Do opening mistakes disappear as you climb?
Six classic opening traps measured in every rating band of the Lichess opening book: most fade, one barely moves, one peaks in the middle of the ladder, and the punishment is not always found.
Everyone falls for the Scholar's mate once. But does a 2000 player still walk into the opening traps that catch beginners? We followed six classic traps through every rating band of the Lichess opening book (games 2013–2025), and measured two things in each band: how often the mistake is played when the position comes up, and how often the opponent then finds the punishment.
How to read the figures. A game's band is the average rating of its two players, and all time controls are mixed, from bullet to correspondence. "Mistake played" is the share of the games that reach the position in which the mistake is played; "punished" is the share of the games after the mistake in which the winning reply is played. Percentages are cut at one decimal, never rounded up.
The Scholar's mate: 3…Nf6??
| Band (average rating) | Mistake played | Punished (4.Qxf7#) |
|---|---|---|
| 1000–1199 | 8.4 % | 66.9 % |
| 1200–1399 | 7.2 % | 71.2 % |
| 1400–1599 | 6.1 % | 77.5 % |
| 1600–1799 | 5.0 % | 84.5 % |
| 1800–1999 | 4.3 % | 89.6 % |
| 2000–2199 | 3.9 % | 92.5 % |
| 2200–2499 | 3.6 % | 93.9 % |
| 2500+ | 3.0 % | 90.9 % |
The mistake fades slowly, from 8.4 % to 3.0 %. The striking column is the other one: in games averaging 1000–1199, a third of the players who get the chance do not play mate in one. 85,693 of them (27.6 %) retreat the queen with 4.Qf3 instead. Even at 2500+, 25 of 277 games miss it — most likely in bullet.
The Two Knights: 5…Nxd5?
| Band (average rating) | Mistake played | Punished (6.Nxf7 or 6.d4) |
|---|---|---|
| 1000–1199 | 67.7 % | 44.6 % |
| 1200–1399 | 59.1 % | 51.6 % |
| 1400–1599 | 46.8 % | 60.1 % |
| 1600–1799 | 31.6 % | 69.7 % |
| 1800–1999 | 17.5 % | 78.7 % |
| 2000–2199 | 8.2 % | 85.2 % |
| 2200–2499 | 4.1 % | 89.3 % |
| 2500+ | 2.7 % | 90.5 % |
The steepest slope of the six: two players out of three take back on d5 at the bottom, fewer than three in a hundred at the top. It is a mistake rather than a blunder — in our catalogue, Stockfish (depth 20) goes from 0.00 to +1.07 for White — and Black still scores between 35.0 % and 43.3 % after it in every band.
The Blackburne–Shilling: 4.Nxe5?
| Band (average rating) | Mistake played | Punished (4…Qg5) |
|---|---|---|
| 1000–1199 | 26.7 % | 86.9 % |
| 1200–1399 | 25.0 % | 92.6 % |
| 1400–1599 | 22.4 % | 95.8 % |
| 1600–1799 | 19.2 % | 97.7 % |
| 1800–1999 | 16.5 % | 98.5 % |
| 2000–2199 | 15.1 % | 98.0 % |
| 2200–2499 | 15.5 % | 95.9 % ⚠ |
| 2500+ | 17.7 % | 96.6 % ⚠ |
⚠ Smaller samples: 33,258 games reach the position at 2200–2499 and 2,177 at 2500+.
This one does not go away. Taking the e5 pawn stays above 15 % in every band, while the punishment 4…Qg5 — hitting the knight and g2 — is found almost every time. In our catalogue, Stockfish (depth 20) gives +0.96 before 4.Nxe5 and −0.73 after it.
The Englund Gambit: 6.Bc3??
| Band (average rating) | Mistake played | Punished (6…Bb4) |
|---|---|---|
| 1000–1199 | 22.1 % | 92.1 % |
| 1200–1399 | 23.0 % | 94.0 % |
| 1400–1599 | 24.3 % | 95.2 % |
| 1600–1799 | 24.5 % | 96.4 % |
| 1800–1999 | 21.3 % | 97.2 % |
| 2000–2199 | 14.8 % | 97.0 % |
| 2200–2499 | 8.9 % | 96.6 % |
| 2500+ | 6.4 % | 96.6 % |
The one trap that peaks in the middle: the mistake is most common in games averaging 1600–1799 (24.5 %), not at the bottom. One plausible reason — our reading, not something the data shows — is that the gambit itself is played by people who know this trap. Stockfish (catalogue, depth 20): +1.83 → −3.48.
The Stafford Gambit: 6.Bc4??
| Band (average rating) | Mistake played | Punished (6…Ng4) |
|---|---|---|
| 1000–1199 | 34.0 % | 51.1 % |
| 1200–1399 | 32.8 % | 52.1 % |
| 1400–1599 | 31.2 % | 53.6 % |
| 1600–1799 | 28.4 % | 54.9 % |
| 1800–1999 | 24.4 % | 56.1 % |
| 2000–2199 | 20.5 % | 57.8 % |
| 2200–2499 | 16.2 % | 61.3 % |
| 2500+ | 10.9 % | 63.0 % |
About one in nine games averaging 2500+ that reach this position still walk into it. And the punishment is far from automatic: 6…Ng4 is the only move that turns the game (Stockfish 19 at depth 22: −0.61 after Ng4, +0.45 after Bxf2+, +0.24 after castling), yet even at 2500+ it is found only 63.0 % of the time.
The Damiano Defence: 2…f6? and 3…fxe5??
After 1.e4 e5 2.Nf3, 2…f6? almost disappears with rating: 2.5 % of the games in the 1000–1199 band, 0.1 % at 2500+. If White then sacrifices with 3.Nxe5 and Black takes the knight (3…fxe5??), 4.Qh5+ follows in 98.3 % to 99.5 % of the games, in every band: the players who sacrifice on e5 know the follow-up.
And yet, at 2500+, Black scores 56.0 % in the 906 games where 3…fxe5 was played, although after 4.Qh5+ Stockfish gives White +4.90. In fast games, the extra knight often decides more than the engine's evaluation.
What it means for your games
- Some traps fade, others don't. The Two Knights capture falls from 67.7 % to 2.7 %; the Blackburne–Shilling bait is still taken in 17.7 % of the games at 2500+ that reach it.
- Knowing the punishment matters as much as avoiding the mistake. At 1000–1199, mate in one after 3…Nf6?? is missed a third of the time, and the Stafford refutation is found only one time in two.
- The bottom of the ladder is not always where mistakes are most common. The Englund trap is played most at 1600–1799.
To see which of these positions show up in your own games, analyze them in the app: Stockfish finds the mistakes and the coach explains each one.
Method
Source: ClickHouse table chess.opening_book (Lichess games 2013–2025), queried on 1 October 2026. For each trap and band: the games of every move from the position before the mistake, then from the position after it, transpositions included. "Punished" counts the reply Stockfish 19 (depth 22) ranks clearly best; for the Two Knights, 6.Nxf7 and 6.d4 are equally good and both count. Evaluations come from our mistakes catalogue (Stockfish, depth 20) when the mistake is in it, otherwise from Stockfish 19 at depth 22 on one thread. Bands where too few games reach the position are flagged (⚠). Two traps were left out because their trend is not reliable: the Tennison Gambit 3.Nd4?? (the free knight is taken less often as ratings rise, probably because of premoves in bullet) and the Blackburne–Shilling 5.Nxf7?? (too few games above 2200).
Check it on your own games
Import a game from Lichess or Chess.com: Stockfish finds the moments that mattered, the coach explains them move by move.