ACWR (Acute:Chronic Workload Ratio)
Also known as: Workload Ratio, Acute-to-Chronic Ratio, Spike Ratio
The ratio of acute training load (typically the last 7 days) to chronic training load (typically the last 28 days). Originally proposed as a marker of injury risk: ratios above ~1.5 were claimed to indicate dangerous training spikes.
Formula
ACWR = acute load (7-day rolling sum or EWMA) / chronic load (28-day rolling sum or EWMA)
Typical interpretations (now contested):
- ACWR 0.8-1.3: "sweet spot"
- ACWR > 1.5: "danger zone" (claimed)Example
Acute load (last 7 days): 2,400 AU. Chronic load (last 28 days, weekly average): 1,800 AU. ACWR = 2,400 / 1,800 = 1.33. Under the original framework, this would be "safe"; if the ratio jumped to 1.6 the next week, the original framework would flag it as injury risk.
How Afitpilot Uses This
Afitpilot deliberately does NOT surface ACWR as a number, badge, or alert. We compute and display EWMA-smoothed acute (7-day) and chronic (28-day) load trends in the coach drawer for description only, with explicit "does not predict injury risk" disclaimer in the legend. See the EWMA Load Trend term for the full reasoning. We treat ACWR as a metric to understand and explain, not one to act on.
ACWR — the rise and fall
| Who / Context | Value | Note |
|---|---|---|
| Gabbett (2016) | ACWR > 1.5 = injury risk | Original paper; rugby league cohort; highly cited and operationalised everywhere |
| Banister & Calvert (1980s) | Acute/chronic conceptual lineage | Training-impulse models predating ACWR by 30+ years |
| Impellizzeri (2020) | ACWR critique published in BJSM | Methodological flaws, mathematical coupling, no individual predictive value |
| Impellizzeri (2023) | Stronger statistical case for the same conclusion | Re-examines original cohorts with corrected methodology |
| Modern practice (2024+) | Track load, don't badge ratios | Consensus shifting toward EWMA trends as descriptive, not predictive |
Known Limitations
- •The strong claim that ACWR > 1.5 predicts individual injury risk has been comprehensively critiqued (Impellizzeri 2020, 2023) — the predictive signal in the original Gabbett papers largely disappears once methodology is corrected for mathematical coupling and confounders.
- •Population-level statistics (group averages) cannot be applied to individuals without large error. Even if 100 athletes with ACWR > 1.5 have slightly elevated group-mean injury rates, any given athlete cannot predict their own risk from their ratio.
- •The acute and chronic windows share data points (a session counted in the last 7 days is also in the last 28), which creates mathematical coupling that artificially correlates the ratio with injury without causal mechanism.
- •ACWR ignores absolute load. A 1.5 ratio from 1,000 → 1,500 AU is very different from 4,000 → 6,000 AU — the second represents a much larger absolute training stress.
Science Context
ACWR had a dramatic decade-long arc in sport science: introduced by Gabbett (2016) on rugby league data, rapidly adopted by elite team sports, then thoroughly critiqued by Impellizzeri and colleagues (2020, 2023). The critique is now widely accepted at the research level but persists at the practitioner level because the original framing was clean and actionable. Afitpilot's stance — compute the smoothed acute and chronic loads for descriptive context, never surface the ratio as a risk signal — follows current research consensus. This isn't a refusal to monitor load; it's a refusal to monitor it through a metric that doesn't carry the predictive weight it was sold with.