The Fitness-Fatigue Chart: What Training Load Scores Can and Can't Tell You
The fitness, fatigue and form lines in training apps come from a 1970s model that fits performance data surprisingly well but can't predict injury. What the research supports, and how to use the chart without obeying it.
Sprint Summary
The short version — read this if you're short on time.
Many training apps show three lines: fitness, fatigue and form. They come from the Banister impulse-response model, first proposed in the 1970s, which treats each session as adding a slow-fading fitness effect and a fast-fading fatigue effect. The model can fit performance data well: in 9 elite swimmers it explained much of the change in performance over a season. But the numbers that describe how long fitness and fatigue last were very uncertain for each swimmer, and apps use fixed defaults for everyone. In a triathlon study, a taper shaped by the model improved cycling performance more than a simple cut in training.
The chart can't predict injury. The acute:chronic workload ratio, a close cousin of the form line, was promoted as an injury predictor, but a systematic review found methods too varied for firm advice, a critique found no evidence that managing it reduces injury, and in 5,205 runners it wasn't linked to injury in the expected way. A single run much longer than your recent longest was the clearer risk. Use the chart to see trends, plan gradual build-ups and time your taper, and give equal weight to how you feel.
No training load chart or ratio has been shown to predict injury for an individual. Pain that persists, worsens during sessions or changes how you move needs assessing by a clinician, whatever your form score says.
A healthy-looking fitness line doesn't rule out overreaching, illness or low energy availability. Persistent fatigue, poor sleep, frequent illness or, in women, missed periods are reasons to ease off and see a GP or sports doctor; TriForward's "Under-Fuelled, Not Undertrained" covers low energy availability.
Don't chase a steeper fitness line by stacking load. The models suggest more training eventually stops helping, and a single run much longer than your recent longest was linked to higher injury rates in a large cohort.
Full Distance
The complete research and analysis.
Open almost any training platform and you'll find a chart with three lines. One, usually called fitness, climbs slowly through a training block. Another, fatigue, jumps after every hard week. The third, form or freshness, is the gap between them: negative when you're tired, positive when you've rested. Many age-group triathletes check this chart daily, plan weeks around making the fitness line rise, and worry when form dips deep into the negative.
Where does the chart come from, how good is the science behind it, and what can it tell you? This guide looks at the model behind it, how well it fits real performance, the separate claim that load ratios predict injury, and how to use the chart without obeying it.
Where the lines come from
The fitness-fatigue chart is a version of the Banister impulse-response model. A 2026 paper by Kontro and colleagues in PLOS One traces it back to a systems model published by Banister and colleagues in 1975, later reduced to two components: fitness and fatigue. The idea is simple:
- Every training session has two effects: it makes you a little fitter and a little more tired.
- Both effects fade over time, following an exponential decay, each with its own time constant. Fatigue is normally modelled as fading faster than fitness.
- Performance at any moment depends on your baseline plus the accumulated fitness, minus the accumulated fatigue.
In the full model, each athlete has their own values for how strongly each session builds fitness and fatigue, and how long each lasts. These values are estimated by fitting the model to repeated performance tests.
Training apps simplify this. They turn each session into a load score, usually from power or heart rate and the session's duration and intensity. Kontro and colleagues note that popular platforms calculate the fitness line, often labelled chronic training load, as an exponentially weighted moving average of daily load, using fixed decay periods. Inference: the app's chart is therefore a weighted running average of your recent training, not a measure of your body. It doesn't know how you slept, what you ate or whether you're fighting a cold.
How well does the model fit?
Nine elite swimmers over a season (2006)
Hellard and colleagues, in the Journal of Sports Sciences in 2006, tested how useful the model is for monitoring training. They fitted it to the training loads and performances of 9 elite swimmers across one season.
- The model fitted performance well for every swimmer: it explained on average 79% of the variation (R² = 0.79 ± 0.13).
- But the key parameters were very uncertain. One of the time constants describing how long training effects last was estimated at 38 days, with a 95% confidence interval from 17 to 59 days. Another was 19 days, with a range from 6 to 32 days.
