October 17, 2026 · 12 min read

Am I training in the right proportions? Zones, energy systems and polarisation

I was convinced I trained in a polarised way: lots of easy riding, some hard intervals, little of the “middle”. Three cards on the Performance tab — Time in zones, Energy systems and Intensity distribution — showed something else. This is the third article in the series about the Progress page, after the Form tab and the power curve and rider profile.

Three cards, one source of data

It's worth knowing straight away: all three cards are computed from the same record — the seconds spent in each of the seven power zones, saved after every ride. They differ in how they group those seconds:

Power zones are only saved for rides computed with a known FTP — which is why 135 of my 164 hours this season are in zones. The missing hours are mostly rides without a power meter and two weeks around the turn of April and May that have no zone breakdown.

Time in zones

The first card shows every week as a bar split into zones. The Minutes / Share switch turns hours into percentages — useful when you compare a 3-hour week with a 12-hour one. The Power / Heart rate switch shows the same in heart rate zones. This is my whole season, added up:

73% of the time in Z1 and Z2 looks like a textbook aerobic base. Careful with Z1: 45 hours is not just recovery, but also warm-ups, cool-downs, descents and sitting in a group. How the zones are divided and where the Coggan model gets it wrong, I describe in the articles on power zones and on the limits of the Coggan model.

Heart rate zones tell a different story from power zones

The app has only been saving heart rate zones since mid-September, when it started storing the zone limits valid on the day of each ride — older rides have nothing to show. So for those five weeks I have both views:

Heart rate shows a little more time at the top. That is normal: heart rate rises with a delay and drifts upwards towards the end of a long ride, so it stays high for a while after an interval. Power zones tell you what your legs did, heart rate zones how much it cost your heart. Why it's worth having both, I explain in the article on combining heart rate and power.

Energy systems

The second card groups the same seconds into three systems: aerobic (Z1–Z4), glycolytic (Z5–Z6) and neuromuscular (Z7). For me that is 93%, 6% and 1%. The second view, “Below / above threshold”, splits the time in two at the top of Z4, which is 105% of FTP — not exactly at FTP. The reason is technical and honest: Z4 straddles FTP, and the split cannot cut a zone in half.

93% aerobic looks like a lot, but for an amateur it is a typical number — most riding, even hard riding, is aerobic work. Above all, this card tells you one thing: the shares are of time, not of training load. A minute in Z6 is far more tiring than a minute in Z2, yet here they count the same. How much load each system produced is shown by a separate card — more on it in a moment.

Load by energy system

This card sits on the Form tab, below the ACWR chart, and is the most “modelled” card on the whole page. It takes each ride's TSS — the same single number the PMC is computed from — and splits it between the three systems according to time in zones. Not evenly per second: every zone has a typical intensity assigned to it, from 0.45 of FTP in Z1 to 1.75 of FTP in Z7, and a second weighs as much as the square of that intensity — exactly the way TSS itself grows. So a minute in Z6 weighs more than four times as much as a minute in Z2. Then the app runs the same CTL and ATL over each of the three parts as in the ordinary PMC.

system's share of a ride's TSS = Σ(seconds × IF²) in the system's zones ÷ Σ(seconds × IF²) over the whole ride

Weighting by intensity changes the picture from the previous card:

Six per cent of the time above threshold produced almost a fifth of my training load. This is what the three fitness lines looked like over the season:

The most interesting part is June versus October. On 18 June aerobic CTL was 24, glycolytic 7, neuromuscular 2.7. On 9 October: 40, 7 and 1.3. In autumn I built almost nothing but an aerobic base — the glycolytic line stood still and the neuromuscular one halved. It is the same conclusion as from polarisation, only shown as load: I am short of hard minutes, not of easy hours. The switch above the chart shows the same for fatigue (ATL): on 15 October it was 46 aerobic, 3 glycolytic and under 1 neuromuscular — the last week was almost purely aerobic.

