Polarised training: what 80/20 actually asks of you

What the model says

Sort your sessions into three bands: below the first ventilatory threshold (easy, conversational), between the two thresholds (the middle — hard breathing, short sentences, sustainable but unpleasant), and above the second (hard intervals). Polarised distribution puts roughly 75 to 80% of volume in the first band and 15 to 20% in the third, leaving very little in the middle. A systematic review of that distribution found it improved VO₂ max and work economy — and the authors flag that the number of studies is small enough to limit how far the finding generalises.

That caveat is worth repeating, because the online version of this advice has none. The literature contains papers arguing polarised training is optimal for endurance athletes and papers arguing it is not, published in the same year in the same journal. Anyone telling you it is settled has not read past the first result.

The middle is where recreational training goes to die

Here is the failure mode the model is really pointing at. Left to instinct, most people train almost everything at a moderate intensity: hard enough to feel like training, easy enough to complete. Every run at the same tempo. The result is a training week with no easy sessions to accumulate volume in and no hard sessions to drive adaptation — everything is a compromise, and compromises do not produce much of either.

The practical instruction inside polarised training is therefore simpler than the percentages suggest: make your easy days genuinely easy, so that your hard days can be genuinely hard. Most of the benefit people report comes from the first half of that sentence.

Zones, and why yours may be wrong

The bands are usually approximated from a percentage of maximum heart rate, and maximum heart rate is usually estimated from age. That estimate is decent as a population average — the best-known equation predicts it from age alone, independent of sex and habitual activity — but it carries real spread between individuals. If your true maximum sits ten beats away from the formula, every zone boundary derived from it is wrong by the same amount, and a session you file as easy may not be.

The cheap correction is the talk test. Easy means full sentences without gasping. If you cannot hold a conversation, the session is not in the band you think it is, whatever the watch says.

What your data can and cannot tell you

It can tell you your actual distribution, which is usually the surprise. Add up your minutes by zone across a fortnight rather than judging a single session, and compare that to the 80/20 you believe you are doing. Most people discover a large middle band they did not intend.

What it cannot do is tell you the distribution is working. That shows up over months, in a resting heart rate that drifts down and in hard sessions whose overnight cost keeps arriving on schedule rather than stacking. And the model is not a licence to add hard days: the consensus statement on overreaching is clear that overload without adequate recovery is how you go backwards, and the 20% is a ceiling, not a target to beat.

Where to start

Count sessions before you count minutes. Four easy and one hard is close enough to the ratio to begin with, and it removes the arithmetic. Then check yourself against your own data: mark the block as a period in Vitra and look at your zone distribution, your resting heart rate trend and how quickly you recover from the hard day — computed on your own machine, against your own history, which is the only baseline that means anything for zone boundaries this personal.

Frequently asked questions