Skip to content
SeedProject · active

Building a chess openings trainer

I'm often improvising by move ten, so I'm exploring a trainer that helps me practise opening decisions.

planted June 26, 2026 · last tended November 9, 2026

A captured spark. Unverified, unpolished, possibly wrong.


I started playing chess after watching The Queen’s Gambit, along with what felt like everyone else. The initial excitement stayed with me, and I still love the game, although the way I spend my time playing it doesn’t always help me improve.

I play a lot of bullet. The one-minute clock, premoves, and quick return to the next game make it very easy to keep going. A longer game asks for more patience, while bullet gives me another decision almost immediately. That pattern is familiar from my attempts to manage attention while studying.

The result is that I often move by feel. I recognise something about the position, choose a move that looks plausible, and deal with the consequences when they arrive. That has got me to around 1700 rapid and 1500 blitz, but I’ve been feeling stuck.

Openings are an obvious place where it shows. By move ten, I’m often spending time figuring out a position my opponent already knows. Improvising is part of chess, of course, but I’d like to understand more of those early decisions before the clock starts.

What I want to practise

I’m thinking about building a web app that lets me rehearse openings by making moves on a board. There are already plenty of databases to browse, but I find it difficult to stay interested when I’m only looking through lines.

The part I want to borrow from my Anki deck is having to choose a response. I’d see a position, make a move, and get feedback about why it fits or what I overlooked. Spaced repetition could help bring back positions I haven’t practised for a while.

That raises a few design questions:

  • How should a repertoire be organised? I could start with one opening for White and a defence against each of e4 and d4. The harder job is choosing lines I understand well enough to practise, rather than collecting a huge tree of moves.
  • How do I check understanding? Recognising the next move in a familiar sequence might be easier than finding it in a game. I’d like to practise from different positions and explain the plan behind the move.
  • What happens when the opponent leaves the main line? This is the part I’m especially interested in. I can know the expected reply and still have no idea why an unexpected move is weak, so the trainer should include those deviations.

I want to be careful about what I’m practising. Memorising a move without understanding the position could leave me just as confused a few turns later. That’s related to checking whether I’ve understood what I’m studying, rather than mistaking familiarity for understanding.

What I haven’t decided

I’m still working out the stack. A local-first app would suit me, but move trees, engine evaluation, and importing games from Lichess or Chess.com would each add a different amount of work.

There’s also the repertoire itself. I could write my own, which would take longer but force me to think through the choices, or start from an existing repertoire and work out which parts suit the positions I actually reach.

And then there’s the question I keep coming back to: would this help more than spending the same time on tactics or reviewing my games? Building a trainer is appealing to me, which makes it worth checking whether I’m using it to avoid a less exciting kind of practice.

For now, this is a project idea with a problem I recognise from my own games. I’d like to try a small version and write down what happens, rather than assume that finishing an app will automatically improve my chess.

Paths that lead here

Where this note points

More from these beds