Playing10 min read

Chess Gambits: How Often Each Rating Accepts Them

Every list of chess gambits ranks them by fame. Chessdrive's Maia3 and Stockfish data shows how often each rating accepts six of them, from the King's Gambit to the Benko, and that accepting is the engine's top move in all six.

The starting position

A gambit is an opening in which a player sacrifices material, usually a pawn, hoping for a positional advantage in return, and the other side then accepts or declines the offer. Every page that ranks gambits does it by reputation or by a win rate with no rating attached. We asked a different question: at each rating, how often does the defender actually take the pawn, and does the engine agree?

We ran the maia3-5m checkpoint, see how Maia chess actually works, on the position where the pawn is offered in six gambits, at seven rating bands from 1000 to 2200. We then scored the capture with Stockfish 18 at depth 20, see what the Stockfish chess engine is.

One point to hold onto. Maia3 is a model that predicts which move a player of a given rating is likely to choose. Every percentage below is a model prediction of what players at that rating play, not a count from real games. We tried to check against real games from the Lichess explorer and access was refused, so no real game shares appear here.

What is a gambit, and what does accepted or declined mean?

The test is simple: the player offering the gambit gives up material on purpose. Wikipedia describes it as sacrificing material with the aim of a later positional advantage. The offer is then called accepted if the opponent takes the material and declined if they do not.

Gambits sit on a spectrum of soundness. Against a moderately sound one, the usual advice is to accept and give the material back at a favourable moment. Against a less sound one, the accepting side is more likely to keep the extra pawn. Often a gambit can be declined with no disadvantage at all. That is why "accept or decline" is a real choice and not a formality, and why the percentages below are worth knowing.

Is the Queen's Gambit a gambit?

Strictly, no. After 1.d4 d5 2.c4, White appears to give up the c-pawn, but Wikipedia says Black cannot retain it without incurring a disadvantage, and that the gambit label has been questioned for this reason. In practice White recovers the pawn, so nothing is really sacrificed. The name is too established to drop, and the Queen's Gambit Accepted is still filed with the gambits.

That is also why it is not in our table. The data we have for it starts after 2...dxc4, so it has no row where the pawn is first offered. For the lines themselves, see our Queen's Gambit guide and the Queen's Gambit Accepted.

How often each rating accepts six gambits

A note on signs before the table. In each row, the evaluation shown is from the point of view of the side deciding whether to accept, the side to move. For the King's Gambit, Evans, Danish and Smith-Morra that is Black. For the Englund and Benko it is White. A positive number means the accepting side is ahead by the engine's count. In each row the bracketed score is the capture, and in all six it is also Stockfish's top move.

Gambit and accepting move (Stockfish score)1000120014001600180020002200
King's Gambit, 2...exf4 (+0.64)42.7%44.0%43.4%42.9%42.8%42.1%40.1%
Englund Gambit, 2.dxe5 (+1.38)52.9%58.9%61.8%63.1%64.5%67.3%72.3%
Evans Gambit, 4...Bxb4 (+0.21)61.0%61.3%66.3%72.8%79.1%84.7%87.3%
Danish Gambit, 3...dxc3 (+0.33)72.7%73.9%74.0%73.9%72.1%65.2%51.1%
Smith-Morra Gambit, 3...dxc3 (+0.37)66.8%70.6%72.4%73.1%71.7%64.3%52.3%
Benko Gambit, 4.cxb5 (+0.66)30.4%25.3%24.3%26.8%34.9%53.0%70.4%

Four patterns stand out.

  • Accepting is the engine's choice every time. The capture ranks first in all six, so the cost of accepting, measured against the engine's best move, is zero. The scores are small, though. The Evans capture is worth only +0.21, so Black's edge there is thin.
  • Some gambits get accepted more as rating rises. The Evans climbs from 61.0% at 1000 to 87.3% at 2200, the Englund from 52.9% to 72.3%. Stronger players find the engine move more often.
  • The Danish and Smith-Morra go the other way at the top. The Danish peaks at 74.0% at 1400 and falls to 51.1% at 2200. The Smith-Morra peaks at 73.1% at 1600 and falls to 52.3%. At club level most defenders grab the pawns, and only the strongest bands start to decline.
  • The Benko has a dip. White accepts with 4.cxb5 30.4% of the time at 1000, drops to 24.3% at 1400, then climbs to 70.4% at 2200. At 1000 to 1600 the likeliest move is not the capture but 4.b3, at 33.2%, 37.3%, 40.1% and 40.1%.

The King's Gambit is the odd one out. The accepting share barely moves, 40.1% to 44.0% on every band, so whatever your rating, roughly four in ten defenders take the pawn.

