How to build a chess opening repertoire that survives contact

Almost every repertoire fails in the same place. Not in the first six moves, which everybody learns, and not in the deep theoretical arguments, which almost nobody reaches. It fails around move nine, in a perfectly normal position that was never in the book, against a move the engine calls the fourth best, played by someone who has no idea they have just left theory.

That gap is the whole problem. A repertoire is usually built as a list of moves, and moves run out. What you actually need is something closer to a map: a set of positions you understand, a handful of plans attached to each of them, and a clear idea of what you are trying to make the game look like. This is how to build that.

01Start by deciding what the repertoire is for

Before choosing a single line, be honest about the job. A repertoire built to reach objectively equal positions and a repertoire built to reach positions you enjoy playing are different objects, and the second one wins far more games below master level.

There are really only three sane goals, and you should pick one as primary:

1. 01Reach a middlegame you understandYou want the same structures over and over so that your plans come pre loaded. Slower to feel exciting, fastest to actually raise your rating.
2. 02Reach a middlegame your opponent does not understandYou accept slightly worse objective positions in exchange for unfamiliarity. Effective, and expensive to maintain once people start preparing for you.
3. 03Reach a middlegame quickly with no memory loadThe right answer if you play mostly rapid and blitz, or if you are returning to chess after years away.

If you cannot say which of the three you are chasing, you will end up collecting lines the way people collect browser tabs. Pick one, write it down, and use it as the tie breaker every time two lines look equally good.

02How much theory do you actually need

Far less than the internet implies, and much more than most club players carry. The useful measure is not "how many moves deep" but "how many moves past the point where the game usually leaves the book". If your opponents deviate on move seven, knowing move eighteen is worth nothing and knowing the plan on move eight is worth everything.

As a rough calibration, this is where the real work sits at each level:

The trap at every level is the same. Depth feels like progress because it is measurable, and understanding does not. Adding another eight moves to a line you already play is a comfortable way to spend an evening without getting better at chess.

03Choose lines by the positions they produce

Opening choice is usually presented as a question of evaluation. It is really a question of taste, and taste is a legitimate criterion because you are the one who has to play the resulting position for the next three hours.

Ask three questions of any candidate line, in this order:

• What does the pawn structure look like on move fifteen? Not the evaluation. The shape. Closed centre or open, pawn majority on which wing, which side has the break.
• Where do the pieces want to go? If you cannot name the good square for each minor piece without thinking, you do not understand the line yet.
• What happens when they play the natural move rather than the best one? This is the position you will actually get, and it is usually missing from the book.

Take the Carlsbad structure, which arises from the Queen’s Gambit Declined after an early exchange on d5. It appears in dozens of move orders, and if you understand it once, you understand all of them.

A player who knows this structure has an opinion on move five and a plan on move fifteen. A player who memorised the same five moves from a video has a plan until the video ends. Both spent an evening. Only one of them can still use it a year from now.

Learn one structure properly and you have learned a hundred move orders by accident.

04Build it from your own games, not from a table of contents

This is the part that most guides skip, and it is the part that makes the difference. Your repertoire should be assembled in response to evidence about what actually happens to you at the board.

The method is unglamorous. Take your last fifty or a hundred games, sort them by the opening you reached, and count. Not wins and losses, which are noisy at this sample size, but two other things: where the game left your preparation, and which positions you dislike having on the board.

1. 01Export your gamesPull the PGN out of your online accounts or your tournament records. In CipherChess you can import straight from Lichess and Chess.com, or drop a PGN file onto the board.
2. 02Find the exit moveFor each game, mark the move where you stopped knowing and started guessing. That number is the honest depth of your current repertoire.
3. 03Cluster the exitsYou will find they are not scattered. Three or four positions will account for most of them. Those positions are your repertoire’s real to do list.
4. 04Check what the field plays thereLook up each exit position in a database and see what the common replies are, weighted by the rating band you actually play in.
5. 05Write one line and one plan per exitNot a tree. One main answer, plus a sentence about what you are trying to do. You can branch later, once you have played it a few times.

This inverts the usual order. Instead of learning theory and hoping to reach it, you find the positions you are already reaching and equip yourself for them. The first pass usually takes two hours and improves your results more than the previous six months of watching opening videos.

05The shape of a working repertoire

A complete repertoire has four compartments, and they are not equally expensive to fill.

