The Grob Opening is 1.g4, an irregular flank move with ECO code A00. White does not fight for the centre with a pawn. The plan is a bishop on g2 that rules the long diagonal and presses on d5 and b7, which only works if White keeps the centre free of its own pawns. Wikipedia says it is widely considered one of the worst possible first moves for White, and the same page quotes John Watson: against an informed or skilled opponent it is simply masochistic. Yet it keeps turning up below master level. This page shows why, with numbers.
The pages we opened give the reputation, the names and a handful of lines. None shows what players at 1000 or 2000 actually reply to 1.g4, or which replies punish it. We ran the maia3-5m checkpoint (see how Maia chess actually works) on two Grob positions at seven rating bands from 1000 to 2200, and graded every legal move with Stockfish 18 at depth 20. Percentages in the tables are the model's predicted move probabilities for a player at that rating. They are not game database frequencies and they are not win rates. We have no game results for this opening.
Every table evaluation is from White's side, so -1.27 means Black is better by 1.27 pawns. The two dataset positions are the one after 1.g4, and the one after 1.g4 e5 2.Bg2, both with Black to move. The second is the position reached by the most popular continuation at 1400. Our line checks are separate local Stockfish runs, called local checks below. In prose we say which side is better to avoid sign confusion.
Where the Grob comes from
The opening is named after Henri Grob, a Swiss international master born on 4 June 1904 who died in 1974. Per Wikipedia he won the Swiss championship in 1939 and 1951, played four Olympiads and played 3,614 correspondence games against readers of the Neue Zuercher Zeitung, winning 2,703. He wrote a book on the idea, Angriff g2-g4, in 1942. In his own newspaper column he called it the Spike, and the list of other names is long: Ahlhausen's Opening, the Genoa Opening, the San Pier d'Arena Opening, Fric's Opening and the Kolibri.
The modern champion is Michael Basman, the English IM who died in 2022. ChessBase says he popularised the Grob and its inverse 1...g5, the Borg, and wrote books on both. Wikipedia calls his 1978 win over John Nunn perhaps the highest-profile Grob game. We could not find the move list or the event, so we do not analyse it here.
Move 1: the engine wants 1...d5, humans often play 1...e5
After 1.g4, Stockfish ranks 1...d5 first at -1.27, then 1...c5 at -0.82, 1...e5 at -0.79 and 1...Nc6 at -0.76. Every reply we list is good for Black, which tells you how bad 1.g4 is. The question is which good reply humans find.
| Rating | 1000 | 1200 | 1400 | 1600 | 1800 | 2000 | 2200 |
|---|---|---|---|---|---|---|---|
| 1...d5 (-1.27) | 25.9% | 29.0% | 32.6% | 36.8% | 41.5% | 46.8% | 51.4% |
| 1...e5 (-0.79) | 47.1% | 42.7% | 35.9% | 28.6% | 21.7% | 15.6% | 11.6% |
| 1...c5 (-0.82) | 3.0% | 4.2% | 5.5% | 6.7% | 7.7% | 8.3% | 8.3% |
| 1...e6 (-0.51) | 4.7% | 5.3% | 6.1% | 6.8% | 6.9% | 6.2% | not top five |
| 1...g6 (-0.49) | not top five | 3.0% | 3.7% | 4.4% | 4.8% | 4.9% | 5.0% |
The pattern is steady and almost linear. The best move 1...d5 rises at every step, from 25.9 percent at 1000 to 51.4 percent at 2200. The rival 1...e5 falls from 47.1 to 11.6 percent. The crossover is at 1600, where 1...d5 is predicted at 36.8 percent and 1...e5 at 28.6 percent. At 1400 the order is still reversed, with 1...e5 at 35.9 and 1...d5 at 32.6.
That is the Grob's real habitat. Below 1600, the single likeliest reply is 1...e5, which the engine rates -0.79: still a clear Black edge, but 0.48 short of the best move. In practice the player who tries 1.g4 against a 1200 is facing the strongest reply less than a third of the time. From 1600 up, the best move is also the favourite, and the surprise is gone.
