# Readiness: Factors, Baselines and Coverage

Canonical HTML: https://effortsignal.com/learn/readiness/

Last content update: 2026-09-10

Readiness combines baseline-relative HRV, resting heart rate, sleep and workload. It is a model with explicit missing-data gates.

## Aggregate score

Configured weights are HRV 0.35, resting HR 0.15, sleep 0.30 and workload 0.20. Each available component has a 0–100 score s. Available weights are renormalized:

```text
Readiness = round_half_up(exp(sum(w × ln(s)) / sum(w)))
```

The displayed score needs available configured weight of at least 0.65 and an available physiological component (HRV or resting HR). Missing components are not replaced with neutral values. The geometric mean makes a low factor matter more than in an arithmetic average.

## HRV and resting heart rate

Use compatible measurements from the preceding 42 local days; the current day is excluded from that baseline. At least 14 observations are required, with 28 for a reliable baseline. Metric, provider account, device and measurement method define compatibility. HRV is transformed with natural logarithms; resting HR remains in bpm. Baseline center is the median; scale is max(1.4826 × median absolute deviation, floor), with floors 0.05 for log HRV and 1.5 bpm for resting HR.

The deviation is 70% today's z-score plus 30% the seven-day median trend z-score when both exist. A trend needs three observations. Today-only or trend-only evidence can still be used with a quality flag. HRV scores interpolate through (z, score): (−2.5,10), (−2,28), (−1.5,48), (−1,68), (−0.5,82), (0,90), (0.5,95). Resting-HR mapping reverses the direction: (−0.5,95), (0,90), (0.5,82), (1,68), (1.5,48), (2,28), (2.5,10). Values beyond endpoints are clamped.

## Sleep

Main sleep plus confirmed naps is compared with an explicit sleep target. Otherwise, with at least 28 compatible nights in 60 days, the target is the 75th percentile clamped to 7–9 hours; without enough history the default is 8 hours. The sleep state is 70% today's target ratio plus 30% a seven-day weighted ratio with a three-day half-life. Missing nights are omitted, not filled with zero sleep.

The sleep ratio/score points are 0.50/10, 0.60/20, 0.70/38, 0.80/58, 0.85/70, 0.90/80, 0.95/88, 1.00/95 and 1.05/100, with linear interpolation. A missing completed main sleep makes that component unavailable.

<h2 id="workload">Workload uses its own load model</h2>

The component requires 42 complete dated load rows. It compares the sum of the seven complete days ending yesterday with the mean totals of the five non-overlapping weeks before those seven days. This is not ATL/CTL. Where available, activity PSS is used. HR-only activity TRIMP is calibrated into that scale using the median PSS/TRIMP ratio from at least three overlapping activities. Mixed power and HR-only history without enough calibration makes the component unavailable. With only HR evidence, the component uses TRIMP. Missing activity load is flagged; zero previous-week average prevents the ratio.

Workload ratio/score points are 0.75/100, 0.90/95, 1.00/90, 1.10/82, 1.25/68, 1.40/50, 1.60/32, 1.80/18 and 2.00/10. This separate policy does not change the TRIMP-only [fitness/fatigue series](/learn/training-load/).

## Verified examples and limits

With all four component scores equal to 90, Readiness is 90. HRV alone (weight 0.35) cannot produce a displayed score. Sleep plus workload has weight 0.50 and no physiological component, so it also cannot produce one. Stable sleep at exactly the target maps to a sleep component score of 95, not 100.

A score can be provisional while baselines mature. High confidence requires all components, sufficiently mature physiological baselines and no quality flags. The live Dashboard may show the most recent persisted result within its 14-day history when today has none; the date identifies that older result. A historical selection stays on the requested date. Changing sources or settings can change later calculations. This is not a diagnosis or a promise of training performance. [Recovery and Readiness](/features/recovery-readiness/) shows the factors and their dates.

## Sources and implementation scope

[Walsh and colleagues (2021), athlete sleep consensus](https://pubmed.ncbi.nlm.nih.gov/33144349/) provides background, not a specification of every EffortSignal rule. The formulas, windows, quality gates and examples above describe EffortSignal 0.1.13, checked September 10, 2026. Product-specific thresholds and rounding should not be attributed to the background source.