Injury risk & load progression

The intuition here is sound and worth keeping: when training load rises fast, whether that is your weekly volume or your single longest run, injuries tend to follow. Building gradually is sensible. What is not settled is the popular arithmetic that claims to tell you exactly how fast is too fast. Two rules get quoted constantly, and both are on shakier ground than their confidence suggests.

The "10% rule"

The rule of thumb: do not increase your weekly volume, or your longest run, by more than 10% from one week to the next. It is tidy, easy to remember, and repeated everywhere.

It is also coaching folk wisdom with no primary derivation. Nobody measured runners and arrived at 10% as a threshold. When it was finally put to a proper test, the result was awkward for the rule: in a randomized controlled trial (Buist et al., 2008), novice runners built up on a 10%-graded program were injured at essentially the same rate (about 20.8% versus 20.3%) as runners who built up on a shorter, steeper ramp (an 8-week program versus 13 weeks). The gradual group did not come out ahead.

So treat 10% as a rounded, memorable nudge toward "increase gradually," not as an evidence-based line. There is nothing special about the number itself.

The acute:chronic workload ratio

The second rule is more numerical. It compares your recent load against your habitual load. Acute load is roughly the last 7 days. Chronic load is roughly the last 28 days, expressed as a weekly average. Divide one by the other:

ACWR = load_{acute (7d)} ÷ load_{chronic (28d avg)}

A ratio near 1.0 means this week looks like a normal week for you. A popular reading calls 0.8-1.3 a "sweet spot" and anything above 1.5 a "danger zone." Hold on to those numbers loosely, for reasons we get to below.

Why the ratio is on thin ice

The ACWR has a structural flaw that is easy to miss. The acute load is part of the chronic load: the last 7 days are also inside the last 28 days. The numerator and denominator are coupled, and dividing two quantities that share a component manufactures correlation that is not really about injury at all. The ratio is also sensitive to bookkeeping choices most people never see: how you bin the days, whether you use a flat average or an exponentially weighted one, where you set the windows.

The mathematical-coupling problem was formalized by Lolli et al. (2019), and Impellizzeri et al. (2020) laid out the full set of pitfalls; several independent groups have then failed to reproduce the injury-prediction claims. Crucially, no study has properly estimated a causal effect, so the 0.8-1.3 sweet spot and the 1.5 danger line are not validated decision rules.

In fairness, the ratio still has defenders, and watching how your recent load compares to your habitual load is genuinely useful for spotting when you have ramped up quickly. The dispute is specifically about treating a fixed threshold as an injury predictor, not about tracking load trends at all, which is sensible. What is not earned is dressing a single number up as a red line.

What cadennce actually does

Given all that, we do not gate anything on these numbers and we do not throw hard warnings. Instead we surface the underlying pattern as a quiet, suggestive nudge: "your longest run jumped a lot this week," or "your recent load is climbing quickly." The point is to draw your eye, not to diagnose you or tell you to stop.

The honest takeaway to carry with you is simpler than any ratio. Build gradually, avoid large sudden spikes, and let how you feel and any niggles carry more weight than a single number. A twinge in your shin tells you more than an ACWR of 1.4 ever will.