Do chess opening traps work? 26 traps in Lichess data

Opening traps are easy to collect and hard to judge. A trap video shows the game where everything works. It leaves out how often the opponent steps around the bait, and how often the trapper misses the punishment.

So we counted. For 26 well-known opening traps, we used the Lichess opening explorer to find how many games reached the trap position, how often the next move fell for the trap, and how often the trapper then played a punishing reply. The games are blitz, rapid and classical games from the Lichess database, retrieved on October 16, 2026.

The main findings:

• The median trap caught 24% of the players who reached it. The range ran from 1.2% for the Two Knights Defense ...h6 trap to 92% for the French Defense Classical trap.
• After a player fell for a trap, the median punish rate was 94%. Five traps were punished in less than 60% of games, and the French trap in only 4.4%.
• The traps that caught the most players began with an ordinary-looking move such as a recapture. Traps whose threat is already visible caught few players.
• Rating changes some traps far more than others. The Fried Liver Attack caught 72% of players under 1200 and 7.1% at 2000 and above. The Halosar Trap caught 46% to 48% in every band.

You can download the full dataset as a CSV, with every trap split by rating band.

How to read the numbers

Each trap has a trap position, where one player chooses a move. The mistake is the move that falls for the trap, and the punishment is the other player's reply that takes advantage of it. We call the player who chooses the mistake the victim and the player who can punish it the trapper.

• Games at the trap: games that reached the trap position, by any move order.
• Fell for it: the share of those games where the next move was the mistake.
• Punished: the share of games after the mistake where the trapper's next move was a punishing reply. Only that first move counts, so a player who found the first move and went wrong later still counts as punished.
• Edge: what the best punishment leaves the trapper according to Stockfish 19. Mate means a forced mate, Winning means an edge of three pawns or more, Clear means one to three pawns, and Small means less than one pawn.

Lichess groups games by position, so one game can pass through several trap positions. Every Fried Liver game also reaches the Two Knights position, for example. Read each row on its own and do not add rows together.

All 26 traps ranked

The table is sorted by how often players fell for each trap. Move numbers follow the usual move order of each trap. Trap names link to the matching line in our trap trainer where one exists.

Two columns tell different stories. Punish rates sit high for most traps: 15 of the 26 were punished in 90% of games or more. Fall rates spread across the whole range, from 1.2% to 92%.

Traps that start with a natural move catch the most players

The traps at the top of the table share a pattern. The mistake looks like normal chess, such as recapturing material or holding on to an extra pawn, and the problem only shows after the move.

The French Defense Classical trap is the clearest case. After 7...cxd4, 92% of White players recaptured with 8.Nxd4. It looks like the most natural move on the board, and it allows Black to take on e5.

The same pattern runs through the next rows. In the Damiano Defense, 78% of Black players took the knight with 3...fxe5 and walked into 4.Qh5+. In Noah's Ark Trap, 67% of White players recaptured with 8.Qxd4. In the Queen's Gambit Accepted line, 50% of Black players defended the extra pawn with 4...c6 and lost it to 5.axb5. In the Fried Liver Attack, 46% took back on d5 with the knight.

Traps whose threat is already visible caught far fewer players. Only 1.2% of Black players answered 4.Ng5 with 4...h6, which allows 5.Nxf7 to fork the queen and rook. In the Hungarian Defense line, 7...gxh6 is also a recapture, but the queen on d5 already aims at f7, and only 2.5% took back. Scholar's Mate caught 6.0% of players who faced 3.Qh5.

Low rates still add up. Scholar's Mate positions appeared in more than 17 million games, so 6.0% means more than a million games where Black played 3...Nf6 and allowed mate in one.

Many punishments go unplayed

Falling for a trap often costs less than the trap videos suggest, because the trapper does not always punish it. Five traps were punished in less than 60% of games.

The French trap is the extreme case. Only 4.4% of Black players answered 8.Nxd4 with 8...Ndxe5. The most common replies were 8...Bc5 at 35% and 8...Nxd4 at 32%.

Noah's Ark Trap was punished in half the games. After 8.Qxd4, the move 8...c5 starts a sequence that traps the bishop on b3. Compare it with 8...Nf6, the most common alternative.

Fool's Mate is the strangest row. After 1.f3 e5 2.g4, mate in one is on the board, and only 23% of Black players played 2...Qh4#. The most common move was 2...d5, at 31.5%. The explorer shows moves, not reasons. Premoves in fast games, joke games, and players who did not expect a blunder that early could all play a part.

Scholar's Mate was punished in 69% of games. A quarter of White players, 24.8%, answered 3...Nf6 with 4.Qf3, which keeps the queen and misses the mate.

The London System row needs a footnote. Only 42% of White players met 4...Qxb2 with 5.Nxd5 or 5.Bb5+. Another 49.8% played 5.Nb5, which keeps a clear edge by Stockfish's count but falls short of the three-pawn mark this study uses for a winning punishment. The Fried Liver Attack, at 54%, is a similar case: its punishment only gives a small edge, and 6.Qf3 at 21.8% and 6.d3 at 13.5% were common alternatives.