- Some parameters were highly correlated with each other, which the authors said made interpreting them "worthless".
So the model can reproduce the shape of a season after the fact, but it's much less certain about the specific numbers that would let you predict an individual's response. Inference: if the time constants were that uncertain for elite swimmers tested regularly, the fixed defaults in an app are at best a rough approximation for any one age-grouper.
Six untrained people over 15 weeks (2003)
Busso, in Medicine & Science in Sports & Exercise in 2003, applied the model to 6 previously untrained people over a 15-week programme, with performance tested three times a week using a 5-minute all-out effort.
- Performance improved by 30% (± 7%).
- A version of the model that let fatigue vary with training load fitted better than the original: a standard error of 6.47 W versus 9.20 to 10.31 W.
- The modified model implied an inverted-U relationship between daily training and performance: beyond a point, more daily training reduced performance.
Inference: the original chart assumes each session's fatigue is the same however much you've been doing. The Busso study suggests fatigue builds disproportionately when training is heavy, which the simple chart can understate.
A model-shaped taper in triathletes (1999)
The model has been used to plan tapers. Banister, Carter and Zarkadas, in the European Journal of Applied Physiology in 1999, simulated different tapers in the model and then tested them in triathletes in field trials.
- An exponential taper, cutting training quickly at first then more gradually, improved cycle ergometry performance more than a single step reduction.
- A fast exponential taper beat a slow one in cycle ergometry.
- The 5 km run improved in the same direction, but the differences were not statistically significant.
The abstract doesn't give the total number of triathletes or a sex split, so this is small, older evidence. TriForward's "Tapering for a Sprint or Olympic: Shorter Race, Shorter Taper?" covers taper evidence in more detail.
What a load score can and can't capture
The chart is only as good as the load scores feeding it. A 2017 review by Passfield and colleagues in the Journal of Sports Sciences, looking at power data from cycling, concluded that simple summaries can't reflect how variable a session's effort was or its likely metabolic stress, and that current methods of quantifying training load are inadequate.
Kontro and colleagues make a related point: impulse-response models squeeze all training stress into a single number and assume the body adapts the same way whatever type of training produced it, despite clear evidence that different intensities and durations cause different responses.
For a triathlete there are extra problems. Inference, from how the scores are built:
- Three sports, one number. A swim, a ride and a run with the same score may not cost the same. Running involves impact and muscle damage that swimming doesn't.
- Different inputs. Bike scores often come from power, run scores from pace or heart rate, swim scores from pace or a manual estimate. Each brings its own errors. TriForward's "Do You Need a Bike Power Meter?" and "Wrist or Chest Strap?" cover device accuracy.
- What's missing: strength sessions, work stress, poor sleep and illness rarely appear, yet all affect how tired you are.
The injury question
The chart is often read as an injury warning: if fatigue rises too fast relative to fitness, you're at risk. This idea comes largely from research on the acute:chronic workload ratio (ACWR), which compares the last week's load with the average of the last three or four weeks. Inference: the ACWR is a close cousin of the form line, which compares short-term and long-term load as a difference rather than a ratio.
The original proposal (2016)
In a 2016 paper in the British Journal of Sports Medicine, Gabbett described the "training-injury prevention paradox": athletes used to high loads tend to have fewer injuries, while large, rapid increases in load are likely responsible for many soft-tissue injuries. The paper proposed the ACWR as a best-practice predictor of training-related injury. It was hugely influential, especially in team sports.
A systematic review (2020)
Andrade and colleagues, in Sports Medicine in 2020, reviewed 20 studies of the ACWR and injury in professional team sports, covering 1,234 athletes (all men, average age 24) and 2,375 injuries.
- Most studies suggested that a higher ratio went with more time-loss injuries.
- Methods varied widely: the studies used 14 different ways of grouping ratio values, and almost none used the same categories.
- The authors concluded that the varied methods limit the strength of any recommendation.
A methodological critique (2020)
Impellizzeri and colleagues, in the International Journal of Sports Physiology and Performance in 2020, went further. They argued that:
- No study had properly tested whether changing the ratio changes injury rates, so using it to manage injury risk is conjecture.