Chess.com says 1...d5 is the most critical line, and gives White's choices after it as 2.h3 or the more common 2.Bg2. Chess.com calls 2.h3 inaccurate because of 2...h5. Our dataset does not cover that position, so we leave it at the claim.
The 1...e5 branch: where the slip lives
The popular road at 1400 continues 1.g4 e5 2.Bg2, which the model plays with a probability of 79.4 percent at that rating. Stockfish ranks the replies close together: 2...Nc6 first at -0.85, 2...h5 at -0.84 and 2...d5 at -0.76.
| Rating | 1000 | 1200 | 1400 | 1600 | 1800 | 2000 | 2200 |
|---|---|---|---|---|---|---|---|
| 2...d5 (-0.76) | 31.3% | 40.0% | 45.7% | 50.6% | 56.1% | 60.9% | 61.0% |
| 2...Nc6 (-0.85) | 16.4% | 16.5% | 15.8% | 14.8% | 13.6% | 13.1% | 15.4% |
| 2...Nf6 (+0.67) | 17.9% | 13.3% | 9.5% | 6.9% | 4.8% | not top five | not top five |
| 2...Bc5 (-0.63) | 8.5% | 8.1% | 8.3% | 8.2% | 7.7% | 6.8% | 5.6% |
| 2...d6 (-0.24) | 7.3% | 8.1% | 8.8% | 8.9% | 8.3% | 6.9% | 5.3% |
Two things stand out. First, 2...d5 is the likeliest move at every band, from 31.3 percent at 1000 to 61.0 percent at 2200, and it is only 9 centipawns behind the best move 2...Nc6. So the engine move is almost never missed in a way that matters. A local check of the position after 2...d5 shows White's best is 3.d4, and Black stays better by 0.73. That is a middling edge compared to the Black advantage at move 1.
Second, the real rating-linked slip is 2...Nf6, which scores +0.67 for White. It is predicted at 17.9 percent at 1000, 13.3 at 1200, 9.5 at 1400, 6.9 at 1600 and 4.8 at 1800. At 1000, roughly one Black player in six turns a Black edge of 0.85 into a White edge of 0.67, a swing of 1.52 pawns. The reason is tactical. After 2...Nf6 our local check says White's best is 3.g5, kicking the knight, at +0.69 for White. Black has spent a tempo to have the knight chased again, and the Grob suddenly looks fine. If you play 1.g4 e5 2.Bg2 Nf6 as White against a 1000 player, this is the position you hope for.
The Fool's mate trap
The first trap is the one you should avoid as White. After 1.g4 e5, the natural looking 2.f3?? loses at once to 2...Qh4#. Our local check confirms mate in one. The other candidate, 2...h5, comes out at only +2.55 for Black in the same position, so the mate is the clear move.
This is the Fool's mate pattern with the move order flipped. Wikipedia's page on the Fool's mate lists 1.f3 e6 2.g4 Qh4# and names the Grob as a parent opening, and it cites the Oxford Companion to Chess for the version where White starts with g4 and plays f3 or f4 second. It matters because 1.g4 e5 is the most likely reply at 1000, and 2.f3 looks like a harmless support move to anyone who forgets that g4 and f3 together open the h4 to e1 diagonal straight to the king.
There is also a plain blunder for White: 1.g4 e5 2.g5, which simply hangs the pawn to 2...Qxg5. A Lichess game between players rated 1001 and 1236 shows how fast it can go wrong after that: 3.Nf3 Qf6 4.Nxe5 Qxe5 5.Bg2 Qf6 6.Bxb7 Bc5 7.Bxa8 Qxf2#. We replayed it and every move is legal and the final position is mate, but it is one game between club players, not a forced sequence.
If White replies 2.h3 instead of 2.Bg2, a local check says Black is better by 1.26, with 2...Nc6 the best move. So there is no safe waiting move to hide behind on move 2.
