Nothing yet. The Coach opens its first experiment once the muse has run and the mutation cooldown has elapsed.
Open experiments
muse.prompt: v1 challenging v0 P(challenger better)0.845 after 20 paired run(s) - challenger 94/100, incumbent 90/100, 6 voided needs 0 more run(s), 0 more candidate(s), and P ≥ 0.9 to promote
Gauges
North star:calibration.brier with its sample count - the one number improvement is for, because a lucky week cannot move it and a well-calibrated view can. Guardrails (must not degrade, both directions): sizing.refused_rate and exit.uncorroborated_decisives (rule compliance - a seam losing the conditional payoff, or a stop firing on quote noise), the trade/decline balance behind attribution.attributable_rate (always-trading and never-trading are both drift, and this book has measured both), and coach.cost_usd_today with model.cal_age_days (cost and staleness). Everything else here is diagnostic.
attribution.attributable_rate
0.5
min 0 / max 0.5 / n=18
book.beta_delta_dollars
299,264.23
min -516,369.42 / max 601,188.96 / n=59
book.vega_dollars
251.57
min -14.53 / max 253.04 / n=59
calibration.brier
0.2332
min 0.1712 / max 0.2332 / n=90
calibration.n
71
min 27 / max 102 / n=160
calibration.n_eff
9.68
min 9.68 / max 11.94 / n=90
calibration.resolved
95
min 3 / max 95 / n=160
coach.cost_usd_today
0
min 0 / max 2.033 / n=160
coach.open_experiments
1
min 0 / max 1 / n=160
coach.promotions_total
0
min 0 / max 0 / n=160
discovery.days_since_run
0.2
min 0 / max 0.67 / n=19
discovery.gauntlet_survival
0.05
min 0 / max 0.4286 / n=142
exit.uncorroborated_decisives
0
min 0 / max 0 / n=116
gates.admitted_hold_rate
0.4225
min 0.4225 / max 0.4225 / n=26
gates.no_chain.regret
0.4
min 0.4 / max 0.4 / n=26
gates.no_chain.resolved
5
min 5 / max 5 / n=26
model.cal_age_days
400
min 0 / max 400 / n=152
model.inflation_5d
1.05
min 1.05 / max 1.3 / n=152
muse.candidates_per_run
5
min 3 / max 5 / n=155
muse.funnel_overlap_rate
0.3
min 0.3 / max 0.38 / n=110
muse.runs_total
28
min 10 / max 28 / n=155
muse.seed_entropy
29
min 12 / max 29 / n=155
muse.survival_rate
0.92
min 0.4333 / max 0.94 / n=155
research.days_since_run
0.96
min 0.37 / max 0.96 / n=19
research.opportunities_per_run
0
min 0 / max 2 / n=142
sizing.refused_rate
0
min 0 / max 0 / n=46
Coach actions on the same series: experiment_opened v0 vs v1 (2026-08-29)
Levers and how to steer
lever
incumbent
since
state
muse.prompt
v07809d229
seed
running
playbook.catalogue
v05a8b5667
seed
running
Variants live in data/state/levers/*.json. Set "paused": true to stop experimenting on a lever - it closes any open experiment and freezes the incumbent, which is how you hold behaviour still for a demo. Editing the incumbent text by hand is supported; the fingerprint is recomputed from the text on load. Nothing here can reach code, gate thresholds, sizing, or the constitution.