48%: the anatomy of an alert rule that cried wolf
On a simulated season a common rule lit up one day in two. The first fix brought it down to 8%, but for the wrong reason
MarkerDrift team3 November 2026intermediate
MarkerDrift Journal · method, data, sources
Method articles for people who monitor athletes: definitions, formulas, worked examples and checked sources. Every number can be reproduced, and the dossier data can be downloaded.
Outside his usualHigher than his last 8 MD-3 days. Worth a look, not a verdict.
Featured
A dossier measures a real question on the demo club’s data, with the calculation to download.
On a simulated season a common rule lit up one day in two. The first fix brought it down to 8%, but for the wrong reason
MarkerDrift team3 November 2026intermediate
The library
Filter by type or by topic. A new one every 10 days.
On a simulated season a common rule lit up one day in two. The first fix brought it down to 8%, but for the wrong reason
13 min read3 November 2026intermediate
Why a number should be read against the athlete themselves, and what the squad average is really for
7 min read26 October 2026basic
Basics
The words every article uses. Tap an entry for the formula.
z = (x − mean) ÷ SDsample SD, with n − 1
TE = SD of the differences between two trials ÷ √2
CV = SD ÷ mean × 100
sRPE = duration (min) × RPE (CR-10)in arbitrary units (AU)
ACWR = acute (7 days) ÷ chronic (28 days)uncoupled: the chronic load leaves out the acute days
λ = 2 ÷ (N + 1)today = λ × load + (1 − λ) × yesterday
monotony = mean ÷ SD of the 7 days (rest = 0)strain = weekly load × monotony
compare the same MD and the same kind of daynever “Tuesday against Tuesday”
Calendar
A new article every 10 days: we announce it on the MarkerDrift LinkedIn page and in the RSS feed.