The Training Log 其不善者而改之
The artefact that compounds. After a few months, the pattern in your own log will tell you more than a coach's first session could — because it is drawn from your games, not from a stranger's impression of them. Everything here works in the page; use Export to keep a copy, since this site deliberately stores nothing in your browser.
This week
Clean-game streak
Ten consecutive games at 15|10 or slower without losing a piece to a one-move tactic. One slip resets it to zero — that is the point. This is the exit criterion for Rung 1.
Game log
Failed-puzzle queue
Add anything you get wrong. Revisit at ~7 days → ~3 weeks → ~2 months — Cepeda's ridge puts the optimal gap at roughly 20% of your target retention horizon. Re-solve cold from the diagram; never reread the solution (Principle Ⅰ).
The reference card
Before every move
- What did their move just do — including what it stopped defending?
- Is my move safe? Assume not. Try to prove yourself wrong.
- After I play it: all their checks, captures and threats. Say the list.
After every game (worked through on Analyse Your Games)
- Three sentences from memory — engine off.
- Find the turning point yourself. Commit to a better move.
- Engine on. Look only where you disagree.
- Log one 善 and one 不善, the 不善 as a rule.
Every week
Tactics on most days, mixed themes · long games at 15|10 or slower · every game reviewed · no opening study you did not earn by losing.
How to read your own log after two months
Look for the single most frequent error class and work only on that — the "your leak" panel above computes it for you, and the six Train pages are where each class is treated. Then look at your confidence column: if your losses cluster at 4 and 5, your problem is calibration rather than knowledge, and the fix is different. And look at the rules you wrote. If the same rule appears three times, you have not been applying it; make it part of the safety check rather than writing it a fourth time.