- The ratio has statistical problems, including how the numerator and denominator are linked, which can create artificial relationships.
- Their conclusion: there is no evidence supporting the ACWR in training-load management or to reduce injury risk.
The largest running study (2025)
The most relevant data for triathletes come from Frandsen and colleagues, in the British Journal of Sports Medicine in 2025. They followed 5,205 adult runners (average age 45.8, 22% women) for 18 months, using data from their GPS watches: 588,071 runs and 1,820 runners with overuse injuries.
- A single run more than 10% longer than the runner's longest run in the previous 30 days was linked to higher injury rates: hazard rate ratios of 1.64 for a 10 to 30% spike, 1.52 for 30 to 100%, and 2.28 for more than double.
- The ACWR showed a negative dose-response: higher weekly ratios were not linked to more injuries in the way the theory predicts.
- The week-to-week change in distance showed no relationship with injury.
TriForward's "Forget the 10% Weekly Rule: The 5,205-Runner Study" covers this study in more detail. Inference for the chart: a smooth fitness line built from weekly totals can hide the single long run that matters most.
What the chart is good for
The evidence above, and the evidence on subjective measures, point to a modest but real role for the chart.
Seeing trends
The chart shows at a glance whether your training is building, holding or falling, and whether recent weeks were unusually heavy. That's useful for planning and for spotting when life has quietly eroded your training.
Planning gradual build-ups
A steadily rising fitness line, without sudden jumps in fatigue, is a reasonable picture of a sensible block. Inference: the chart doesn't tell you the right rate of increase for you, and the running study suggests watching your longest single session matters more than weekly ramps.
Timing a taper
The model's main practical success is tapering: in the triathlon study, a model-shaped taper beat a simple cut. Inference: watching fatigue fall and form rise in the last two weeks is a reasonable check that you're actually resting.
Pairing it with how you feel
A 2016 systematic review by Saw, Main and Gastin in the British Journal of Sports Medicine looked at 56 studies measuring athletes' well-being. Subjective measures, such as mood, perceived stress and recovery, tracked training load more sensitively and consistently than objective measures such as blood markers and heart rate. Inference: a quick daily rating of fatigue, soreness, sleep and mood adds what the chart can't see.
When the chart misleads
Because the chart only knows the load scores it's given, some everyday situations make it misleading. Inference, from how the lines are calculated:
- After illness. A week off with a virus lowers fatigue and pushes form positive, so the chart says you're fresh. You may be anything but. TriForward's "Train Through a Cold? The 'Neck Check' Rule and When to Rest" covers returning after illness.
- After a threshold change. Most load scores are calculated relative to your threshold power or pace. Update the threshold and the same session scores differently, so the lines jump without any change in your body.
- Missing or faulty data. A forgotten watch, a dropped power signal or a swim logged without a score leaves holes that look like rest.
- Heat and altitude. The same power or pace costs more in the heat, but a power-based score doesn't know that.
In each case the fix is the same: read the chart alongside what actually happened and how you feel.
A triathlon scenario
The following example is illustrative, not drawn from a study. A 41-year-old age-grouper building for a middle-distance race watches his fitness line closely. Over three weeks he raises it quickly, partly by turning his usual 75-minute long run into a 2-hour run. His form drops to a number his app colours red, and he wonders whether to take a rest day or push on.
- He checks the single sessions, not just the line. The 2-hour run was well over 10% longer than his longest run in the past month, the kind of spike linked to higher injury rates in the running study.
- He rates how he feels: heavy legs, poor sleep, a niggle in his Achilles.
- He eases back, shortens the next long run and builds it more gradually.
- In race week, he uses the chart to check that fatigue has fallen and form has turned positive, alongside how fresh he feels.
Common mistakes
- Treating the fitness line as fitness. It's a summary of recent training.
- Chasing a steeper line by stacking load.
- Reading form as an injury forecast. The evidence doesn't support it.
- Ignoring single big sessions hidden in weekly totals.
- Trusting one number across three sports without question.
- Letting the chart overrule pain, illness or fatigue.