White gambits

King's Gambit: the disputed one

The line is 1.e4 e5 2.f4. Wikipedia traces it to Lucena's 1497 book, and it was one of the most popular openings until the late nineteenth century. Steinitz argued it was flawed, and Fischer wrote in the 1960s that it loses by force. Our data does not back that up as far as the capture goes: Stockfish ranks 2...exf4 first at +0.64 for Black. So the honest summary is that the gambit's soundness is disputed, but taking the pawn is the engine's choice. Wikipedia adds that the gambit is very rare at the highest levels and has never lost its popularity at club level.

Our model has 42.7% of 1000 rated players accepting, and it stays between 40.1% and 44.0% all the way up. The full move by move breakdown, including the Falkbeer Countergambit and the Muzio and Salvio, is on our King's Gambit page.

Evans Gambit: the one strong players take

The line is 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.b4. It is named for the Welsh sea captain William Davies Evans, and the first recorded game, Evans against McDonnell, was played in London in 1827. Wikipedia says it has proven resilient under modern analysis with no clear refutation, and that most commentators think declining it with 4...Bb6 is weaker than accepting, because of the wasted tempo.

Our numbers agree on the engine side: 4...Bxb4 ranks first at +0.21. They also show the strongest rating effect in the table. Maia3 gives the capture 61.0% at 1000, 72.8% at 1600 and 87.3% at 2200. If you play the Evans, expect a stronger opponent to take the pawn and make you prove the compensation. The rest is on our Evans Gambit page.

Danish Gambit: two pawns for development

The line is 1.e4 e5 2.d4 exd4 3.c3, a variation of the Center Game. Wikipedia names the Danish player Martin Severin From, who played it at Paris 1867, but says it may have been known earlier, so we do not say he invented it. Hans Lindehn beat Steinitz with 3.c3 in London in 1864.

Here a source disagrees with the engine. A Chess.com blog warns that Black will find themselves in trouble very soon if not extremely accurate, and, as of November 2026, calls the gambit's 58% win rate massive, with no rating filter given. Wikipedia says Black can accept one or both pawns safely, and that most theorists rate the position after Schlechter's defence (4.Bc4 cxb2 5.Bxb2 d5) as equal. Stockfish ranks 3...dxc3 first at +0.33. The blog's warning is about practical trouble, not about the capture being a mistake, and a win rate for the whole opening cannot tell you whether taking the pawn was the problem.

The model says 72.7% accept at 1000, a peak of 74.0% at 1400, then a fall to 51.1% at 2200. The details are on our Danish Gambit page.

Smith-Morra Gambit: where strong players start declining

The line is 1.e4 c5 2.d4 cxd4 3.c3, named after Ken Smith of the Dallas Chess Club and the Frenchman Pierre Morra, and usually just the Morra Gambit in Europe. Wikipedia says accepting gives Black an extra pawn and a central majority, that many Sicilian players decline altogether, and that a Black who keeps the pawn has good chances in the endgame.

Our data splits that claim by rating. Declining is a strong player's habit, not a club player's. The model has 66.8% accepting at 1000 and a peak of 73.1% at 1600, then 52.3% at 2200, so at the top nearly half the field plays something else. Stockfish still ranks 3...dxc3 first at +0.37. See our Smith-Morra Gambit page for the lines.

Black gambits

Benko Gambit: the dip at 1400

The line is 1.d4 Nf6 2.c4 c5 3.d5 b5, a Benoni variation also called the Volga Gambit. Wikipedia credits Pal Benko with promoting it from the late 1960s, and his book appeared in 1974. Black's compensation after 4.cxb5 a6 5.bxa6 Bxa6 is speed of development, the long diagonal and pressure on the half open a and b files.

White is the one deciding here, and the numbers are unusual. The capture 4.cxb5 is Stockfish's top move at +0.66, yet only about one in four players at 1400 plays it (24.3%). At 2200 it is 70.4%. Wikipedia treats 4.cxb5 as the main line and says Black has good practical chances at amateur level, which fits: at club ratings, White often declines before Black's plan gets going. Our Benko Gambit page goes through it.

Englund Gambit: the weakest of the six

The line is 1.d4 e5. Wikipedia lists it as the Englund Gambit and the Charlick Gambit, after Henry Charlick, who introduced 2...d6 in the early 1890s. Fritz Englund only sponsored a thematic tournament in 1932 and 1933, and the main line 2.dxe5 Nc6 3.Nf3 Qe7 was introduced by Karlis Betins. Avrukh calls 1...e5 the worst possible reply to White's first move.