One first move as White

One. Not two. The cost of a second first move is not double, it is roughly triple, because the branching multiplies rather than adds. Choose the move whose typical middlegames you like and stay there for at least a year. White scores somewhere around 54 to 55 percent across large databases regardless of which reasonable first move is played, which tells you that the choice matters far less than what you do with it.

One answer to 1.e4 and one to 1.d4

This is where most of your study time goes, and where most repertoires are weakest, because Black has to be ready for a wider range of setups. A good economy here is to choose defences that share a structure or a piece setup, so that the ideas transfer between them.

A plan for the sidelines

The English, the Bird, the Réti, the various move order tricks and the openings named after people who died in the 1800s. You will meet these more often than you expect, and almost nobody prepares for them. A single sensible setup you can play against all of them, learned once, quietly wins games for years.

The anti systems aimed at you

Once you have a defence, learn the two or three lines specifically designed to avoid your main setup. These are the lines your opponents will actually pick when they know what you play, and they are the ones most often missing when a repertoire collapses.

06Write the plan, not just the moves

A repertoire file that contains only moves is a file you will have to relearn every time you open it. Attach three things to every line you keep:

1. 01The idea in one sentence. "I am trading the light squared bishops so his queenside pawns have no defender." If you cannot write the sentence, you have not finished studying the line.
2. 02The move you are afraid of. Every line has one. Naming it is how you stop being surprised by it.
3. 03The endgame it tends to reach. Openings do not end at move twelve. Knowing that your favourite line usually produces a rook endgame with a queenside majority is worth more than the last four moves of theory.

These notes are what make a repertoire recoverable. Come back after three months away and the moves will have faded. The sentences will not.

07Train it so it survives contact

Reading a repertoire and knowing a repertoire are different states, and the gap between them is only closed by retrieval. You have to be asked the position and produce the move, under mild pressure, repeatedly, with gaps between the repetitions.

That is not a chess insight, it is how memory works. The practical version:

• Train from the position, not the move list. Being shown a diagram and asked for the move is the thing you will have to do at the board. Reading a line top to bottom is not.
• Space the repetitions. A line you got right today should come back in three days, then a week, then a month. A line you got wrong should come back tomorrow.
• Train both colours. Play the opponent’s side occasionally. You will find out immediately which of your lines you only half understand.
• Stop when you are guessing. A session where you are hunting for the move you half remember is teaching you the wrong retrieval habit. Look it up, note why, move on.

Twelve focused minutes a few times a week will hold a repertoire in place. Two hour marathon sessions will not, which is a relief, because nobody has two hours.

08Maintain it, because it is a living document

The repertoire is never finished, but maintenance is cheap if you do it continuously and brutal if you do not. The habit is one pass after each tournament or each batch of twenty online games:

1. 01Where did I leave the book this time?New exit positions get added. This is the only source of genuinely new work.
2. 02Which line did I dread?If you were unhappy to see a position, either learn it properly or change the move order that allows it. Dread is information.
3. 03What is now dead weight?Lines nobody has played against you in a year can be demoted to a note. Delete without guilt.

Two thirds of the maintenance is deletion. A repertoire that only grows will eventually be too large to hold, and the first thing to fall out of an overloaded repertoire is always the thing you needed most.

09A four week plan to get from nothing to something

If you are starting from scratch, or restarting, this order works and does not require you to buy anything.

1. 01Week one · EvidenceGather fifty of your games. Mark every exit move. Cluster them. You now have a list of five or six positions that matter more than everything else combined.
2. 02Week two · DecisionsChoose your first move as White and your two defences as Black, using structure and taste rather than evaluation. Write one sentence explaining each choice.
3. 03Week three · Fill the holesWork through your exit positions one at a time. Database first, engine second, and for each one write the move, the idea, and the move you are afraid of.
4. 04Week four · RetrievalStop reading. Train from positions, spaced out, ten minutes at a time, both colours. Play games and let the failures tell you what to add next month.

At the end of four weeks the repertoire will still be full of gaps. That is the correct state. A repertoire with no gaps is one that has never been tested against anybody.

Once the repertoire exists, the next question is how much to trust the numbers you used to build it, because a database will happily tell you that a losing move scores 60 percent. That is the subject of how to read a chess database. And when your lines meet a specific opponent rather than the field in general, opponent preparation is a different exercise again.