PrimerForthcoming
Correlation isn't a coaching decision
Stack enough signals and something always lines up. How to tell a real effect from a coincidence wearing a trend line.
Est. 6 minJun 2026
Field Notes
Notes from the engine.
Working notes on causal inference, honest measurement, and building coaching agents that don't bluff. Written the way we'd explain it to a coach — exact, and no longer than it needs to be.
Latest note
The same load that built them last week is breaking them this week.
An average effect hides the one thing a coach most needs: the condition under which it flips its sign. A field guide to effect modifiers — and the threshold where more stops helping.
Read the note →The queueDrafting in order
MethodForthcoming
Why “insufficient” is an answer, not an error
Most tools render a verdict on three data points as confidently as on three hundred. The case for withholding one.
Est. 5 minJun 2026
AI coachingForthcoming
Giving an agent permission: reading the license ladder
DECLINE through ACT. What each rung authorises, and why the estimate — not the prompt — should set the ceiling.
Est. 7 minJul 2026
PrimerForthcoming
On average, about no one: the case for within-athlete
Population studies describe a cohort your athlete may not resemble. Why the answer has to come from their own history.
Est. 6 minJul 2026
PrimerForthcoming
How to read a confidence interval like a coach
A wide interval is a quiet way of saying “not sure yet.” Turning a range into a training decision — or a pass.
Est. 5 minJul 2026
MethodForthcoming
What survives refutation checks
How placebo shifts, uncertainty, and evidence gates separate a finding from a flattering fit.
Est. 6 minAug 2026
One note when one ships. Nothing in between.
No digest, no drip. We email when a field note is published — about once a month — and never about anything else.