The Grob Gambit
The main line against 1...d5 is 1.g4 d5 2.Bg2, after which Black can take the pawn: 2...Bxg4 3.c4. This is the Grob Gambit. White offers a pawn for pressure on d5 and b7, per Wikipedia. Chess.com says most traps in the Grob occur in this accepted line, and that Black can still gain a significant edge by giving back a pawn or two for development.
Our checks agree that taking is the strongest decision, and the margin is large. After 2.Bg2, Stockfish's top move is 2...Bxg4 at +1.36 for Black. Then 2...Nc6 is +1.19 for Black and 2...c6 +1.08. So the gambit declined is slightly better for White than the gambit accepted, but no line is close to equal.
- After 3.c4: the best move is 3...c6 at +1.21 for Black, with 3...Nf6 at +1.12 close behind.
- After 3...c6 4.cxd5 cxd5 5.Qb3: White hits b7 and the d5 pawn. 5...Nf6 keeps +1.43 for Black, with 6.Nc3 e6 7.Qxb7 Nbd7 as the engine line.
- The slips: 5...Qc8 is only +0.54 for Black, and 5...e5 +0.51. Both give back about a pawn.
That gap is the lesson of the whole opening. Black can be a pawn up and still lose most of the advantage with one natural-looking move, because the b7 pawn and the long diagonal never stop mattering. Chess.com calls the declined line 1.g4 d5 2.Bg2 c6 the most popular and best scoring, with no figures, and we cannot verify that. Our local check puts White's best reply to 2...c6 at 3.g5, with a Black edge of 0.90.
The named defences
Wikipedia describes the Keene Defence as 1.g4 d5 2.h3 e5 3.Bg2 c6, a strong Black setup, and says Richard Palliser prefers 3...Nc6 to 3...c6. Our dataset does not hold those positions and we did not run a check on them, so we repeat the claim as sourced and no more. TakeTakeTake also lists the Spike Attack and the Romford Countergambit as variations without any detail. Do not treat any of these names as universal, and expect other pages to label the same position differently.
Engine versus database
Here the evidence splits. Wikipedia says that in the TCEC season 12 superfinal, Stockfish and Komodo both rated the position after 1.g4 as a clear advantage for Black. Our data agrees: the best reply is -1.27 for White, and 12 of the 16 replies we list are at least 0.4 better for Black. TakeTakeTake puts it plainly: according to engines, 1.g4 is one of White's worst possible first moves.
The same Wikipedia page cites a chessgames.com database, 477 games as of October 2026, in which the Grob wins at least as often as it loses against almost any Black reply. These two facts do not contradict each other. A database of games records what people played, and we just saw how often people miss the engine reply, especially under 1600. The count will change, and it gives no rating split, so we do not read it as a rating-by-rating result.
Chess.com calls it a dubious flank opening, rarely used by masters but relatively common among club players. It lists surprise and the temptation it creates for reckless play as pros, and a weakened kingside, weaker performance at longer time controls and a Black edge after the right replies as cons. NM Kevin Bordi is the streamer chess.com names as playing it regularly.
Who should play it, by rating
Our data has no game results by rating for the Grob, so this is what the model and the sources suggest, not a result table.
- Around 1000 to 1400, as White. This is the only band where the surprise is real. Black replies 1...e5 more often than the best move 1...d5, and 2...Nf6 appears 9.5 to 17.9 percent of the time after 2.Bg2. Learn the 3.g5 answer and never play 2.f3.
- Around 1600 to 1800, as White. 1...d5 is now the likeliest move at 36.8 to 41.5 percent. You will meet the Grob Gambit and the pressure on b7 more often than the Fool's mate error.
- 2000 and above. More than 46 percent of predicted replies are 1...d5 and the 2.Bg2 position is met by 2...d5 about six times in ten. You are playing a position the engine thinks is over a pawn worse.
- As Black, at any rating. Play 1...d5, take on g4 if offered, answer 3.c4 with 3...c6 and 5.Qb3 with 5...Nf6. After 1.g4 e5 2.Bg2, play 2...d5 or 2...Nc6 and avoid 2...Nf6.
