KnightyLearnChess
Understand the idea

How to build a chess opening repertoire you actually understand

By Nikolas Burk11 min read

Build a repertoire by choosing one opening per side that leads to a pawn structure you understand, learning the plan for that structure instead of long move lists, and re-testing your lines every few months. For most players under 1800 that is one or two openings as White and one defence each against 1.e4 and 1.d4.

That is the whole method. The rest of this post is why it works, which openings fit it, and how to stop your repertoire quietly falling apart.

Why memorised repertoires collapse at move 8

I sat around 1600 on Lichess for a long time with a repertoire that looked respectable on paper: lines memorized well past move ten. And in almost every game, somewhere between move 6 and move 10, my opponent played something I hadn't seen, and I was on my own in a position I didn't understand. My "preparation" had bought me nothing except the illusion of being prepared.

That is the normal failure mode. Below 1800, your opponents don't play the main lines either. A memorised repertoire is a dictionary of answers to questions that nobody at your level asks. The moment the sequence breaks, you have to fall back on something, and if the only thing you learned was the sequence, you fall back on nothing.

What survives contact with a real opponent is structure. Once the pawns settle, every position asks the same question: where are the pawn chains pointing, which pawn is weak, and which break opens the game in your favour. If you know the answer for the structure you are in, it doesn't matter whether White played 4.♘︎f3 or 4.c3. You have a plan either way.

Caro-Kann Advance after 3...♗︎f5. White's chain runs e5–d4; Black's ...c5 break hits its base on d4, and the bishop on f5 is already outside the chain.

Look at the pawns on d4 and e5, not the pieces. That chain fixes the whole game. White has space on the kingside and will build behind it. Black's job is to attack the base of the chain with ...c5, develop around it, and keep the light-squared bishop outside the pawns, which is exactly what 3...♗︎f5 does. Every Caro-Kann Advance game you will ever play comes back to that plan, whatever White does on move 4.

The point of this post is that most openings are like this. There are far fewer structures than there are openings, and a repertoire is really a set of structures you have agreed to play.

How do I build a chess opening repertoire?

Four steps, in order. Pick by structure, pick one opening per side, learn the plan before the moves, and re-test rather than re-read.

Step 1: pick by structure, not by hype

Openings come and go with whoever is winning on YouTube this month. Structures don't. Three families cover most of what you will meet under 1800, and each has a plan you can learn in an afternoon.

The e5 chain (French Advance, Caro-Kann Advance, and the 3.e5 lines of the Alapin Sicilian). White's pawns stand on d4 and e5, pointing at the kingside. White attacks where the chain points. Black attacks the base of the chain with ...c5 and sometimes ...f6, and the single most important question is whether Black's light-squared bishop is inside or outside the pawns. The two boards in this post show the same chain with the bishop in both places.

French Advance after 3...c5. Same chain, same break on d4, but the c8 bishop is shut in behind e6 and d5. One structure, one plan, two openings.

The isolated queen's pawn (Queen's Gambit Accepted, the Caro-Kann Panov, the Sicilian Alapin after ...cxd4, the Tarrasch). One side has a d-pawn with no neighbours. That side gets open lines and active pieces and must attack before the pawn becomes a weakness. The other side blockades the square in front of the pawn, trades pieces, and heads for an endgame where the pawn is simply a target. If you play 1.d4 or the Alapin against the Sicilian, you will have this pawn in half your games, so learn both sides of it.

Caro-Kann Panov after 7...♘︎xd5. White's d-pawn has no neighbours. White attacks with pieces while the position is open; Black sits on d5 and trades.

The open and symmetrical centre (Italian Game, Four Knights, Vienna, Scotch, and the Petroff). Both sides have a pawn on e4 or e5 and the centre is not yet fixed. The game is about development, the timing of the d4 or ...d5 break, and small differences in piece placement. This is where the famous traps live, because an open centre punishes a single slow move. It is also the best family for a first repertoire, because the plans are the most general: develop, castle, fight for the centre with c3 and d4.

Pick the family you enjoy: the chain for slow, closed games with a long plan, the isolated pawn for open positions and piece play, the open centre for learning chess from first principles. The right structure is the one whose plan you can explain in two sentences without looking it up.

Step 2: how many openings should you learn, and which ones?

Three. One as White that you play against everything, one defence to 1.e4, and one defence to 1.d4. You can add a second White opening once the first one is boring you, not before.

Which three depends on how much tactical noise you can handle right now. The table pairs each pick with the structure it leads to and the guide that teaches the plan.

Rating bandAs WhiteAgainst 1.e4Against 1.d4Structures you will meet
Under 1200Italian Game or Vienna, with the Sicilian Alapin against ...c5ScandinavianQueen's Gambit basics, declined with ...e6Open centre; the isolated pawn from the Alapin; a Caro-Kann-style centre after ...c6
1200–1500London System, or Four Knights with the Sicilian Alapin against ...c5Caro-Kann AdvanceQueen's Gambit Declined, from the same guideThe London's d4–e3 pyramid; the symmetrical centre and the Alapin's isolated pawn; the chain as Black
1500–1800QGD Exchange (moving to 1.d4)Caro-Kann, now knowing the Panov tooQueen's Gambit Declined, adding the ...c5 and ...dxc4 ideasThe Carlsbad minority attack; the isolated pawn from the Black side against the Panov

Two things to notice. The Black defences barely change as you climb. The Scandinavian gives you the same kind of structure every single game, which is why it is the right first defence: you reach your position whether your opponent is 900 or 1500, and you practise one plan a hundred times. The Caro-Kann is its natural successor, same centre, richer plans. The White openings move from "develop and castle" to "choose a structure on purpose", which is the actual skill you are building.

