What Are CTL, ATL and TSB? Fitness, Fatigue and Form Explained
If you use a training app like TrainingPeaks, Strava or Yama, you've certainly run into these three acronyms: CTL, ATL and TSB. The model is widely used in digital training management — and is often interpreted more precisely than the underlying data allows.
The good news: the concept behind them is surprisingly simple. It comes down to three questions every athlete knows: How fit am I? How tired am I? And am I in form right now? In this article I'll explain all three metrics from the ground up — no sports science degree required, but with the real math behind them.
CTL = long-term weighted training load (a fitness proxy). ATL = short-term weighted training load (a fatigue proxy). TSB = the difference between CTL and ATL (training-load balance).
The foundation: what is TSS?
Before we get to the three metrics, we need a unit for training load. That unit is called TSS (Training Stress Score). It captures how demanding a single session was — from duration and intensity.
In the classic power-based cycling version, one hour at FTP is defined as 100 TSS. An easy one-hour session might produce roughly 50 TSS; duration and relative intensity determine the value together.
The formula above applies to power-based cycling using Normalized Power (NP), Intensity Factor (IF) and FTP. Depending on the platform, running may use pace-based rTSS, heart-rate-based hrTSS or another load model. These methods can produce different scores for the same workout, so a consistent data basis matters. If you want to go deeper, our article on the FTP test helps.
What is CTL? Your fitness
CTL stands for Chronic Training Load. Many platforms label it simply as “fitness”. More precisely, CTL is a model-based fitness proxy derived from training load.
By default, CTL is an exponentially weighted average of daily TSS with a 42-day time constant. It is not a hard 42-day window: older workouts continue to contribute, but with progressively less weight. TrainingPeaks therefore notes that CTL effectively reflects roughly three months of training.
A high CTL first tells you that you have accumulated substantial relative load over time. It does not automatically mean that you are faster or can tolerate any further load: the model does not know about sleep, illness, injury, technique or race specificity.
Sport, load formula and threshold settings can substantially change CTL. An incorrect FTP or threshold pace shifts the entire curve. Your own trend under consistent settings is more informative than comparing the absolute number with another athlete.
What is ATL? Your fatigue
ATL stands for Acute Training Load. Platforms often label it “fatigue”. This is also a proxy: ATL does not measure tiredness directly. It weights recent training load using a shorter default time constant of 7 days.
Because the time constant is short, ATL reacts more quickly than CTL. A hard training week pushes it up rapidly; rest days bring it down rapidly. Whether you actually feel tired must be checked separately — the model cannot measure your experience.
On its own, ATL says little. It only becomes meaningful in combination with CTL — and that's exactly what TSB is.
What 42 and 7 really mean mathematically
These numbers are time constants, not cutoff windows. On a day with no new TSS, standard CTL retains 41/42 of its previous value, while ATL retains only 6/7. Short-term load therefore falls much faster than long-term load during recovery.
| Metric | Time constant | Remaining after 7 days with zero TSS | Model half-life |
|---|---|---|---|
| CTL | 42 days | 84.5% | 28.8 days |
| ATL | 7 days | 34.0% | 4.5 days |
Example: if CTL and ATL both start at 60 and seven days record zero TSS, CTL is still 50.7 in the model, while ATL has fallen to 20.4. TSB consequently turns positive. This describes the model only; it does not mean biological fitness and fatigue decay with identical half-lives in every athlete.
What is TSB? Your form
TSB stands for Training Stress Balance and is a useful number for race planning. In the TrainingPeaks convention, it is calculated as follows:
Fitness minus fatigue = form
A negative TSB means short-term weighted load is higher than long-term weighted load; a positive number means the opposite. TrainingPeaks explicitly notes that TSB is not a performance predictor. It describes modelled adaptation to recorded load, not guaranteed race-day form.
What the TSB numbers mean
| TSB range | State | Meaning |
|---|---|---|
| Positive | Relatively fresh | ATL is below CTL; recent load has fallen relative to the long-term baseline |
| Near zero | Balanced | ATL and CTL are at similar levels |
| Negative | More recently loaded | ATL is above CTL; common during build phases |
| Unusually negative for you | Review | Check against symptoms, sleep, recovery and the intended training stimulus |
A positive TSB does not imply exceptional fitness; mathematically, it comes from ATL being lower than CTL. A taper reduces training load before competition. Meta-analyses commonly support a substantial reduction in volume while largely maintaining intensity, but the appropriate size and duration remain individual. To plan your taper in practice, use the taper calculator.
Why is my form dropping? What a negative TSB means
Many people come across TSB because their app shows a falling form curve — often exactly when they are training hardest. That is not a contradiction but how the model works: every hard week pushes the fast-reacting ATL up more than the slow-moving CTL. The difference, your form, turns negative.
A dropping form is therefore usually intended during build phases. It shows that you are currently training more than you have been used to over recent weeks — which is exactly the stimulus your body adapts to. Freshness only returns once you reduce the load, for example in a recovery week or during your taper.
When a negative TSB is a warning sign
The model cannot tell whether you are productively tired or overloaded. Both look the same on the chart. The difference shows up in signals that TSS does not capture:
- Performance stalls or declines even though you have recovered in between.
- Resting heart rate and HRV deviate clearly from your personal baseline for several days.
- Sleep, motivation and mood noticeably get worse.