If you like the idea of a surprise opening that bets on the opponent's mistakes, the Englund Gambit page covers the Black side of that bet, and the Dutch Defense page covers a more respectable flank reply. For how all of this fits with other openings by strength, see chess openings by rating. If your real problem is the middlegame, how to get better at chess will do more for you than any opening trick.
Where other resources are better
This page is narrow on purpose. The dataset holds two positions, and Maia3 5m is a small model of human choice, so its percentages are predictions. Our line checks are quick local runs, not a repertoire book. Chess.com's opening page is the best place for a plain tour of the main branches, and Wikipedia is better for the names, the history and Henri Grob's biography. TakeTakeTake and chess.com both discuss traps in general, while we verified only a few exact lines. We have no game results by rating, and we did not open the Lichess explorer, which is the place for real game counts. Several pages in the search results were unreadable. None of the pages we did read compares the engine's best move with the likeliest human move at each rating, which is the gap this one fills.
What to take from it
After 1.g4, the best reply 1...d5 at -1.27 is predicted at 25.9 percent at 1000 and 51.4 percent at 2200, and 1...e5 is the favourite until 1600. After 1.g4 e5 2.Bg2, 2...d5 is the favourite everywhere and only 9 centipawns behind best, while 2...Nf6 at +0.67 is the low-band slip, fading from 17.9 percent at 1000 to 4.8 percent at 1800. The Fool's mate trap 2.f3?? Qh4# is a danger to White, not to Black. The Grob is objectively dubious, over a pawn worse by the engine, and the database result that it scores fine says more about the replies under 1600 than about the opening. Keep every percentage for what it is: the prediction of a model, not a count of real games.
Frequently asked
- Is the Grob opening good?
- Objectively no. In our Stockfish data the best reply to 1.g4 is 1...d5 at -1.27 from White's side, over a pawn for Black. Wikipedia says it is widely considered one of the worst possible first moves for White, and that Stockfish and Komodo both gave Black a clear advantage after 1.g4 in the TCEC season 12 superfinal. It still surprises players, which is why it lives at club level.
- Why does the Grob win so many games if it is bad?
- Because most players do not find the best reply. Our model predicts 1...d5 at only 25.9 percent at 1000 and 36.8 percent at 1600, while 1...e5 is the likeliest reply up to 1400. Wikipedia also cites a chessgames.com database (477 games as of October 2026) in which the Grob wins at least as often as it loses against almost any reply. That database is practical evidence and the engine is objective evidence, and the two disagree.
- What is the best reply to 1.g4?
- 1...d5, which hits the g4 pawn and opens the bishop on c8. Our data rates it -1.27 for White, ahead of 1...c5 at -0.82, 1...e5 at -0.79 and 1...Nc6 at -0.76. After 2.Bg2 a local Stockfish check puts the best reply 2...Bxg4 at +1.36 for Black.
- What is the Grob Gambit?
- 1.g4 d5 2.Bg2 Bxg4 3.c4, where White gives a pawn for pressure on d5 and b7, according to Wikipedia. Our local check says Black keeps the edge: after 3.c4 the best move is 3...c6 at +1.21 for Black. Chess.com says most Grob traps live in this accepted line.
- Who invented the Grob?
- It is named after Henri Grob, a Swiss international master born 4 June 1904 who died in 1974. He called it the Spike in his newspaper column, and Wikipedia lists other names including Ahlhausen's Opening, the Genoa Opening and the Kolibri. Michael Basman popularised it and its inverse 1...g5, the Borg.
- Is there a Fool's mate trap in the Grob?
- Yes. After 1.g4 e5, the move 2.f3?? allows 2...Qh4#, and our local Stockfish check confirms mate in one. Wikipedia's Fool's mate page, citing the Oxford Companion to Chess, lists this g4 first move order. It is the Fool's mate pattern with White's two pawn moves swapped, and it only happens if White plays the second move carelessly.
Want to face 1.g4 against a model that answers like 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 the Grob from either side against a model that picks its moves the way your band does, then see where the game turned with a separate Stockfish review. Available now on iOS.
Keep reading
Updated