Guide · Banister model

Fitness, fatigue, form.

Q: Is CTL really a measure of fitness?

It's a proxy — the best single-number proxy we have from training data alone, but a proxy. CTL measures the rolling average of training stress you've absorbed. Actual fitness also depends on sleep, nutrition, genetics, and how smartly you trained. Two athletes at CTL 90 can race very differently. Treat CTL as the floor of your fitness, not a benchmark against other people.

Q: What's a safe CTL ramp rate?

Rule of thumb: 3.5 CTL points per week for experienced athletes in a build block, 2.5 for returning or beginner athletes. Above 5 per week for more than a week or two is explicit injury territory. In practice, ramps are never smooth — expect 4–5 point weeks followed by a deload week that holds or drops CTL. That zigzag is healthy.

Q: Can I race at negative TSB?

You can, but you shouldn't for A-races. Most athletes race fastest between TSB +5 and +20. Going into a race below TSB −5 means you haven't shed enough fatigue; going above +25 usually means you've cut volume so hard you've lost sharpness. For B-races and tune-ups, TSB 0 to +10 is fine. For an A-race, aim for +15.

Q: My CTL dropped during a vacation — should I panic?

No. Two weeks fully off drops CTL roughly 15–20%. A week off drops it 8–12%. The fitness itself doesn't vanish that fast — detraining studies show aerobic capacity is preserved for 2–3 weeks with only minor losses. CTL overstates the short-term fall. When you come back, the first 10 days of training rebuild most of what the number dropped.

Q: How long until CTL is accurate for a new user?

Meaningful: about 6 weeks of consistent training history loaded. Stable: 4–5× the 42-day time constant, so roughly 6 months. We warm up CTL with 126 days of data before the visible chart, so if you've been on Strava 6+ months your CTL is honest on day one. If you've just connected, we flag lower confidence in the weekly review for the first month.

Three numbers on one chart — and every coaching platform uses them. Here’s what they mean, how they’re calculated, and what ranges signal what.

The three numbers

CTL — Chronic Training Load (“fitness”)

42-day exponentially weighted moving average of daily TSS. Climbs slowly as you train consistently. A well-trained amateur cyclist sits at CTL 60–90. A cat 2 racer at 90–120. An off-season amateur drops to 30–40.

ATL — Acute Training Load (“fatigue”)

7-day EWMA of TSS. Responds fast. A single 200-TSS ride shifts ATL by 30+ points the next day. Recovers in 3–5 days.

TSB — Training Stress Balance (“form”)

CTL minus ATL. Positive = fresh, negative = loaded. This is the number you check before a race.

TSB bands — what each range means

TSB > +25 — Stale / overtapered

You’re too fresh. If this is mid-season, you’re undertraining. If this is race week, you’ve cut too much.

TSB +5 to +25 — Fresh (good for races)

Peak performance window. Most athletes race fastest at TSB around +15. Cut volume, keep some intensity.

TSB −5 to +5 — Fit (productive training)

The sweet spot for building. You’re absorbing the load and adapting. CTL should be climbing slowly.

TSB −10 to −30 — Productively overloaded

Short-term block of hard training. Fine for 2–3 weeks if you follow with a deload. Don’t race in this band.

TSB < −30 for 14+ days — Overreached

Injury and illness risk climbs sharply. Our rule engine flags this as NEGATIVE_TSB_PROLONGED and will force a recovery block.

Safe ramp rates

Experienced amateur: CTL should climb no faster than 3.5 points per week. Beginner or returning: 2.5 per week. Faster than 5 per week is explicit injury risk and shouldn’t be prescribed.

Real ramps are rarely smooth — you’ll see 2-week build steps of 4–5 points, then a deload week where CTL plateaus or drops by 2. That’s healthy.

How we compute it

Both CTL and ATL are standard single-pole EWMAs with time constants 42 and 7 days:

CTL(today) = CTL(yesterday) × e^−1/42 + TSS_today × (1 − e^−1/42)

To give CTL a stable starting point we run a 126-day warmup window (5 × time-constant rule) before the chart you see. If you’ve been on Strava 6+ months, your CTL is real. If you’ve just connected, we show you the last 60 days and flag lower confidence in the review.

Worked example — a 12-week build into a race

Amateur cyclist, FTP 260 W, starting CTL 55 in early January, target gran fondo in late March. Here’s the curve, week-by-week:

Weeks 1–3 — Re-entry (+2.5 CTL/wk)

380, 420, 440 weekly TSS. CTL climbs 55 → 57 → 60 → 62. ATL around 55–65. TSB slightly negative (−3 to −7). Legs feel normal, no alarms. This is the “get your tax paid” phase.

Weeks 4–6 — Build (+3.5 CTL/wk)

490, 520, 480 TSS. CTL 62 → 66 → 70 → 72. ATL peaks around 85 after the 520-TSS week. TSB dips to −13 for a couple of days then rebounds. Monotony stays under 1.4 because the long day still dominates.

Week 7 — Deload (CTL plateaus)

310 TSS. CTL holds at 72 (a deload week should hold CTL, not crash it). ATL drops to 60. TSB climbs to +12 by Sunday — the athlete does a tune-up event feeling sharp.

Weeks 8–10 — Peak build (+3/wk)

520, 560, 540 TSS. CTL 72 → 75 → 78 → 81. The athlete is flirting with their ceiling. TSB runs −10 to −18. The rule engine doesn’t flag overreach because monotony stays healthy and no 7-day ATL spike exceeds 40% of CTL.

