| 2000–2199 | 1.4% | 8.1% | 5.6% | 8.1% |
| 2200+ | 0.7% | 5.4% | 3.9% | 6.1% |
Does it always hold?
- up to 999
- 1400–1599
- 2200+
| Rating | Loss from | in theory | leaves it | later |
|---|---|---|---|---|
| up to 999 | 6 | 15.4% | 43.4% | 31.7% |
| up to 999 | 10 | 7.9% | 28.6% | 21.0% |
| up to 999 | 20 | 2.5% | 13.1% | 9.6% |
| up to 999 | 30 | 0.8% | 6.4% | 5.0% |
| 1400–1599 | 6 | 9.1% | 28.2% | 23.1% |
| 1400–1599 | 10 | 4.0% | 15.2% | 13.1% |
| 1400–1599 | 20 | 0.9% | 5.2% | 4.5% |
| 1400–1599 | 30 | 0.3% | 2.1% | 1.9% |
| 2200+ | 6 | 2.2% | 13.5% | 13.3% |
| 2200+ | 10 | 0.7% | 5.4% | 6.1% |
| 2200+ | 20 | 0.1% | 1.4% | 1.4% |
| 2200+ | 30 | 0.0% | 0.5% | 0.5% |
| Rating | Time control | in theory | leaves it | Factor |
|---|---|---|---|---|
| up to 999 | Bullet | 6.8% | 26.9% | 3.9× |
| up to 999 | Blitz | 7.4% | 26.5% | 3.6× |
| up to 999 | Rapid | 8.9% | 31.6% | 3.5× |
| up to 999 | Classical | 9.1% | 34.4% | 3.8× |
| 1400–1599 | Bullet | 4.0% | 16.8% | 4.2× |
| 1400–1599 | Blitz | 3.7% | 13.9% | 3.8× |
| 1400–1599 | Rapid | 4.4% | 15.9% | 3.6× |
| 1400–1599 | Classical | 4.8% | 17.6% | 3.7× |
| 2200+ | Bullet | 1.4% | 8.7% | 6.3× |
| 2200+ | Blitz | 0.5% | 4.0% | 8.3× |
| 2200+ | Rapid | 0.6% | 4.5% | 7.2× |
| 2200+ | Classical | 0.5% | 3.1% | 6.6× |
Half-moves 8 to 24, loss of 10 points or more.
Questions
- What is the end of theory?
- The first position of a game that appeared in fewer than 100 games of the study at that half-move. Before it, the crowd plays known moves.
- Why the same stretch of the game?
- Mistakes get more common as a game goes on. Comparing moves from different stretches mostly measures the phase of the game. All numbers here are for half-moves 8 to 24.
- What to do with it?
- If you know where your theory ends, you know where the risk starts. That is where practice pays, especially for moves that keep coming back.