What the punishment is worth

To see whether finding the punishment mattered, we compared the trapper's score after the punishing reply with the score after every other reply. The score counts a win as one point and a draw as half a point. The table lists every trap with at least 4,000 games of other replies, sorted by the gap between the two scores.

These are results of whole games, so they also reflect every later move by both players. Read them as a record of how the games went.

Missing the punishment can hand the advantage back. White scored 46% after missing mate in one in Scholar's Mate, below an even score. In the Halosar and Englund traps, the trapper scored 36% and 37% after other replies.

Some positions stay good even when the punishment is missed. Black still scored 76% in Fool's Mate games without 2...Qh4#, probably because White's king stays exposed after 1.f3 and 2.g4.

A small edge can still pay off. The Fried Liver and Ruy Lopez Exchange punishments give only a small edge by Stockfish's count, and they still lift the trapper's score by 13 to 14 points. The Blackburne Shilling Gambit adds a second effect: 4...Qg5 is a small edge by itself, but it sets the next trap. 35% of White players then took on f7 with 5.Nxf7 and lost to 5...Qxg2, and Black scored 78% after 4...Qg5.

Rating changes some traps more than others

Trap advice often assumes one kind of opponent. The rating bands show that each trap has its own curve. The bands use the average rating of the two players.

Fell for it, by rating band:

The first four traps fade as rating rises. The Fried Liver Attack drops from 72% to 7.1%. The next ones hold up better. Noah's Ark Trap still caught 39% of players at 2000 and above, the Halosar Trap caught 46% to 48% in every band, and the French trap caught 88% at 2000 and above. The Queen's Gambit Accepted line moves the other way, from 37% under 1200 to 52-56% from 1600 up.

Punished, by rating band:

Punish rates mostly rise with rating. Under 1200, Black found 6...Nxd5 in only 31% of Elephant Trap games; at 2000 and above, in 93%. Noah's Ark Trap was punished in about a third of games below 1600.

The London row runs the other way, from 57% under 1200 to 40% at 2000 and above. 5.Nb5, the clear-edge alternative, may explain part of it, but this study did not split replies by band. The French trap stays near 5% at every level.

How to use this in your training

The numbers suggest one habit for each side of a trap.

When you can set the trap, learn the punishment as carefully as the bait. Most traps catch only a minority of players, and several common ones went unpunished in about half the games or more. Start with the replies players miss most often: 8...Ndxe5 in the French trap, 8...c5 in Noah's Ark Trap, 6.Nxf7 in the Fried Liver Attack and 4.Qxf7# in Scholar's Mate.

When you might be the victim, train the positions where the natural move is the mistake. The Halosar Trap, the Damiano Defense and the French trap caught players at every rating. Below 1600, the Fried Liver Attack, the Ruy Lopez Exchange Nxe5 line and the London System ...Qxb2 line also caught more than a third of players. Playing a trap from the trapper's side is a quick way to recognize it from the other side.

Our Trap Lines · White course and Trap Lines · Black course collect these lines for recall practice. 22 of the 26 traps in this study have a matching line there. The Caro-Kann smothered mate, the Hungarian ...gxh6 line, the Petroff copycat line and the Fishing Pole Trap do not.

Methodology

Source. All counts come from the Lichess opening explorer, which is built from the Lichess game database released under CC0. We retrieved them on October 16, 2026. The CSV has one row per trap and rating band plus an all-ratings row. You are welcome to reuse it, and a link back to this article helps others find the method.

Games. We used blitz, rapid and classical games. Bullet is left out because at that speed a trap measures the clock more than the opening. The four bands use the explorer's rating groups by average rating: under 1200, 1200-1599, 1600-1999, and 2000 and up. The all-ratings figures are the sums of the bands. Joke games and games where one side was not trying count like any other game.

Positions. The explorer groups games by position, so transpositions are included and the same game can count toward more than one trap. Counts should not be added across traps.

Rates. Fell for it is the share of games at the trap where the next move was the mistake. Punished is the share of games after the mistake where the next move was a punishing reply. Only the first move of the punishment counts. Score is the trapper's wins plus half the draws, divided by games. In this article we quote a rate for a band only when at least 500 games reached the trap and at least 100 games reached the position after the mistake, and the tables mark thinner bands as too few games. The CSV includes every band.

Which traps. We chose 30 well-known traps from common trap collections and our own trap courses, then checked each with Stockfish 19 at depth 20. A trap stayed in the study only when the mistake scored at least half a pawn worse than the victim's best move in the trap position. Four candidates failed that check and were dropped. A deeper check at depth 26 put two of the remaining lines at or just under that mark: the French Classical trap at about 0.3-0.5 pawns and the Ruy Lopez Exchange Nxe5 line at about 0.4. We kept both because their punishments still score well in practice, and the CSV lets you drop them for a stricter cut.

Which punishments. In a mating trap, only the mate counts. When the best reply wins by three pawns or more, every reply worth three pawns or more counts. Otherwise, every reply within a quarter pawn of the best counts. The Edge column reports the best of those replies.

Sources

• Lichess opening explorer
• Lichess open database
• Chessmate trap trainer