What chess players actually play at every rating

Key findings

• At 800–999, 63% of players answer 1.e4 with 1…e5; at 2200–2399 only 19% do.
• The Sicilian Defense (1…c5) overtakes 1…e5 as the most common reply to 1.e4 at 2000–2199, 29% to 25%.
• At 800–999, 44% of players accept the Queen's Gambit with 2…dxc4; at 2200–2399, 9.8% do.
• At 800–999, 74% of Black players answer 5.exd5 in the Two Knights Defense with 5…Nxd5, which allows the Fried Liver Attack; at 2200–2399, 4.3% do.
• After 6.Nxf7, the Fried Liver Attack scores 76% for White at 800–999, 25 percentage points above White's average in that band; from 2200–2399 up it scores below that average.
• The King's Gambit stops outscoring White's average at 1800–1999, the Danish Gambit at 2000–2199, and the Stafford Gambit (a Black gambit) at 2000–2199.

How players answer 1.e4

Below 1600, half or more of players answer 1.e4 with 1…e5. Its share falls in every band, and the Sicilian Defense (1…c5) takes over at 2000–2199. The French Defense (1…e6) and the Caro-Kann Defense (1…c6) more than double from the lowest bands to the highest. The Scandinavian Defense (1…d5) is the second most common reply below 1200 and fades above 2000.

Share of games in which Black played each reply after 1.e4, by average rating. The most common reply in each band is in bold.

Who accepts the Queen's Gambit

After 1.d4 d5 2.c4, players below 1600 take the pawn with 2…dxc4 more often than any other move. From 1600–1799 the Queen's Gambit Declined (2…e6) is more common, and from 1800–1999 the Slav (2…c6) is played more often than the capture too. If you play the Queen's Gambit as White, expect 2…dxc4 below 1600 and 2…e6 or 2…c6 above it.

Share of games in which Black played each move after 1.d4 d5 2.c4, by average rating. The most common move in each band is in bold.

Which gambits still work, and where they stop

For each gambit we take every game that reached the gambit position and score it for the side that gambited: a win is 1, a draw ½. We then subtract that colour's average score in all games of the same band. A positive number means the gambit did better than that colour usually does at that level; a negative one means it did worse.

Fried Liver Attack (White)

1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.exd5 Nxd5 6.Nxf7. Stops outscoring White's average at 2200–2399.

King's Gambit (White)

1.e4 e5 2.f4. Stops outscoring White's average at 1800–1999.

Danish Gambit (White)

1.e4 e5 2.d4 exd4 3.c3. Stops outscoring White's average at 2000–2199.

Stafford Gambit (Black)

1.e4 e5 2.Nf3 Nf6 3.Nxe5 Nc6. Stops outscoring Black's average at 2000–2199.

Gambiteer's score minus that colour's average score in the band, in percentage points. Below each: the gambiteer's score and the number of games. Cells with fewer than 1,000 games show a dash.

Who falls for the traps

For each trap we take every game that reached the position just before the mistake and count how often the player to move walked into it.

Fried Liver: 5…Nxd5

After 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.exd5, 5…Nxd5 allows 6.Nxf7, the Fried Liver Attack. Train the attack: Fried Liver Attack.

Blackburne Shilling: 5.Nxf7

After 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nd4 4.Nxe5 Qg5, 5.Nxf7 allows 5…Qxg2 6.Rf1 Qxe4+ 7.Be2 Nf3#. It comes up in the Italian Game.

Stafford Gambit: 6.Bg5

After 1.e4 e5 2.Nf3 Nf6 3.Nxe5 Nc6 4.Nxc6 dxc6 5.d3 Bc5, 6.Bg5 allows 6…Nxe4!, when 7.Bxd8 Bxf2+ 8.Ke2 Bg4# is mate. Train the trap: Stafford Gambit.

Scholar's Mate: 3…Nf6

After 1.e4 e5 2.Bc4 Nc6 3.Qh5, 3…Nf6 allows 4.Qxf7#.

Share of players who made the trap move, out of all games that reached the position. Below each: the number of games that reached it. Cells with fewer than 1,000 games show a dash.

Methodology

Dataset

The Lichess open database of standard rated games, released under CC0 (public domain). We read it through Chessreps' copy of the Lichess opening explorer, which holds every monthly dump from January 2022 to December 2025: 4.48 billion games at all time controls.

Time controls

Blitz, rapid and classical games only. Bullet, UltraBullet and correspondence games are left out.

Ratings

A game's rating is the average of both players' Lichess ratings at the time. Bands are 200 points wide, from 600–799 to 2400–2599; 2.80 billion games fall inside them.

Positions, not move orders

Each figure counts every game that reached a position, by any move order. A share is the fraction of those games in which the player to move chose a given move.

Sample sizes

Every table shows its game counts. We hide any figure based on fewer than 1,000 games. The JSON download has the exact counts behind every number.

Limits

Gambit scores describe the games where the gambit was played. People who choose a gambit may differ from people who don't, so a score is not proof of what the gambit does for you. Online play is faster and less careful than over-the-board chess.

Chess.com ratings

Lichess ratings run higher than Chess.com ratings. The table below adds a rough Chess.com equivalent from the conversion Chessreps uses elsewhere (Lichess ≈ 0.77 × Chess.com + 700). It is an approximation from published comparison tables, and we don't extend it below Lichess 1000.

Last updated

October 30, 2026.

Games in each rating band (blitz, rapid and classical).

Data and how to cite

Download the data (JSON). It has every figure on this page with its game counts, the positions as FEN, and the method notes. The URL stays the same when we update it.

How to cite

Chessreps, "What chess players actually play at every rating" (2026), https://chessreps.com/openings/what-players-play-by-rating