A study published in Scientific Reports last week has reopened one of the longest-running debates in endurance training: is the polarized model, with most miles run very easy and a smaller hard slice, really better than the pyramidal model that adds a steady tempo middle? The answer the authors offer is the most precise to date. Across 601 marathon runners tracked through a full training block, the polarized group improved on average by 11.3 minutes versus 8.7 minutes for the pyramidal group, a roughly 30 per cent larger improvement despite a lower total training volume.
The interesting bit is the personalisation. Using a machine learning classifier trained on heart rate, pace and self-reported effort, the authors split runners into four response groups. About 31 per cent were "polarized responders" who improved sharply on the hard-easy plan. Another 32 per cent were "pyramidal responders" who did better with the steady-state-heavy plan. Roughly 19 per cent were "dual responders" who improved similarly on either, and the remaining 18 per cent were classed as non-responders, with no statistically meaningful improvement from either approach. That split is the most novel finding in the paper.
It rhymes with a separate body of work also published this spring. A Sports Medicine paper looking at the 2022 Boston Marathon found that the runners who improved most increased their running volume and quality sessions in the 12-to-4 months before race day, but reduced weekly running frequency in the final four months in favour of cross-training. The combination, the authors argue, allowed for higher stimulus quality without breakdown. Read alongside the Scientific Reports paper, the practical message becomes a little less polemical: more easy miles do help, but only up to the point at which the runner's body absorbs them.
Coaches reading the new study should be cautious about turning a population-level finding into an individual prescription. The polarized advantage is real but the size of that advantage depends on the runner's response cluster, which can only be measured with several weeks of training data. For mid-pack marathon runners in particular, the dual-responder cluster is large enough that swapping one plan for another may produce no detectable benefit and a meaningful disruption to routine. The authors' own suggestion is that the first four weeks of any new block should be treated as diagnostic rather than performance-defining.
There is also a deeper finding tucked into the discussion. The training plans that produced the largest gains were those whose hard sessions sat at clearly threshold or VO2-max paced efforts, not in the grey zone above tempo but below intervals. The non-responder cluster, the 18 per cent who did not improve on either plan, were disproportionately those who logged a large volume in that grey zone. For runners targeting an autumn marathon in Berlin, Chicago or New York, the practical takeaway is sharper than the headline number: keep easy days truly easy, take hard sessions to a clearly hard intensity, and do not be afraid of a lower weekly mileage if recovery is the limiting factor.