Cycling

FTP is a population average. It only matches reality at E ≈ 0.70.

Functional Threshold Power assumes all riders can sustain their lactate threshold for approximately 60 minutes. This is wrong — threshold duration depends on endurance. TrueZone replaces FTP with Lactate Threshold Power (LTP), an individualized metric that separates threshold intensity from threshold duration.

297
Cyclists validated
13,887
Sessions analyzed
6.0 bpm
Median MAE
99.5%
Session success rate
42 avg
Sessions per cyclist

The Problem

FTP assumes everyone's threshold lasts 60 minutes. It doesn't.

Functional Threshold Power is defined as the highest power a rider can sustain for approximately one hour. But how long a rider can actually sustain their lactate threshold depends on their aerobic endurance — and that varies enormously between individuals. FTP conflates two independent physiological qualities: threshold intensity and threshold duration.

FTP overestimates low-endurance riders

38 cyclists with E < 0.65 cannot sustain their lactate threshold for 60 minutes. FTP assumes they can, inflating their threshold estimate and setting zones too high.

FTP underestimates high-endurance riders

186 cyclists with E > 0.75 can sustain threshold well beyond 60 minutes. FTP caps their capacity, setting zones too low and leaving performance on the table.

FTP is only correct at E ≈ 0.70

The 60-minute assumption matches reality for riders near the population mean endurance. For everyone else — the majority — it is systematically wrong.

The Solution

Lactate Threshold Power is individualized.

TrueZone separates threshold intensity from threshold duration. LTP is the power at your actual lactate threshold. How long you can sustain it depends on your endurance (E). The model derives both from ordinary ride data — no 20-minute test, no ramp protocol, no calibration ride.

FTP gets two things wrong, not one. First, the threshold power itself (because duration is assumed, not measured). Second, the zone boundaries around it. FTP-based zones use fixed percentages of FTP — the same ratios for every rider. But endurance shifts where thresholds actually fall relative to each other. A high-endurance rider has their aerobic zones compressed closer to threshold; a low-endurance rider has them spread further apart. TrueZone derives zone boundaries geometrically from each individual's threshold alignment, not from population averages.

LTP sustainable duration by endurance

E = 0.40Low endurance
~30 min
E = 0.55Below average
~42 min
E = 0.70Population average
~60 min
E = 0.80Above average
~75 min
E = 0.90High endurance
~95 min

FTP's fixed 60-minute assumption is only correct at E ≈ 0.70. For the majority of cyclists, it is systematically biased. LTP eliminates this bias by modeling threshold intensity and duration independently.

FTP vs LTP across the endurance spectrum

204060801001200.40.50.60.70.80.9Endurance (E)LTP duration (min)FTP overestimatescan't sustain 60 minFTP underestimatessustains well beyond 60 minFTP assumes 60 min
Actual LTP duration
FTP assumption (60 min)

Validation

297 cyclists. 13,887 sessions. Median MAE 6.0 bpm.

Validated on the GoldenCheetah OpenData dataset — real-world cycling data from a diverse population of riders. The model achieved a 99.5% session success rate, with 71% of riders reaching full parameter convergence.

0.771

Mean endurance (E)

The GoldenCheetah population skews toward experienced cyclists. Mean E well above the 0.70 population average confirms that FTP systematically underestimates their threshold durability.

456 W

Mean peak aerobic power

Peak aerobic power (~VO₂max intensity) derived from the model. Sprint power is estimated separately using an endurance-dependent multiplier.

6.0 bpm

Median MAE

Median absolute error in heart rate prediction across all fitted sessions. Comparable to running validation and sufficient for accurate zone and threshold derivation.

Bayesian Learning

Parameters converge from ordinary rides.

No FTP test required. Each ride refines the model's estimate of the rider's endurance. The confidence interval narrows as evidence accumulates, stabilising within 3–10 rides.

Cyclist — 49 rides, GoldenCheetah dataset
0.750.800.850.900.951.0011121314149SessionEndurance (E)
Bayesian convergence
Confidence interval

Interactive

Power-duration explorer

Set your lactate threshold power and endurance to explore how the model predicts sustainable power across durations. See how LTP and FTP diverge as endurance moves away from the population average.

Lactate Threshold Power
250 W
Endurance
70%

LTP vs FTP

250 W

LTP (Lactate Threshold)

Sustainable for 1:00:08

250 W

FTP (60-minute power)

Sustainable for exactly 60 min

349 W

VO₂max power

~5 min sustainable

1349 W

Sprint power (est.)

3.0× aerobic peak

At E ≈ 0.70, LTP and FTP are approximately equal — this is the population average where the 60-minute assumption holds. LTP duration: 60 minutes.

Power-duration curve

Predicted sustainable power at key durations, derived from your lactate threshold power and endurance. Power values account for aerodynamic and rolling resistance physics.

DurationPower% of LTP
5 min445 W178%
10 min382 W153%
20 min327 W131%
30 min297 W119%
60 min(FTP)250 W100%
90 min225 W90%
2 hr209 W84%
3 hr187 W75%

Power thresholds

Exercise thresholds reflect transitions between muscle fiber recruitment patterns. T2 is the lactate threshold — the boundary between sustainable and unsustainable intensity.

ThresholdDescriptionPower
T1Endurance threshold (LT1)114 W
T1.5Aerobic threshold173 W
T2 (LTP)Lactate threshold250 W
T2.5VO₂max threshold349 W
SprintNeuromuscular peak1349 W

Replace estimated FTP with measured LTP.

Give your cyclists individualized threshold power and endurance-adjusted zones. TrueZone is SDK and API ready today.