What we don't know
- How well the model fits age-group triathletes, with three sports and irregular testing.
- What decay periods suit individuals, given how uncertain they were even in elite swimmers.
- Whether managing load ratios reduces injury in any sport. No trial has tested it.
- How to weight swim, bike and run load in a single score.
- Whether using a chart improves performance compared with training by plan and feel.
References
- Kontro H, Mastracci A, Cheung SS, MacInnis MJ. The three-dimensional impulse-response model: modeling the training process in accordance with energy system-specific adaptation. PLOS One, 2026;21(2):e0341721. doi:10.1371/journal.pone.0341721
- Hellard P, Avalos M, Lacoste L, Barale F, Chatard JC, Millet GP. Assessing the limitations of the Banister model in monitoring training. Journal of Sports Sciences, 2006;24(5):509–520. doi:10.1080/02640410500244697
- Busso T. Variable dose-response relationship between exercise training and performance. Medicine & Science in Sports & Exercise, 2003;35(7):1188–1195. doi:10.1249/01.MSS.0000074465.13621.37
- Banister EW, Carter JB, Zarkadas PC. Training theory and taper: validation in triathlon athletes. European Journal of Applied Physiology and Occupational Physiology, 1999;79(2):182–191. doi:10.1007/s004210050493
- Passfield L, Hopker JG, Jobson S, Friel D, Zabala M. Knowledge is power: issues of measuring training and performance in cycling. Journal of Sports Sciences, 2017;35(14):1426–1434. doi:10.1080/02640414.2016.1215504
- Gabbett TJ. The training-injury prevention paradox: should athletes be training smarter and harder? British Journal of Sports Medicine, 2016;50(5):273–280. doi:10.1136/bjsports-2015-095788
- Andrade R, Wik EH, Rebelo-Marques A, Blanch P, Whiteley R, Espregueira-Mendes J, Gabbett TJ. Is the acute:chronic workload ratio (ACWR) associated with risk of time-loss injury in professional team sports? A systematic review of methodology, variables and injury risk in practical situations. Sports Medicine, 2020;50(9):1613–1635. doi:10.1007/s40279-020-01308-6
- Impellizzeri FM, Tenan MS, Kempton T, Novak A, Coutts AJ. Acute:chronic workload ratio: conceptual issues and fundamental pitfalls. International Journal of Sports Physiology and Performance, 2020;15(6):907–913. doi:10.1123/ijspp.2019-0864
- Schuster Brandt Frandsen J, Hulme A, Parner ET, Møller M, Lindman I, Abrahamson J, Sjørup Simonsen N, Sandell Jacobsen J, Ramskov D, Skejø S, Malisoux L, Bertelsen ML, Nielsen RO. How much running is too much? Identifying high-risk running sessions in a 5200-person cohort study. British Journal of Sports Medicine, 2025;59(17):1203–1210. doi:10.1136/bjsports-2024-109380
- Saw AE, Main LC, Gastin PB. Monitoring the athlete training response: subjective self-reported measures trump commonly used objective measures: a systematic review. British Journal of Sports Medicine, 2016;50(5):281–291. doi:10.1136/bjsports-2015-094758
Frequently asked questions
What do the fitness, fatigue and form lines actually mean?
They come from the Banister impulse-response model. Each session adds to a slow-fading 'fitness' line and a fast-fading 'fatigue' line, and 'form' is the difference. It's a mathematical summary of your recent training, not a measurement of your body.
Can the chart predict my race performance?
Roughly, for groups and with lots of test data. In 9 elite swimmers, the model fitted performance well, but the parameters that describe how long fitness and fatigue last were very uncertain for each individual. Apps use fixed defaults, so treat the lines as a rough guide.
Does a high acute-to-chronic load ratio mean I'll get injured?
Not reliably. Reviews of team-sport studies found inconsistent methods, a methodological critique found no evidence that managing the ratio reduces injury, and in 5,205 runners the ratio wasn't linked to injury the way you'd expect. A single run much longer than your recent longest was the clearer warning sign.
How should I use the chart?