- Complaints such as persistent muscle or joint pain appear.
Short-term overload that disappears after a few days of recovery is part of hard training. If the drop in performance persists for weeks despite recovery, this is referred to as overtraining — a medical issue that should be assessed by a doctor, not diagnosed from a form curve. Our article on using HRV for training explains how to interpret resting heart rate and HRV; the recovery calculator estimates how long a single session keeps affecting you.
A real-world example over 12 weeks
Let's say you're preparing for a half marathon. Here's how the three values might develop over a typical training cycle:
| Phase | CTL | ATL | TSB | Feeling |
|---|---|---|---|---|
| Week 1 (start) | 45 | 45 | 0 | Neutral |
| Week 4 (build) | 58 | 70 | −12 | Tired, productive |
| Week 8 (peak block) | 72 | 88 | −16 | Really cooked |
| Week 11 (taper) | 70 | 55 | +15 | Coming fresh |
| Week 12 (race day) | 68 | 48 | +20 | Modelled fresh |
Note race week: CTL has only dropped slightly (from 72 to 68 — you've barely lost fitness), but ATL has fallen sharply (from 88 to 48 — the fatigue is gone). Exactly this combination produces a positive TSB of +20. Whether that becomes a strong race performance also depends on specificity, health, pacing and conditions on the day.
The most common misconceptions
"High TSB is always good"
A high TSB initially means only that short-term load is below long-term load. That may fit a taper, but it can also result from interrupted training. Without context, the number means neither “peak form” nor “too little training”.
"I need to keep raising my CTL"
The change in CTL is often called ramp rate. There is no universal boundary below which training is automatically safe and above which it is automatically dangerous. Interpret the increase relative to your history and alongside symptoms, sleep, perceived recovery and workout quality.
"The numbers are directly comparable"
No. A CTL of 70 means something different for a runner than for a cyclist, because TSS accumulates differently per sport. Always compare your values only with yourself over time, never with other athletes.
CTL, ATL and TSB are only as good as your TSS — and TSS is only as good as your threshold values (FTP, threshold pace, threshold heart rate). If your FTP is outdated, all three metrics are skewed. Keep your thresholds current: you can calculate your FTP here, and how to determine the lactate threshold without a lab is covered in our article on the lactate threshold.
Fitness, fatigue and form in TrainingPeaks and Strava
Many people first see these values in their training app — just under different names. The mapping is simpler than the labels suggest:
| Model term | TrainingPeaks | Strava |
|---|---|---|
| CTL | Fitness (CTL) | Fitness |
| ATL | Fatigue (ATL) | Fatigue |
| TSB | Form (TSB) | Form |
| Chart | Performance Management Chart | Fitness & Freshness |
In Strava, the chart is called Fitness & Freshness and requires a subscription. Strava describes its method as an impulse-response model — the same basic idea as CTL and ATL. As its load measure, however, it uses Training Load from power data and, where power is missing, Relative Effort based on heart rate or perceived exertion.
That is why numbers from Strava and TrainingPeaks are not directly comparable, even though they share the same names. A form of −19 in Strava and −19 in TrainingPeaks do not describe the same state. Only compare your values within the same platform and against your own history.
Why these metrics aren't everything
CTL/ATL/TSB are powerful, but they have one weakness: they're distance-agnostic. A positive TSB tells you you're fresh — but not whether you have the specific endurance for your target distance. For a marathon, what counts isn't just freshness but also whether you have enough long runs in your legs.
That's exactly the gap Race Shape fills — a metric that additionally accounts for volume and distance specificity. How that works is explained in our article on Race Shape.
Yama shows all three live
Connect your training data with Yama — CTL, ATL and TSB are computed automatically from your activities and displayed as a performance management chart. See at a glance how your training-load balance changes. The curve complements your plan and how you feel; it does not automatically decide when you need rest.
Try YamaConclusion
Three acronyms, one simple principle: CTL weights load over the longer term, ATL over the shorter term, and TSB is their difference. Fitness, fatigue and form are useful shorthand labels, not direct physiological measurements.
Anyone who understands these three can steer their training deliberately: ramp up fitness during build, reduce recent load before a race, and review unusual trends in time. The concept is simple — the discipline to stick to the plan even in the tired phase of a training block is the harder part.
One last bit of perspective: CTL, ATL and TSB are a model, not a law of nature. They reflect how much you've trained — not how well you've slept, eaten or recovered. Take the numbers seriously, but not as the whole truth.
Sources and calculations
The equations shown here follow the TrainingPeaks default definitions. The values 84.5%, 34.0% and the model half-lives are calculated directly from those recurrence equations. They are mathematical properties of the model, not measured biological constants.
- TrainingPeaks: Fitness (CTL) — definition, 42-day time constant and formula.
- TrainingPeaks: Fatigue (ATL) — definition, 7-day time constant and formula.
- TrainingPeaks: Form (TSB) — previous-day calculation and the note that TSB is not a performance predictor.
- TrainingPeaks: Training Stress Scores — distinctions among TSS, rTSS, sTSS and hrTSS.
- Morton, Fitz-Clarke & Banister (1990) — the fitness-fatigue model and exponential decay of modelled responses.
- Wang et al. (2023) — systematic review and meta-analysis of tapering in endurance athletes.
- Strava Support: Fitness & Freshness — labels, subscription requirement and calculation basis (Training Load and Relative Effort).