EloLeak

The member desk for chess improvers

Find where your Elo leaks.

Stockfish reads your last 150 rated games and hands you a coach's verdict: the openings that bleed points, the phase where you blunder, the wins you let slip. Then it tells you exactly what to train, and scouts your next opponent.

Become a member · $9.99/month Log in

Billing via the newsletter. Cancel anytime. No password, no install.

Diagnosis · blitz_addict_1400 member desk
Endgame blunders
4.1
per 100 moves
Conversion
69%
strong players: ~90%
Openings
31%
Caro-Kann as Black
1 Endgames: 3x your middlegame blunder rate -38 pts
2 61 winning positions, only 42 wins -17 pts
3 42% of blunders under 30 seconds -9 pts

What members get

A full improvement desk, not another puzzle app

Engine diagnosis

Stockfish grades every move of your last 150 rated games, on the same blunder scale Lichess uses. Five leaks, each priced in points lost, worst first.

Coach's plan

Every leak comes with the fix: targeted drills, and the one book worth reading for your exact weakness. Diagnosis without a prescription is just bad news.

Opponent Scout

Paste your next opponent's username before the game. Their weak openings by color, their blunder phase, whether they crack in time trouble. 30 scans a month.

Weekly review

Every Monday, three lines: what you played, whether your number one leak is closing, one number that moved.

Progress re-scans

Re-run your diagnosis on fresh games and see the deltas. Your endgame blunder rate is a number now; watch it drop.

Strength estimator

Any username on any platform: playing strength estimated from results against rated opposition, never from the badge.

The diagnosis

Five leaks, measured on your real games

Opening repertoire

Which openings you actually score badly in, by color. "Your worst opening as Black: Caro-Kann, 31% over 16 games."

Blunders by phase

Opening, middlegame or endgame: where your games really break. "You blunder 3x more in endgames: 4.1 per 100 moves vs 1.3."

Time management

The blunders you play in time trouble you created yourself. "42% of your blunders happen under 30 seconds."

Converting winning positions

The wins you reached and then let slip. "61 winning positions, only 42 wins. Strong players convert ~90%."

Opponent strength

Whether you play down to weaker players and hand back points. "Against weaker players you score 63% when Elo expects 71%."

Ranked by cost

Every leak gets a price in points dropped, and your desk opens on the most expensive one. You always know what to fix first.

Before the game

Scout your opponent like titled players do

Preparation from public games is standard practice at the top. Your desk does it in two minutes: steer the game into the lines where your opponent already loses.

Scout · your_next_opponent ~2 min
As Black vs 1.e4 he scores 38% (13 games). Play 1.e4.
Blunders spike in the endgame. Keep pieces on.
Flags in 18% of his losses. Make him think.

Pricing

$9.99/month. Everything included.

Membership comes with the EloLeak newsletter. One subscription unlocks the whole desk: diagnosis, coach's plan, 30 Scout scans a month, weekly reviews, re-scans, the strength estimator.

Billing runs through the newsletter platform. Cancel anytime, in one click, no questions.

Become a member · $9.99/month

Fair questions

Before you ask

How do I get in after subscribing?

Come back here, hit Log in, and use the same email as your subscription. We email you a one-time sign-in link. No password to remember.

Why not just paste my games into ChatGPT?

Try it: language models hallucinate chess analysis with total confidence. Every number here comes from Stockfish running on our servers, over your full sample. Your puzzles are your own blunders with engine-verified solutions, your tilt profile is computed over your whole history, and the desk re-checks your account every week without being asked. A chat window does none of that.

Is this engine analysis or AI guessing?

Every evaluation comes from Stockfish, the strongest open chess engine. Prose explains the numbers; it never invents them.

Lichess Insights and Chess.com Insights already exist. Why pay?

They are chart explorers, and good ones. Four differences. Coverage: their engine stats only count the games someone bothered to analyze; your desk runs Stockfish on every game in the sample. A verdict, not thirty charts: Insights hands you filters and lets you play data analyst; your desk ranks five leaks by the points each one costs and opens on the most expensive. A prescription: every leak maps to drills and the one book worth reading. Your opponents too: no platform tool scouts the player you face tonight, or reads your Chess.com and Lichess games in one place.

Do you need my password?

No. Both platforms publish rated games through free, official public APIs. We only need the username.

Is scouting an opponent fair play?

Preparing against a known opponent from their public games is what titled players do before every round. Scout does it in two minutes.

Your next 100 rated games can pay for this.

One plugged leak is worth more than a month of puzzle grinding.

Become a member · $9.99/month