Best Chess Openings for Beginners, Ranked by Data

Every list of the best chess openings for beginners ranks them by taste. This one ranks them by how they actually score: 2.08 billion real games between 1000 and 1600, one number per opening, pulled from the Lichess database and shown below with the method. Four of the openings that appear on almost every beginner list score below average for the side playing them, and the one every list opens with is only the fifth best White opening measured here.

What "best" means here, and why the percentages you have seen are misleading

"Best" on this page means one thing: the opening scores well when players at your strength play it. Not that a grandmaster would approve of it. Every number comes from the Lichess opening explorer, filtered to blitz and rapid games in the 1000 to 1600 rating band, and each opening is measured at the position that defines it. Score is the normal chess convention, (wins + draws/2) / games, from the point of view of the side who chose the opening.

Here is the part every other list gets wrong. White already scores above 50% before choosing anything. So a White opening at 52.9% is not 2.9 points ahead of the field, it is 1.0 point ahead. And a Black defense at 48.3% is not losing, it is exactly average. You have to compare each opening against its own side's baseline, or you will rate every White opening as good and every Black one as bad. It also means White scores and Black scores are never comparable to each other, only each to its own baseline.

Two honest caveats, because a number without its limits is just a claim. Each opening's score is conditional on reaching it: the Italian Game figure only counts games where Black cooperated with 1...e5 and 2...Nc6. That is the right question for a beginner ("if I get to play this, how does it go?") but it is not a measure of objective strength. And these are averages over millions of players who all had to keep playing after move three, so what they really measure is how easy an opening is to play well. That is exactly what you want when choosing your first repertoire, and it is not the same list an engine would give you.

One more thing about the five picks, so the ranking is auditable. They are the highest scorers that also do a job: one opening as White, one answer to 1. e4, one answer to 1. d4, and a free course to drill each. Two openings outscore some of them and are deliberately not in the list. The Queen's Gambit is +2.6 and beats the London, at the price of learning answers to the Slav, the Queen's Gambit Declined and half a dozen Indian defenses. The gambits outscore everything and get their own section at the end. Both are named with their numbers rather than quietly dropped.

The four principles that beat any opening choice

Compare like with like and the whole spread on this page is small: from the best White opening (56.6%) to the worst (50.2%) is six points, and the Black defenses run from 50.8% down to 42.6%. Hanging a piece in a club game costs you far more than any of that. So before the list, the four rules that decide the other 30 moves. An opening is only a way of reaching these faster.

1. Scotch Game, 55.5% for White: your opening with the white pieces

1. e4 e5 2. Nf3 Nc6 3. d4. White scores 55.5% across 48.0 million games, which is 3.6 points above what White scores after 1. e4 in general. Among openings a beginner can safely be handed, nothing else on the board comes close, and it beats the Italian Game, the pick every other beginner list opens with, by 2.6 points.

Why it works at club level: the third move trades your d-pawn for Black's e-pawn and rips the centre open immediately. That matters because the way beginners lose with 1. e4 e5 openings is by drifting in a closed, slow position with no idea what to do. The Scotch removes the option. Pieces come out fast, lines are open, and the player who counts better wins.

The one line to know: after 4... Bc5, the natural 5. Nxc6 runs into 5... Qf6, which hits the knight and looks down the f-file at once. It is perfectly fine for White, you just need the reply ready: 6. Qd2 or 6. Qf3. If you would rather not deal with it, 5. Be3 holds everything together. Train the Scotch line by line and the board plays Black's most common replies back at you.

2. Caro-Kann Defense, 50.8% for Black: the cheapest answer to 1. e4

1. e4 c6, and next move d5. Black scores 50.8% over 72.8 million games, 2.7 points above Black's 48.1% baseline against 1. e4. It ties the Sicilian for the best margin over baseline of any Black defense measured here, and it costs a fraction of the study.

