Does Training Load Predict Injury Risk?
The acute:chronic workload ratio was treated as settled science for half a decade. Then someone replaced the chronic load with random numbers and the effect survived.
By The Standard
In this articleObservation
The Standard take
Load monitoring is validated for measuring training dose. It is not validated for predicting injury. The acute:chronic workload ratio's association with injury survives when the chronic-load term is replaced with random numbers, which means the relationship was largely mathematical coupling, not biology. No "sweet spot" threshold has causal support.
Evidence rating: Emerging. This rating holds unless Prospective, pre-registered studies using genuinely uncoupled acute and chronic windows that still find a dose-response relationship.
Key Findings
Evidence: Emerging- For most of a decade, a single ratio ran the training week in professional sport.
- Does training load predict injury risk — and does the acute:chronic workload ratio measure anything real?
- Load monitoring is validated for measuring training dose.
What this does not prove: It does not prove training load is unrelated to injury, and it is not a licence to stop monitoring.
Observation
For most of a decade, a single ratio ran the training week in professional sport. Coaches held athletes out of sessions because a number had drifted above 1.5. The number came from a real paper, published in a real journal, and it is now one of the better documented statistical artefacts in sports science.
The Question
Does training load predict injury risk — and does the acute:chronic workload ratio measure anything real?
The Standard take
Load monitoring is validated for measuring training dose. It is not validated for predicting injury. The acute:chronic workload ratio's association with injury survives when the chronic-load term is replaced with random numbers, which means the relationship was largely mathematical coupling, not biology. No "sweet spot" threshold has causal support.
Historical Context
Gabbett's "training–injury prevention paradox" (BJSM, 2016) proposed the acute:chronic workload ratio — one week of load over a rolling four-week average — and reported that spikes outside roughly 0.8–1.3 came with elevated injury likelihood, while high chronic loads built gradually appeared protective. It was intuitive, actionable and immediately everywhere.
What the Evidence Shows
The ratio is mathematically coupled to itself. Lolli and colleagues showed algebraically and by simulation that because acute load sits inside the chronic-load denominator, ACWR generates spurious correlation independent of any biological relationship (BJSM, 2019).
The decisive test. Impellizzeri and colleagues took previously published datasets and replaced the real chronic loads with randomly generated numbers. The resulting "acute:random" ratio was associated with injury just as strongly as the true ACWR (Sports Medicine, 2020). A measurement whose signal survives the deletion of its own content is not measuring what it claims.
The conceptual critique. The same group's IJSPP paper catalogues the arbitrary time windows, ratio-scaling problems and non-independence, and their two-part Journal of Athletic Training series concludes the existing observational literature is unsuitable for deciding how to use load metrics to reduce injury.
The reviews agree. Maupin and colleagues' systematic review found inconsistent, sport-dependent associations with heavy methodological heterogeneity. Verstappen and colleagues' best-evidence synthesis in elite youth soccer found limited and conflicting evidence.
What load monitoring is good for. Session-RPE is well validated as a measure of internal training load, correlating with heart-rate-based indices. Quantifying dose and managing fatigue is a real, useful job. Forecasting injury is a different job.
Where It Breaks
Nielsen and colleagues' work on structure-specific load points at the deeper problem: a hamstring and a metatarsal do not share a dose-response curve, and a single whole-athlete number cannot carry both. Their companion paper, "Seven sins when interpreting statistics in sports injury science", lists the errors — dichotomising continuous exposures, ignoring reverse causation — that let the field mistake an association for a rule.
Reverse causation deserves its own line. Athletes ramp load up because they feel good. Feeling good and getting hurt are not independent.
Counterarguments
The pattern Gabbett described was not invented. Across several team-sport cohorts, rapid load spikes did come with more injuries, consistently enough to launch an entire subdiscipline. Something plausible sits underneath: tissue adapts to loads it has seen. The critique is not that load is irrelevant. It is that ACWR cannot tell you about it.
What would change our mind
Prospective, pre-registered studies using genuinely uncoupled acute and chronic windows that still find a dose-response relationship. Better still, randomised trials that manipulate load prospectively and measure injury incidence, with pre-specified calculation methods and standardised injury definitions. Tissue-specific models rather than one generic ratio.
What this does not prove
It does not prove training load is unrelated to injury, and it is not a licence to stop monitoring. It proves that the specific tool most of the industry adopted does not support the specific decisions most of the industry makes with it.
The Standard
Based on the evidence presented, Second City Standard believes ACWR should be treated as a descriptive plot, never as a gate. Monitor load, because knowing the dose is worth knowing. Do not tell an athlete a ratio says she is at risk this week — that sentence is not supported by anything currently in the literature.
The same shape of error, in a different discipline, in Where sports analytics actually beats judgement. More load and injury work in Sports Science.
Built By
Editorial Operating System v1.0Creator
Erik Chambers
Architecture
Founder, Creator & Editorial Architect
AI Assisted
No
Human Reviewed
Pending
Evidence Reviewed
In progress
Last Updated
September 9, 2026
Confidence
Evergreen medium/10
Research Status
current
We do not claim perfection. We promise transparency. Every investigation shows its work — the question, the evidence, the tools, the humans, and the updates.
Editorial Transparency
This article contains a combination of reporting, publicly available research, and editorial analysis.
Analysis and interpretation. Facts are sourced; conclusions are the author's. Evidence before opinion — facts require sources, analysis requires transparency, opinions require labels.
Meet the creator
Erik Chambers
Founder, Creator & Editorial Architect
Erik originated the central idea, directed the investigation, reviewed the evidence, and approved the final published work.
Read the founder profile →Challenge This
We welcome disagreement
A different way to read the evidence. Research that points in another direction. Where specialists diverge.
Loading challenges…
Submit a challenge
Sign in to submit a challenge. All submissions are reviewed before appearing publicly.
Comments
Ask A Question
What would you ask an editor about this piece?
Reader questions feed our coverage map. The most-asked ones become our next investigations.
Continue Exploring
Guided by the evidenceWhere should this take you next?
Does Training Load Predict Injury Risk?
A composite 0–100 measure of how well this investigation meets the Second City Standard. Scores are auditable — every point comes from the criteria below.
Composite
—
of 100
This investigation is queued for editorial scoring. No score has been assigned yet — the absence of a number is not a judgment on the evidence.
The Standard
· Editorial verdictBased on the evidence presented,
Second City Standard believes The acute:chronic workload ratio was treated as settled science for half a decade. Then someone replaced the chronic load with random numbers and the effect survived.
Remaining uncertainty: Awaiting a final written verdict from the editorial desk.
The Standard · Second City Standard
Continue The Investigation
· Never a dead endThis investigation is one thread. Pull the next one — every path below is another investigation, another question, or another discipline applied to the same problem.
Signal over noise.
One weekly dispatch. The best of 2ND CITY STANDARD, straight to your inbox.
No spam. Unsubscribe anytime.