Trading Journal & Coach

The newest genuinely-useful category of AI-for-trading is not predicting the market — it is analysing your own behaviour. The Trading Journal turns your history of runs and backtests into honest feedback so you can improve the strategy you already have.

Open AI → Trading Journal (/journal).

What it surfaces

From your instances (runs and backtests) it computes, deterministically:

• Win / loss / failure counts and win rate across your backtests;
• Behavioural insights — the leaks that quietly cost retail traders:
  • Over-concentration — most of your activity is in one symbol;
  • Repeated failures — a high share of runs failed to build or configure;
  • Losing bias — more losing than winning backtests (with a nudge to run the Integrity Lab and check the edge is real);
  • a clean bill of health when none of the above applies.

GET /api/journal

Why it is reliable

The behavioural analysis is pure, deterministic domain code (Core.Journal) with no infrastructure dependency — unit-tested for over-concentration, repeated failures, losing bias, the balanced case and the empty account. The facts come first; the AI coach (Portfolio Digest) is an optional narrative layer on top, gated on the Anthropic API key, so the journal works fully without AI configured.