Our data fits. Stockfish gives 2.dxe5 +1.38, the largest edge for the accepting side in the table. Wikipedia calls declining inaccurate. Even so, only 52.9% of 1000 rated players accept, rising to 72.3% at 2200, so a Black player at club level meets a decline about as often as a capture. Another reason to read our Englund Gambit page is its trap with 6.Bc3, which also appears in our chess traps guide.

How to pick a gambit at your rating

  • Playing it as White around 1000 to 1400. The Danish (72.7% to 74.0%) and the Smith-Morra (66.8% to 72.4%) are the ones the model says your opponent will take most often. The Evans is lower at these ratings, 61.0% to 66.3%, and climbs as opponents get stronger.
  • Playing it as White at 2000 and above. Expect more declines. The Danish and Smith-Morra fall to 51.1% and 52.3% at 2200, so have a plan against the decline, not just the accepted lines. The Evans is the exception, with 87.3% of 2200 players taking.
  • Playing it as Black. The Englund and Benko need White to accept, and at club level the Benko is declined more often than not. Know your answer to 4.b3 before you commit to it.
  • Facing a gambit. Accept. The engine says the capture is the best move in all six. Take the pawn, get developed and be ready to give it back, which matches what Wikipedia says about moderately sound gambits.

If you like attacking chess in general, the Fried Liver Attack is another early sacrifice, and it is covered by rating there too. And if you want to know which of your opponent's natural looking moves can be punished, our chess traps guide has the same rating by rating treatment.

Where other pages are better

As of November 2026, Wikipedia's list of chess gambits is the best catalogue, with roughly 290 entries grouped by parent opening, each with its moves and ECO code. It makes no soundness statements and does not include the Englund. The Chess.com blog "Top 5 Best Chess Gambits" gives Lichess win rates of 55% to 69% for five gambits, but with no rating filter, no lines and no acceptance data.

What none of the pages we could see shows is how often a gambit is accepted at a given rating, or whether the engine prefers the capture. That is the gap this page fills, with the limits stated at the top: model predictions, not real games, and the position where the pawn is first offered, not what happens afterwards.

What to take from it

Accepting a gambit is the engine's top move in all six we measured, but the scores are modest and humans do not always play it. The Evans is accepted more as rating rises, the Danish and Smith-Morra less, the Benko dips at 1400, and the King's Gambit does not move at all. Pick the gambit by who your opponent is likely to be, not by its fame.

Frequently asked

What is a gambit in chess?
A gambit is an opening in which a player sacrifices material, usually a pawn, to get a positional advantage later, such as faster development or control of the centre. The offer is then accepted if the other side takes the material and declined if it does not. Wikipedia notes gambits are played more often by White.
What are the best chess gambits?
There is no single answer, because how often your opponent takes the pawn depends on their rating. Maia3 gives the Evans capture 61.0% at 1000 and 87.3% at 2200, while the Danish goes the other way, from 72.7% at 1000 to 51.1% at 2200. Pick the gambit whose accepted lines you will actually reach against the players you meet.
Should you accept or decline a gambit?
In all six gambits we measured, Stockfish 18 at depth 20 ranks the capture first. The scores are modest, from +0.21 in the Evans to +1.38 in the Englund, for the side that accepts. Wikipedia says that against a moderately sound gambit the usual plan is to take the material and return it at a good moment, and that a gambit can often be declined with no disadvantage.
Is the Queen's Gambit really a gambit?
Not in the strict sense. After 1.d4 d5 2.c4, Wikipedia says Black cannot retain the pawn without incurring a disadvantage, so White is not truly giving it up. The name stuck anyway, and the Queen's Gambit Accepted is still catalogued as a gambit.
Which gambit is best for beginners?
At 1000 the model has Black taking the pawn 72.7% of the time in the Danish, 66.8% in the Smith-Morra and 61.0% in the Evans, so all three usually get you the open game you are playing for. The Danish costs two pawns for development, so treat it as a sharp weapon rather than a safe choice.
Did Fischer refute the King's Gambit?
He claimed it loses by force, in his 1960s article "A Bust to the King's Gambit". No refutation has been proven, and in our data Stockfish ranks 2...exf4 first at +0.64 for Black, so taking the pawn is the engine's choice even if the gambit is disputed. It remains rare at the top level and popular at club level.

Want to offer a gambit to a player at your rating?

Chessdrive runs the real maia3-5m checkpoint on-device as your opponent at any rating you set from 600 to 2600, so you can play any gambit in this guide from either side and find out whether a player at your level takes the pawn. Finished games are reviewed separately with Stockfish so you can see exactly where the position turned. Available now on iOS.

Keep reading

All guides

Updated