The trade you are making: you support d5 with a pawn instead of a piece, so you get a solid structure and no early tactics to memorise. You give up speed. This is the defense for a player who would rather understand one position than remember ten. After the near-universal 2. d4 d5 White has four serious tries and only two show up often at club level, so your homework is genuinely small. White can also duck with 2. Nc3, 2. Nf3, 2. c4 or the 3. f3 Fantasy, all rarer and none of them dangerous once you know your structure.

Train the Caro-Kann against all four. It is the course to take if you want one defense and no maintenance.

3. Sicilian Defense, 50.8% for Black: the same score, three times the work

1. e4 c5. Black scores 50.8% across 164.1 million games, the largest sample of any named defense here apart from 1... e5 itself, and the same 2.7 points above baseline as the Caro-Kann. Identical result, very different price.

The reason it is fourth and not first is the bill. White decides the battleground on move two, and roughly half of your club games never reach the Sicilian you studied: they turn into an Alapin, a Closed Sicilian, a Smith-Morra or a bishop out to c4. So a Sicilian repertoire is one system plus a prepared answer to five sidelines, where the Caro-Kann is one structure plus four replies. Take it because you want to play for the win from move one with an unbalanced position, not because it scores higher, because it does not.

If you take it, train the Sicilian with the anti-Sicilians included. A Sicilian course that only teaches you the main line has taught you half your games. Chessigma's teaches the Dragon as its Open Sicilian system, which is the point of picking one: you get the same picture every game.

4. King's Indian Defense, 49.1% for Black: your answer to 1. d4

1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6, the real King's Indian tabiya rather than the 2... g6 junction it shares with the Gruenfeld and the Benoni. Black scores 49.1%, which is 1.8 points above Black's 47.3% baseline against 1. d4, and that makes it the only named answer to 1. d4 measured here that clears its baseline by more than a point. Every classical reply scores worse: the Nimzo-Indian is +0.7, the Dutch +0.1, the Slav 0.4 below, the Queen's Gambit Declined 1.3 below, and the Queen's Gambit Accepted 4.7 below.

It works for the same reason the London works for White: it is a setup, not a theory tree. You play the same five moves against almost anything, then you have one clear plan.

Train the King's Indian move by move, including the e5 break, which is the move that decides whether the whole setup was worth playing.

5. London System, 54.4% for White: one setup against everything

1. d4 d5 2. Bf4. White scores 54.4% over 53.3 million games, 1.7 points above the 52.7% White scores after 1. d4. That figure is measured at the position above, the accelerated move order against 1... d5, which is where roughly half of your 1. d4 games go. It has the smallest edge of the five and the Queen's Gambit beats it at +2.6, and it is still here, because it is the cheapest opening in chess to own: the same eight moves against almost anything.

The honest downside is the mirror of the upside: autopilot gets you a playable position and nothing more, so the London rewards knowing your two or three attacking patterns. Train the London System and you will meet all of them inside an hour.

The openings to skip, and the three that outscore all five

Three openings score higher than the Scotch and are still not on the list. The 4. Ng5 Knight Attack, the gateway to the Fried Liver, is the best-scoring White position measured here at 56.6%. The Danish Gambit is at 55.8% and the King's Gambit at 55.6%. They are missing for two reasons, and neither is that they are bad.

Now the ones to leave alone, all of them measured below their own side's baseline:

Two more names for the list, since neither has a Chessigma course and both turn up on competitor beginner lists: the Slav above, and the Petrov, whose other spelling (Petroff) is how you will see it in older books.

One opening that deserves a correction rather than a warning: the Scandinavian Defense scores 48.3%, which looks awful next to the London's 54.4% until you remember that Black's baseline against 1. e4 is 48.1%. It is exactly average, which is a fair price for the fastest defense in chess to learn. It is not in the top five, and it is not a mistake either.

Here is every opening measured, ranked by how far it beats its own side's baseline. Green means the side named actually gains ground by choosing it.

Source: the Lichess opening explorer, blitz and rapid, rating buckets 1000, 1200 and 1400 (Lichess buckets are lower bounds, so the band runs to 1599), pulled 13 August 2026. Each opening is measured at the position shown in the Moves column. Check any row yourself: the explorer is public.

Frequently asked questions