Notice also what is missing. No Najdorf, no King's Indian, no Grünfeld. Great openings, terrible first repertoires: they rely on long forcing lines and several structures at once. If you are stuck at a rating, it is far more likely to be a tactics or time-control problem than an opening problem, and a sharper opening will not fix it.

Step 3: what's the best way to learn an opening like the Sicilian or the Italian?

Learn the plan first, then eight moves, then stop.

"Knowing an opening" at 1400 means four things. You can say in one sentence what each side wants in the resulting structure. You know the first eight moves of the main line and why each of them is played. You know the two or three most common ways your opponents deviate at your level, and what to do about each. And you know the one trap in the opening that you are most likely to fall into yourself.

That is it. Not twenty moves, not every sideline. If you can explain why Black plays 3...♗︎f5 in the Caro-Kann Advance before ...e6 (the bishop gets out before the pawns lock it in), you know more about that opening than most 1400s who have memorised it to move 15. The same test works for the Sicilian: if you can say what Black's ...d5 break does, you understand it; if you can only recite the Najdorf move order, you don't.

Resources follow from this. Read a guide that explains the moves, not one that lists them. Play through ten master games in the structure, fast, watching only the pawns. Then play the opening in rapid games and lose with it for a month; every loss in a structure you are learning shows you a plan you didn't know yet. Our opening guides are built this way, main line plus the reason for every move, and the ones in the table are where I'd start.

Step 4: the traps worth knowing

A repertoire of traps is a bad repertoire. But one that doesn't know the four traps everyone under 1600 plays is incomplete. Learn these from both sides, not to win cheap games, but to stop losing them.

  • The Fried Liver Attack in the Two Knights: 4.♘︎g5 against ...♘︎f6, then the knight sacrifice on f7. If you play 1.e4 e5 as Black you will see it in your second week; know your answer to 4.♘︎g5 before then.
  • Légal's mate: the queen sacrifice that punishes an early ...♗︎g4 pin when the c4 bishop and f3 knight are both aimed at f7. The general lesson: a pin is only as strong as the piece behind it.
  • The Stafford Gambit after 1.e4 e5 2.♘︎f3 ♘︎f6 3.♘︎xe5 ♘︎c6: a gambit built entirely on White playing natural moves. Objectively poor, practically lethal, which tells you what "objectively" means at 1300.
  • The Englund Gambit after 1.d4 e5: the same idea against 1.d4. Both are worth one evening, so your repertoire is not a liability on move 2.

Each trap works because one side is playing moves rather than plans. That is the structural argument again, from the other direction.

Step 5: re-test, don't re-read

Here is the part that most repertoire advice skips. Opening knowledge decays, and re-reading the lines does nothing about it. Re-reading feels like studying and tests nothing: you recognise the moves on the page and conclude you know them, which is not the same as being able to produce them against a clock.

The honest test is to play the line from memory, from move one, against something that pushes back. Do that about every 90 days for each opening in your repertoire. If you produce the line and can say why each move is played, leave it alone for another quarter. If you miss a move, that is the move you re-learn, and only that move. Three months is roughly how long a line you don't play regularly stays reliable; if you play a defence every week, your games are the re-test and you can skip it.

This is a fixed rhythm, not a memory-science system. It does not need an algorithm. It needs a date in your calendar and the discipline not to look at the lesson before you play the line.

How to switch openings without losing a year

At some point your Scandinavian will feel small, or your London will stop producing positions you enjoy. Switching is fine. Switching badly costs a year, and it costs a year because people switch structures when they only meant to switch openings.

Switch within a family first. Scandinavian to Caro-Kann is cheap: the ...c6 and ...d5 centre is the same and most of what you know carries over. Italian to Vienna is cheap for the same reason. London to QGD Exchange is a real change of structure, chain to Carlsbad, but you keep 1.d4 and the quiet style, so the change is contained.

Switch one opening at a time, keep the old one for a quarter, and give the new one thirty rapid games before you judge it. Most people abandon a new opening after five losses that had nothing to do with it, and go back to one they had already outgrown.

And if you are switching because you keep losing in the opening, check first that you are losing in the opening. Pull up your last ten losses and find the move where the evaluation dropped. If it is after move 15, the opening was fine and the switch will change nothing. The diagnostic that tells you which plateau you are on also tells you whether your repertoire is the problem.

If you want this done for you

Everything above you can do with a database, a notebook and a calendar. We built LearnChess to do it with less friction, and I should say plainly that it is our product.

Each LearnChess opening guide is written around the plan: the main line one move at a time, the reason for each move, and the common deviations at club level. Add a guide to your repertoire and it tracks what you have studied, what you have mastered, and what is due. "Mastered" means you played the whole line from memory and answered engine-checked questions about the ideas behind it; 90 days later the opening comes up for a re-test, which is Step 5 without the calendar. An Opening Run is the harder version of the same test: you play the line against Stockfish, it leaves the book once with a good move you never rehearsed, and you find out whether you knew the position or just the sequence.

None of that replaces the thinking in Steps 1 to 3. It makes Step 5 happen on its own.

Frequently asked questions

Nikolas Burk

Nikolas Burk

Founder of LearnChess

I plateaued around 1600 on Lichess and built LearnChess to understand my own games more deeply: engine lines in plain English, questions about the board in natural language, and a structured path of lessons, practice and tests. On the blog I share what I learn along the way.

More from the blog

All articles