CTL, ATL and TSB Explained: Fitness, Fatigue and Form for Triathletes

CTL, ATL and TSB are the three lines on the Performance Management Chart in TrainingPeaks and similar platforms. They look like a verdict on your fitness. In reality they are three running averages of one number—your daily training stress—and once you see the arithmetic, the chart becomes far more useful and far less mysterious.

Everything on the chart starts with a daily load score. In most triathlon software that is Training Stress Score (TSS), where one hour at threshold equals 100. If the TSS going in is wrong—because of an outdated FTP, a guessed swim threshold or a heart-rate estimate—every line built from it inherits the error. Keep that in mind throughout.

What the Performance Management Chart actually shows

The chart is a practical version of the fitness–fatigue (impulse–response) model proposed by Eric Banister and colleagues in the 1970s. The idea is simple: every workout produces a small, long-lasting positive effect and a larger, short-lived negative effect. Performance at any moment reflects the balance between the two. Andrew Coggan adapted the idea to power-based TSS, and TrainingPeaks popularised it as the Performance Management Chart (PMC).

Strava shows a similar model as “Fitness & Freshness”. Other platforms may use different load inputs, time constants and names, so numbers are rarely comparable between apps.

How CTL, ATL and TSB are calculated

TrainingPeaks documents the same update rule for both averages. Each day, the line moves a fixed fraction of the way towards that day’s TSS:

Because each day only moves the line a fraction of the gap, older workouts never fully disappear—their influence just shrinks. That is why TrainingPeaks notes that CTL in effect reflects about three months of training, even though its time constant is six weeks.

What one workout does to the chart

Take a single 100-TSS session for an athlete starting at zero. CTL rises by 100 × 1/42 = 2.4. ATL rises by 100 × 1/7 = 14.3. The next morning, TSB is 2.4 − 14.3 = −11.9. One hard day can make form look poor, but it barely changes fitness.

How long changes take to show up

If you change your training and then hold the new load steady, the lines close the gap at very different speeds:

Calculated from the default 42- and 7-day constants (1 − (1 − 1/k)^days). ATL at 42 and 90 days is above 99.8% and rounds to 100%.

The most useful rule: CTL ≈ your average daily TSS

If you trained with exactly the same daily TSS for long enough, CTL would settle at that number. In other words, CTL is a smoothed estimate of how much TSS you produce per day. Multiply it by seven and you get the weekly load it represents. That single fact answers most “is my CTL good?” questions better than any league table.

To translate weekly TSS into hours, you need to know how much TSS each hour produces. For power-based cycling, TSS per hour equals 100 × IF². Steady endurance riding at IF 0.71 gives about 50 TSS per hour, mixed training at IF 0.77 about 60, and harder sessions at IF 0.84 about 70.

Arithmetic conversion, not a population benchmark. Running and swimming scores per hour can differ from cycling, so real triathlon weeks will not match a single column exactly.

So what is a good CTL?

There is no universal good number. A CTL of 60 simply says that you have recently sustained something like 420 TSS a week—around six to eight hours of typical training. Whether that is enough depends on your race distance, your history and, above all, whether you can repeat it without breaking down. A better question is: what weekly hours can I sustain, and what CTL does that produce? Our guide to how many hours a week to train for a triathlon is a sensible starting point; CTL then becomes a way to check that you are actually delivering that volume.

Be wary of CTL “targets” for each race distance. Numbers quoted online come from different athletes, platforms and threshold settings, and they are not validated standards. Two athletes with the same CTL can be very differently prepared.

A 70.3 build, modelled day by day

To show how the lines behave in a real plan, we modelled a self-coached half-distance athlete. The assumptions are deliberately ordinary:

• Starting CTL and ATL of 45 (about 315 TSS a week).
• First week at 330 TSS, adding 30 TSS each build week, with every fourth week a recovery week at 65% of the previous week.
• 15 weeks of building, peaking at 660 TSS—roughly 11 hours at 60 TSS per hour.
• A two-week taper at 70% and then 45% of peak load, with race day on the final Sunday.
• A fixed weekly pattern: rest Monday, key sessions Tuesday and Thursday, the long ride on Saturday and the long run on Sunday.

Four things stand out:

1. CTL climbs slowly. Fifteen weeks of progressive training move CTL from 45 to a peak of 73.8. The biggest weekly increase is about 4 points.
2. TSB is negative for almost the entire build. At the end of the biggest week it reaches −27. That is what training looks like in this model, not a warning by itself.
3. Recovery weeks reset TSB, not CTL. Each one pulls TSB back near zero while costing only about 1.5 points of CTL.
4. The taper trades a little CTL for a lot of form. CTL drops 5.3 points from its peak, but TSB swings from −27 to +8.6 on race morning.

