---

Tick-by-Tick Trading Decisions

Full transparency: every trading decision with the LLM reasoning behind it. Browse 202+ ticks across 7 backtest runs. Compare how Treatment, Control, Placebo, and Control-WS arms responded to the same market conditions.

Runs Available

• Run 1: Where It Started Working -- 91 ticks, Sep 1 - Nov 30, 2025, Opus 4.6
• Run 2: Model-Agnostic Confirmation -- 83 ticks, Sep 1 - Nov 22, 2025, Sonnet 4.5
• Run 3: Briefing Restructuring -- 181 ticks, Sep 1, 2025 - Feb 28, 2026, Opus 4.6
• Run 4: Architecture Breakthrough -- 185 ticks, Sep 1, 2025 - Mar 4, 2026, Opus 4.6
• Run 5: Confirmation Run -- 187 ticks, Sep 1, 2025 - Mar 6, 2026, Opus 4.6
• Run 6: First 4-Arm RCT -- 187 ticks, Sep 1, 2025 - Mar 6, 2026, Opus 4.6
• Run 7: Strongest Result -- 202 ticks, Sep 1, 2025 - Mar 21, 2026, Opus 4.6

Each tick shows: portfolio state before decision, BTC price at tick time, the agent decision (long/short/hold/close), position size, and the full reasoning trace explaining why.

Read the full research findings | Back to evidence | How we test

---