Research library

2026-07-29 to 2026-08-28 · 13 sessions · six instruments

None of the four opening shapes earned a playbook

Classify the first minutes of the session into one of four shapes, then run the shape’s playbook. Fifty-eight forward instances later, no shape clears a bar — and the best one is the shape nobody proposed.

VerdictREFUSED · no shape promoted

In simple words

What happened

The two shapes that were actually proposed for promotion are among the weaker cells (29% and 25%). The best cell is the quiet inside-range shape at 50% with a much shallower typical drawdown — and nobody had proposed it, which is the strongest hint that the four-way split was reasoning ahead of the evidence.

Who would pay?

Openings are not all alike. A washout that never recovers, a push through yesterday’s high that holds, a quiet inside range — each is a different market, and each was given its own play.

How it was tested

Each session’s opening was classified once, the shape’s named contract was recorded, and the outcome graded on real prices — best excursion, worst excursion, and whether it reached the playbook’s own definition of a hit. Repeats within a name, day and side were collapsed to one instance.

What would prove it wrong?

Every shape has a lower bound at or below 25%, and three of the four have a typical worst excursion near -99%: the named contract usually goes to nothing before or instead of paying.

Leader-collapsed evidence

What each setup actually did

58 independent shape-to-playbook instances, all resolved on real contract prices

Opening shapeInstancesHit rateLower boundMedian peakMedian worstDecision
FLATQuiet inside range (not a promotion candidate)12+86.8%-47.5%50%Lower bound 25%REFUSED
DOWNINLower inside range10+35.9%-88.3%30%Lower bound 11%REFUSED
PUMPPush through the prior high that holds24+46.2%-98.2%29%Promotion candidateREFUSED
FLUSHWashout below the prior low, no recovery12+22.2%-99.2%25%Promotion candidateREFUSED

What we keep

Learnings

  1. 01

    Write down which cell you expect to win before you look. Here the two promotion candidates finished near the bottom and an unproposed cell finished top, which is only visible because the candidates were named in advance.

  2. 02

    Report the worst excursion next to the best one. A shape with a +46% typical peak and a -98% typical trough is not a playbook, it is a description of a contract that mostly dies.

  3. 03

    Reaching the required sample size is not the same as passing. The bar of 25 forward instances was cleared at 58; the answer is still no.

  4. 04

    The strongest-looking cell here is the quiet one, on twelve instances. That is a hypothesis for a future test, not a finding.