Try it with your own numbers

Change the inputs to see how starting load, progression, recovery weeks and taper length shape the chart. The simulator uses exactly the same model and weekly pattern as the example above.

How to read TSB (form)

TSB is the line athletes most often misread. TrainingPeaks itself describes it as a measure of how adapted you are to your current load, not as a predictor of performance. A useful way to read it is by context:

Notice how much TSB moves within a single week. A Monday morning after a big weekend will always look worse than a Saturday morning. Compare like with like—for example, every Monday—rather than reacting to daily swings.

Ramp rate: how fast should CTL rise?

Ramp rate is the weekly change in CTL. You can predict it with one line of arithmetic. Over seven days at a steady load, CTL closes about 15.5% of the gap between its current value and your average daily TSS:

Daily-step calculation with the default 42-day constant and evenly spread load. Uneven weeks change the result by a few tenths.

The table makes a point that is easy to miss: a ramp of 5 CTL points in one week requires a weekly load roughly 50% above what your CTL represents. Sustaining that for several weeks in a row is a very large progression for most age-group triathletes.

Coaching rules of thumb commonly quote something like 3–7 CTL points per week. Treat these as conventions rather than safety limits. Research on load ratios has not shown that any particular threshold protects you: a widely cited critique of the acute:chronic workload ratio concluded there was no evidence supporting its use to reduce injury risk. Progress load gradually, but let your response—not a target ramp—decide the next step.

Tapering with TSB

The chart is most useful before a race because it shows the trade-off a taper creates. The table below starts from an athlete holding a steady CTL and ATL of 80 (TSB 0) and applies different tapers:

Model outputs with default constants. Real athletes rarely hold perfectly steady load, so their starting TSB will differ.

The model always rewards more rest with a higher TSB—complete rest produces the biggest number of all. It has no way to tell you when freshness becomes staleness, or how much CTL you can afford to lose. That judgement has to come from evidence and experience. A 2023 meta-analysis of endurance athletes found that tapers which cut training volume by roughly 41–60% while keeping intensity and frequency improved time-trial performance. Our triathlon taper guide turns that into practical taper lengths by race distance.

CTL, ATL and TSB in triathlon: three sports, one number

Most platforms add swim, bike and run stress into one combined CTL. That is useful for total workload, but it hides several problems that are specific to triathlon:

• Different scores are calibrated differently. Power-based cycling TSS, pace-based running scores and swim scores use different thresholds and assumptions. A running score depends on an accurate threshold pace; a swim score on a realistic critical swim speed.
• Equal scores are not equal stress. Running is weight-bearing and usually costs more musculoskeletal recovery than the same score on the bike or in the pool.
• Some work is invisible. Strength sessions, unrecorded swims and heart-rate estimates on sessions without power or pace can make the chart overstate or understate your real load.
• Thresholds drift. An FTP that is too low inflates every cycling score. Retest with a consistent FTP test when sessions repeatedly stop matching the numbers.

Check the per-sport lines as well as the combined one. A combined CTL that keeps rising because of extra cycling can hide a run CTL that has stalled—or a run load that is climbing faster than your legs can handle. Our triathlon training zones guide explains how to set consistent thresholds across all three sports.

What the model cannot tell you

The fitness–fatigue model is elegant, but it is a simplification. Research on it points to a few important limits:

• The constants are not personal. When researchers fitted the Banister model to elite swimmers, the individually estimated time constants had very wide confidence intervals, and some parameters were so strongly correlated that they could not be interpreted on their own. The default 42 and 7 days are convenient conventions, not measured properties of your body.
• Different load inputs give different answers. A study comparing training-load quantification methods found that the fitness–fatigue model’s results depended on how training load was quantified.
• It knows only what you upload. Sleep, illness, heat, travel, nutrition and life stress all change how you absorb the same TSS.
• “Fitness” is a label. CTL measures sustained training load. Fitness—your FTP, threshold pace, durability or race result—has to be measured separately.

A 2022 review of fitness–fatigue models in sport also highlighted conceptual issues with how these models represent fitness and fatigue. Our practical reading: the chart is a useful framework for thinking about load, not a precise forecast of how you will perform on a given day.

A five-minute weekly PMC check

1. Check the inputs first. Were all sessions recorded, and are FTP, threshold pace and swim threshold still current?
2. Look at the CTL trend over four to six weeks. Is it rising, flat or falling, and is that what the plan intended?
3. Compare this week’s TSB with the same day last week. Ignore day-to-day swings; look for a pattern of deepening fatigue across several weeks.
4. Look at the per-sport lines. Make sure one discipline is not driving the total while another quietly falls away.
5. Compare the chart with how you feel and perform. If key sessions, sleep and motivation are all deteriorating, reduce load—whatever the chart says.

Frequently asked questions