⚠ [ Diagnosis only — no engine coaching, no rating guarantee, no wagering advice. ] ⚠

Most chess advice is generic because most coaches never see your games. This one starts from your games and nothing else.

What it measures

You send your Lichess username. Clock Leak reads your public games and measures exactly how you spend your clock — not what moves you played, when you played them.

• Losses on time, positioned against real peers at your level
• Clock left when the game ends
• Where your time actually goes: opening (moves 1-10), middlegame (11-25), late game (26+)
• One instruction for your next game, numeric, traced to the one place your clock actually leaks

The measured findings

Built on 93 players and over 11,180 games, cross-checked against an independent 27-player panel rated 2300+ (median 2782).

The myth this kills

"Play faster in the opening" is the most common advice given to players who lose on time. Measured correlation between opening time and time losses: r = +0.11 — noise. It does not predict anything.

What actually predicts it

Time spent in the middlegame, moves 11 to 25: r = +0.41. This is the one lever with real statistical support. Overspending here, not slow openings, is what burns a clock down before the game is decided.

The rate that never improves on its own

Getting stronger does not fix this by itself. The spread within any one band is enormous — from 4% to over 60% of games lost on time among players of identical rating. That gap is what the diagnosis measures.

What the 2300+ panel does differently

They spend 13.4% of their clock in the opening against a 20.5% median in the main panel — and 34.0% late in the game against 23.4%. A real, replicated pattern across 27 independent accounts, not one player's style.

What you get

CLOCK LEAK - yourname, 120 games, rating ~1450

<the clearest finding, first>

WHERE YOU SIT
Losses on time      31% | peers: Q1 9% . median 16% . Q3 27%  -> highest quarter
Clock left at end    24% | peers: ...
Seconds per move     3.8 | peers: ...

WHERE YOUR TIME GOES
Opening (1-10)       19% | peers: 19% | elite: 13%   [not a cause of time loss]
Middlegame (11-25)   61% | peers: 54% | elite: 52%   [this is the lever]
Late game (26+)      20% | peers: 23% | elite: 34%

THE ONE THING TO DO NEXT GAME
By move 25, aim to have at least 40% of your clock left. Check it once, at
move 25, nowhere else.

WHAT IT DOES NOT SAY
<named explicitly, every time>

One page. Read it in two minutes.

Why not just ask ChatGPT

Because it cannot read your Lichess history, cannot compute where 93 real peers and 27 elite players actually sit, and cannot tell a supported finding (middlegame) from a popular myth (opening speed) without the measurement behind it. Clock Leak is a calculation on your real games, not a conversation about chess in general.

What it will not do

It will not evaluate a position or tell you which move was best — that is engine coaching, a different product. It will not promise a rating gain: the measurement is validated, the improvement from acting on it is not, and claiming otherwise would be a promise this product cannot keep. It gives no betting or wagering advice on chess, in any framing.

Who this is for

Players who lose games on the clock and want to know exactly where the time goes, not another list of opening principles.

Who this is not for

Players looking for move-by-move engine review, opening theory, or a guarantee of rating improvement.

FAQ

Do I need to have analysed my games with an engine first?
No. Most players never do — across this panel, only about 10% of games carry engine evaluation. Clock Leak is built entirely on the clock data every game already has.

Does it work for beginners and for strong players?
The reference panel covers 850 to 2050. Outside that range, you still get your raw numbers, but no peer comparison — the agent says so rather than guessing.

Will this get me more rating points?
No promise is made on that. What is measured and shown to you is exactly where your clock goes and how that compares to real players. What you do with it is yours.

Can it review a specific game or position?
No — it works across many games to find a pattern, not inside one game. For move analysis, use an engine tool alongside this one.

Data: Lichess, CC0. No attribution required by the license; given here voluntarily. Diagnosis only, not chess coaching. No rating outcome is guaranteed.