The formula in one line
ACWR = 7-day acute training load ÷ 28-day chronic training load.
Acute load is the average daily training load for the most recent week (the load you're absorbing right now). Chronic load is the average over the trailing four weeks (the load you've adapted to). The ratio tells you whether recent training is sitting close to, below, or above what the athlete has been preparing for.
The two methods: rolling average vs EWMA
There are two common ways to compute the acute and chronic windows.
Rolling average
Take a flat mean of the last 7 days for acute and the last 28 days for chronic. Every day in the window contributes equally. This is the simplest formulation, the easiest to explain to athletes, and what most public calculators (and Performance House) use.
Exponentially weighted moving average (EWMA)
EWMA weights recent days more heavily than older ones, so a hard session yesterday pushes the acute number up more than a hard session four days ago. Williams et al. (2017) argued EWMA gives a faster, more accurate injury-risk signal. The difference in most real datasets is small, and the rolling-average version is much easier to teach and audit.
For the worked example and the calculator below, we'll use the simple rolling average.
A worked example
Imagine you've been logging session-RPE (1–10) for an athlete every day for the last four weeks. The values look like this:
Computing the two windows:
- Acute (last 7 days, week 4): sum 56 ÷ 7 = 8.00.
- Chronic (last 28 days, weeks 1–4): sum 168 ÷ 28 = 6.00.
- ACWR = 8.00 ÷ 6.00 = 1.33.
That sits above the 1.3 threshold. Week 4 was a hard week relative to the four-week baseline. Drop those numbers into the calculator below to verify — and try other combinations to get a feel for how the ratio moves.
ACWR calculator
What ACWR doesn't tell you
ACWR has become a default load-monitoring metric — and the research base behind it is messier than the popularity suggests. A few honest caveats:
- The "sweet spot" finding has been challenged. Impellizzeri et al. (2020) pointed out that because acute and chronic windows overlap, the ratio has built-in mathematical artefacts that can create spurious U-shaped associations with injury. The 0.8–1.3 band is a useful conversation starter; it isn't a clean decision rule.
- Garbage in, garbage out. The ratio is only as good as the underlying load data. If athletes report sRPE inconsistently, or if you mix matches and training sessions inconsistently, the chronic baseline drifts and the ratio means less.
- Context matters more than the number. A 1.6 ACWR after a planned overreach week is not the same thing as a 1.6 after a quiet block. The ratio doesn't carry the context — the coach does.
- Workload is not the whole story. Recovery, sleep, life stress, and individual injury history matter at least as much as the ratio. ACWR sits alongside wellness check-ins, not in place of them.
How Performance House computes ACWR automatically
Inside the platform, every athlete's daily sRPE is logged through the wellness check-in. The 7:28 rolling ratio is computed and plotted automatically using the same computeAcwrSeries() function that powers the calculator above — so the marketing widget, the team dashboard, and the per-athlete chart can never disagree. The chart shades the 0.8–1.3 zone in green and flags when an athlete drifts outside.
- Per-athlete and squad-wide views
- Threshold-based alerts when an athlete moves into the danger zone for two days in a row
- Side-by-side with wellness (sleep, soreness, mood) so the ratio is read in context
- Free on the Starter plan for up to 5 athletes
If you want it set up for your squad without writing the formula into a spreadsheet every week, Performance House handles the daily logging, the rolling math, and the visualization. The same computeAcwrSeries() function powers both the calculator above and the in-app chart.
Frequently asked questions
Should I use rolling average or EWMA for ACWR?
Rolling average is simpler and more commonly used. EWMA (exponentially weighted moving average) weights recent days more heavily and reacts faster to load changes. Williams et al. (2017) argued EWMA gives a more accurate injury-risk signal, but the difference in practice is small. Performance House and most public calculators use the 7:28 rolling average for simplicity and transparency.
Is the 0.8–1.3 sweet spot still valid?
It is widely cited, but the original Gabbett threshold has been challenged. Impellizzeri et al. (2020) showed that the original "sweet spot" finding suffers from spurious correlations introduced by ratio-of-overlapping-windows math. Treat 0.8–1.3 as a useful flag for a conversation, not a hard threshold for prescribing or pulling an athlete.
How many days of data do I need before ACWR is meaningful?
At least 28 days for the chronic window to be properly populated. Most implementations (including Performance House) will compute a value once 7 days of acute data exist, but the chronic baseline isn't stable until you have a full 4 weeks of consistent reporting.
What unit should the daily load be?
Anything consistent across days. Session-RPE × duration (load = RPE 1–10 × minutes) is the most common subjective option. GPS-derived total distance or PlayerLoad works equally well for teams with wearables. The ratio is dimensionless, so as long as both windows use the same metric you can compare.
Can I use ACWR for non-team sports?
Yes. The math doesn't care about sport. Endurance athletes often use TRIMP (training impulse) instead of sRPE; strength athletes use total tonnage or volume load. The same 7:28 ratio applies to any time-series of daily training load.