Week 11 — Taper start

360 TSS. Same intensity, 30% less volume. CTL nudges down to 80. ATL drops to 50. TSB climbs through 0 into +10. Legs start feeling snappy.

Week 12 — Race week

200 TSS through Saturday (race). CTL 78, ATL 42, TSB +36 on race morning. Slightly overtapered — classical literature says aim for +20 to +25, so this athlete could have kept a little more volume. Race still goes well — CTL at 78 is what actually produced the capacity; TSB just decides how fresh that capacity arrives.

Takeaways from the curve

Peak CTL of 81 from start of 55 in 10 weeks = 2.6 CTL/week average. Safe, and shows the ramp isn’t linear — big weeks plus deloads.
Best performance comes from the combination of CTL and TSB, not either alone. High CTL with −20 TSB is broken. Low CTL with +20 TSB is hollow.
The deload in week 7 was what made weeks 8–10 possible. Skip it and weeks 8–10 turn into injury weeks.

Common mistakes

Chasing CTL as a goal.CTL is a byproduct of smart training, not a target. “Hit CTL 100 by June” turns into skipping rest days, inflating monotony, and arriving at June injured. Let the plan prescribe weekly stress; CTL follows.
Racing on negative TSB because the CTL is high. Form trumps fitness on race day. You won’t access the CTL you built if you show up at TSB −15. The taper exists for a reason.
Ignoring monotony because CTL looks healthy. You can have great CTL and toxic monotony at the same time — the classical signature of “grey zone” training. Check both every week.
Confusing ATL drop with lost fitness.After a rest week, ATL falls 30+ points and TSB jumps positive. Some athletes interpret the fresh legs as “losing fitness, need to add training” and crush the first week back, re-loading an already-recovered body. CTL is what matters; ATL is noise around it.
Comparing your CTL to a pro’s.A cat 1 racer at CTL 110 is training 18 h/wk with recovery infrastructure. Your CTL 75 training 8 h/wk around a full-time job is not a lesser version of their training — it’s a different sport.
Resetting the chart after a break.After two weeks off, your CTL will show accurate detraining. Resist the urge to zero it out or manually spike it back up. Train the ramp honestly from wherever CTL lands — you’ll be back faster than you think.

Related metrics

TSS

The daily input. CTL and ATL are both EWMAs of TSS — without accurate TSS, every downstream number inherits the error. Calibrate your FTP and threshold pace so TSS is honest, and CTL follows. What is TSS →

Monotony and strain

CTL hides distribution. Two athletes can have identical CTL 80 but radically different monotony. The Foster monotony metric catches what CTL can’t — whether your 80 came from rigorous hard-easy rhythm or grey-zone slog. Training monotony →

Training readiness

A daily 0–100 score that combines TSB (the form bit of the classical model) with monotony and recovery since last quality. When you open the app on a Tuesday morning, readiness is the number that answers “should I push today?” — CTL and TSB answer “am I on track for race day?”. Training readiness →

The original Banister impulse-response model

The CTL/ATL/TSB framework is a simplified version of Eric Banister’s 1975 fitness-fatigue model. Banister used two exponential decays with opposing signs — positive “fitness” with a long time constant, negative “fatigue” with a short one. Performance equalled fitness minus a weighted fatigue. The modern TrainingPeaks / Coggan PMC is the same math with coaching-friendly labels.

Common questions

FAQ

Is CTL really a measure of fitness?

It's a proxy — the best single-number proxy we have from training data alone, but a proxy. CTL measures the rolling average of training stress you've absorbed. Actual fitness also depends on sleep, nutrition, genetics, and how smartly you trained. Two athletes at CTL 90 can race very differently. Treat CTL as the floor of your fitness, not a benchmark against other people.

What's a safe CTL ramp rate?

Rule of thumb: 3.5 CTL points per week for experienced athletes in a build block, 2.5 for returning or beginner athletes. Above 5 per week for more than a week or two is explicit injury territory. In practice, ramps are never smooth — expect 4–5 point weeks followed by a deload week that holds or drops CTL. That zigzag is healthy.

Can I race at negative TSB?

You can, but you shouldn't for A-races. Most athletes race fastest between TSB +5 and +20. Going into a race below TSB −5 means you haven't shed enough fatigue; going above +25 usually means you've cut volume so hard you've lost sharpness. For B-races and tune-ups, TSB 0 to +10 is fine. For an A-race, aim for +15.

My CTL dropped during a vacation — should I panic?

No. Two weeks fully off drops CTL roughly 15–20%. A week off drops it 8–12%. The fitness itself doesn't vanish that fast — detraining studies show aerobic capacity is preserved for 2–3 weeks with only minor losses. CTL overstates the short-term fall. When you come back, the first 10 days of training rebuild most of what the number dropped.

How long until CTL is accurate for a new user?

Meaningful: about 6 weeks of consistent training history loaded. Stable: 4–5× the 42-day time constant, so roughly 6 months. We warm up CTL with 126 days of data before the visible chart, so if you've been on Strava 6+ months your CTL is honest on day one. If you've just connected, we flag lower confidence in the weekly review for the first month.

Keep reading

What is TSS→

The building block CTL and ATL are computed from.

Training monotony→

Why CTL alone doesn’t tell the full injury risk story.

Training readiness→

How we combine TSB and other signals into one daily score.

Weekly review→

How CTL and TSB drive the theme of every next block.

For cyclists→

CTL ramps specific to power-based cycling progression.

For runners→

Running CTL, long-run progression, and injury caps.

See your own curves.

Connect Strava and we chart 60 days of CTL, ATL, and TSB for you.