Two things are worth knowing before you start trusting these lines. First, it is a model, not three measurements: one load is measured per ride, and the split follows from the assumed zone intensities. Read the proportions between the lines, not their height. Second, rides with no recorded zone time are left out of these lines entirely — for me that is 33 of 138, mostly from spring. That is why in June the three lines add up to 34 even though the ordinary CTL was 48, while in October, when almost every ride has a breakdown, they add up to 48.5 against a CTL of 49. The app says below the chart how many rides it left out — worth a glance before drawing conclusions about spring.

Intensity distribution and polarisation

The third card has three views, and it is the one that answers the question in the title. The first, Distribution, splits every week into three zones: easy (below 75% of FTP), threshold (75–105%) and hard (above 105%).

The second view, Polarisation, computes the published polarisation index for every week [1]:

PI = log₁₀(easy ÷ threshold × hard × 100)

where each zone is its share of the week (e.g. 0.73). A week is polarised when PI is at least 2.0 and there is more easy riding than threshold riding. The second condition matters: the index alone can be high in a week that was a threshold block with one sprint. Weeks with no hard riding or no threshold riding have no index at all — for me that is 4 weeks out of 28.

Of the 24 weeks with an index, one was polarised: 20–26 April, with a distribution of 64% / 13% / 23% and a PI of 2.07. The whole season together is 73% / 20% / 7% and a PI of 1.38. That is not polarised training but pyramidal: most of it easy, less threshold, the least very hard. What my autumn weeks lack for polarisation is not easy riding but hard minutes — in October I spent 1–4% of the time above 105% of FTP.

Is that bad? Studies comparing training models suggest that polarised training gives trained athletes slightly larger gains [2, 3], but the pyramid is also an established, effective model — and for an amateur with 6–9 hours a week often a more realistic one. What matters is knowing which model you actually follow, not which one you claim to. I write about the 80/20 approach on the trainer in the article on polarised training.

Sweet spot

Sweet spot, 88–94% of FTP, crosses the line between “easy” and “threshold” in one view and sits in the middle of Z3–Z4 in the other. That is why the app doesn't draw it as a piece of a bar but gives it as a separate share of the week. For me it averages 3.8% of the time in zones, peaking at 15% in mid-June. When sweet spot makes sense, I discuss in the article on sweet spot training.

Pacing

The third view, Pacing, shows the variability index (VI): normalised power divided by a ride's average power, averaged over the week. 1.00 is a perfectly even effort.

This chart shows where I rode. In spring and summer, outdoors, VI was 1.15–1.26: climbs, descents, junctions. From mid-September, when I moved to the trainer in ERG mode, it dropped to 1.04–1.07, because the trainer holds the target power. A higher VI is not worse — intervals are meant to be uneven. What deserves attention is a long endurance ride with a VI of 1.2 or more: a sign that the “easy” ride wasn't easy at all. I write more about normalised power in the article on WAP / NP.

What I take away from this

Next in this series: efficiency factor, cardiac drift and cadence efficiency — in other words, whether my endurance is improving.

Sources

1. Treff G, Winkert K, Sareban M, Steinacker JM, Sperlich B. The Polarization-Index: A Simple Calculation to Distinguish Polarized From Non-polarized Training Intensity Distributions. Front Physiol. 2019;10:707. doi:10.3389/fphys.2019.00707
2. Stöggl T, Sperlich B. Polarized training has greater impact on key endurance variables than threshold, high intensity, or high volume training. Front Physiol. 2014;5:33. doi:10.3389/fphys.2014.00033
3. Seiler KS, Kjerland GØ. Quantifying training intensity distribution in elite endurance athletes: is there evidence for an “optimal” distribution? Scand J Med Sci Sports. 2006;16(1):49–56. doi:10.1111/j.1600-0838.2004.00418.x
4. Allen H, Coggan AR, McGregor S. Training and Racing with a Power Meter. 3rd ed. Boulder, CO: VeloPress; 2019.

← All blog posts