Use it to see trends, plan gradual build-ups and check that your taper lets fatigue fall before a race. Combine it with how you feel, which a review of 56 studies found tracked training load better than objective measures. Don't let it overrule pain, illness or fatigue.
The evidence behind this guide
2 graded principles, 10 sources. Open one to see the sources.
Acute:chronic load ratios, and by extension form scores on load charts, do not reliably predict injury; in runners, single-session spikes were the clearer signalMixed evidence4 sources
In practice: Don't read a chart's form or ratio as an injury forecast. Watch single-session jumps, especially your longest run relative to the last month, and treat pain, illness and persistent fatigue as reasons to ease off regardless of the chart. No trial has shown that managing load ratios reduces injury, and evidence comes from team sports and runners, not triathletes.
Proposes that athletes accustomed to high loads have fewer injuries, that excessive and rapid load increases likely cause many non-contact soft-tissue injuries, and describes the acute:chronic workload ratio as a best-practice predictor of training-related injury. Influential but later contested.
Most studies suggested higher ACWR was associated with greater time-loss injury risk, but 95% used the coupled calculation, 14 different binning methods were used and almost no two studies matched, so heterogeneous methods limit the strength of recommendations. Men only, team sports.
IJSPP 15(6):907-913. Lays out the methodological pitfalls (mathematical coupling, ratio instability, arbitrary thresholds) in acute:chronic workload ratio metrics — the statistical family that simple percentage-increase 'rules' belong to.
A single session more than 10% longer than the runner's own longest run in the previous 30 days carried a significantly higher injury hazard ratio — 1.64 for a 10-30% jump, 1.52 for 30-100%, and 2.28 for more than double — while week-to-week volume change showed no such relationship. This is the TriForward-referenced study behind 'forget the 10% weekly rule': what matters is the size of a single session jump relative to recent long-run history, not a smooth weekly progression number.
The Banister fitness-fatigue model behind training-load charts can fit performance data, but its individual parameters are very uncertain, so app charts are rough summaries rather than predictionsLimited evidence6 sources
In practice: Use the chart to see training trends, plan gradual build-ups and check that fatigue falls in a taper, alongside subjective ratings, which tracked load better than objective measures in a review of 56 studies (Saw 2016). Don't treat the fitness line as measured fitness or chase a steeper line; fixed app defaults may not fit you, and transfer to three-sport triathlon training is inference.
Methods/theory paper. Traces the impulse-response model to Banister and colleagues (1975), reduced to fitness and fatigue components that decay exponentially with their own time constants; notes popular platforms compute fitness (chronic training load) as an exponentially weighted moving average of daily load. Argues single-metric load inputs wrongly assume adaptation is independent of training type, and proposes a three-dimensional model based on critical power, W' and maximal power. Full text read via Europe PMC.
The Banister model related training load to performance in every swimmer (R2 = 0.79 ± 0.13) and estimation seemed stable, but 95% confidence intervals of key parameters were wide (e.g. 38 days, 17-59; 19 days, 6-32; 19 days, 7-35; 43 days, 25-61) and some parameters were highly correlated, making their interpretation worthless.
Training improved 5-min all-out power by 30 ± 7%. A nonlinear model letting fatigue vary with training load fitted better than earlier models (standard error 6.47 ± 0.71 W vs 9.20-10.31 W) and implied an inverted-U relationship between daily training amount and performance. Very small, untrained sample.
Modelling then field testing in triathletes: an exponential taper improved cycle ergometry performance more than a step-reduction taper, and a fast exponential taper (time constant 4 days) more than a slow one (8 days); 5 km run results went the same way but were not significant.
Mobile power meters provide valid, reliable field measurements and are useful for monitoring training and race demands, but simple summary methods cannot reflect a session's variable power distribution or its metabolic stress, and current methods of quantifying training load are inadequate.
Subjective and objective measures generally did not correlate; subjective measures reflected acute and chronic training load with greater sensitivity and consistency. Subjective well-being was impaired by acute load increases and chronic training and improved with an acute decrease in load. Mixed sports and levels.
Get The Forward
Research-led triathlon guidance, straight to your inbox. No spam, no filler — unsubscribe any time.
