Training Load and Readiness Scores Explained: The Maths Behind the Number Your App Shows

Open two training apps after the same week of running and you may find one calling you “productive” while the other warns of “overreaching”. Both used the same workouts. Both display a confident number. The disagreement is not a bug; it is the natural result of different formulas, different time windows and different baselines applied to identical data. This article walks through what acute and chronic load actually compute, why the ratio between them is fragile, what the morning HRV reading measures and does not measure, and how to read these scores as trends rather than verdicts.

What a single session’s load is made of

Every load model starts by turning one workout into one number. The oldest method, published in 1975, is the training impulse, usually written TRIMP. It multiplies session duration in minutes by a weighting that rises with average heart rate expressed as a fraction of the range between resting and maximum, so a 60-minute run at 75 percent of heart rate reserve scores more than the same hour at 60 percent, and much more than a 30-minute walk. A common variant assigns fixed multipliers per zone, for example 1 for zone 1 up to 5 for zone 5, and sums the minutes spent in each. Another family uses power or pace instead of heart rate: an hour at threshold intensity is defined as 100 points, and everything else is scaled to it by the fourth power of intensity, which punishes hard efforts steeply.

A heart rate model rewards long easy sessions and under-counts short intervals, because the heart lags the effort. A power or pace model does the opposite. The same 10 by 400 metre session can score 40 points in one app and 75 in another, and neither is wrong; they are measuring different things.

Acute load, chronic load and the ratio between them

Acute load is the sum, or the average, of daily session loads over a short window, most commonly the last 7 days. Chronic load is the same calculation over a longer window, usually 28 or 42 days. The idea is that chronic load represents the fitness you have built and acute load the fatigue you are carrying right now. Dividing acute by chronic gives the acute-to-chronic workload ratio. A widely cited range of 0.8 to 1.3 was described as the “sweet spot”, with values above 1.5 associated in some team-sport datasets with higher injury rates in the following week.

Two implementation details change the answer. First, whether the windows use simple rolling averages or exponentially weighted moving averages. A rolling 7-day average treats last Sunday’s long run exactly like this morning’s session; an exponentially weighted version with a 7-day time constant lets it decay, so the acute number falls faster after a hard block. Second, whether the acute window is included inside the chronic one. If the last 7 days count in both numerator and denominator, the ratio is mathematically pulled towards 1 and the extremes are compressed. Apps rarely say which choice they made, yet a ratio of 1.4 in one convention can correspond to 1.6 in another.

The ratio has also been criticised on statistical grounds. Because the same recent data appear in both terms, the correlation with injury partly reflects the arithmetic rather than physiology, and the “danger zone” thresholds came from specific sports with specific load measures. Treat it as a description of how much you have ramped up, not as a prediction.

Fitness, fatigue and the form curve

Many apps display a second model: an impulse-response curve where fitness is a slow accumulation with a 42-day time constant, fatigue a fast one with a 7-day constant, and form is fitness minus fatigue. Numerically, each day fitness moves about 2.4 percent of the way towards today’s load, and fatigue about 13 percent. After a rest day fatigue drops sharply while fitness barely moves, so form rises; that is the mechanism behind every “you are fresh” message. The constants are defaults borrowed from studies on small groups of athletes. Your true recovery speed may be faster or slower, and the model has no way of knowing.

What the morning HRV reading measures

Heart rate variability is the variation in the time between consecutive beats. The metric most apps use is rMSSD, the root mean square of successive differences, expressed in milliseconds and often log-transformed to make the scale more linear. Higher values generally reflect stronger parasympathetic activity, which is the “rest and digest” branch of the nervous system. In practice a well-rested adult might read anywhere from 20 to 100 ms; the absolute number is largely genetic and depends on age, so comparing your value with a friend’s is meaningless. What carries information is your own value relative to your own baseline.

The reading is sensitive to almost everything. Alcohol the night before typically lowers it by 20 to 30 percent. A hard session, a late meal, dehydration, an early flight, the first days of a cold before symptoms appear, and even a stressful email all leave a mark. Seven consecutive mornings that sit below your 30-day normal range tell you something is accumulating, and that is exactly the pattern that precedes most cases of overreaching and illness.

Reading the trend rather than the number

Because every score in this article is a model output, the productive question is never “is 1.35 bad?” but “what is the direction, and does it match how I feel and perform?” A handful of habits make the data useful:

  • Pick one app as the reference and stop comparing absolute values across apps; compare each app’s number with its own history.
  • Take HRV at the same time and in the same position every morning, before coffee, and judge it against a 7-day rolling average inside a 30 or 60-day normal band.
  • Let chronic load rise by roughly 5 to 10 percent per week and hold or cut back for a week every third or fourth week.
  • Log two subjective scores daily, sleep quality and leg soreness, and see whether they move before or after the app’s readiness score does.
  • Cross-check with performance: if pace at a fixed easy heart rate is stable or improving, the load is being absorbed whatever the ratio says.

When two apps disagree

When the scores diverge, look for the cause in this order. Check that both apps received the same activities, since a missing synced session changes a 7-day acute load by up to 30 percent. Check what unit each app uses for session load, heart rate based or pace based, because a week of intervals will split them. Check the baseline: an app that only has 3 weeks of your history has a chronic load that is still climbing, so its ratio reads high for no reason. And check the HRV source, because a watch, a ring and a chest strap measure at different times of night with different filtering, and their baselines are not interchangeable. Once those are aligned, the remaining gap is the models’ opinion, and your own logbook is the tiebreaker.