A multi-threshold measurement framework for evaluating fitness development across the full power-duration spectrum. Developed by Holden Comeau / Half Wheel.
The history of power-based fitness measurement in endurance sport is a history of increasing resolution. Each generation improved on the last by capturing more of the power-duration relationship. PTA is the next step in that progression.
Each step represents an expansion of resolution. CP reduces the power-duration relationship to two numbers. The Power Duration Model fits a continuous curve and extracts three parameters plus derived zones. PTA measures seven individually located transition points, each tracking a specific energy system, each shifting as the rider develops. The resolution increase from 2 to 3 to 7 is not arbitrary. It reflects the number of distinct bioenergetic transitions that are physiologically meaningful and detectable from performance data.
The Critical Power model was developed from laboratory time-to-exhaustion protocols: a rider sustains a fixed power output until failure, repeated at 3-5 intensities on separate days. This produces clean data but under artificial conditions, with 3-5 data points per athlete.
The Power Duration Model improved on this by using field data from training files. But training data carries an inherent limitation: it reflects the prescription, not the physiology. If a coach prescribes 4×8 minutes at 95% of FTP, the data will show the rider's response to that specific stimulus, not what they could have produced across the full duration spectrum under unconstrained effort.
The emergence of large-scale competitive cycling platforms created, for the first time in the history of exercise science, a dataset where hundreds of thousands of riders produce unconstrained maximal efforts across the full power-duration spectrum, multiple times per week, over years. PTA is a direct consequence of this data existing. It could not have been developed from laboratory protocols or training files alone.
PTA identifies seven bioenergetic transition thresholds: the specific durations where the power-duration curve inflects as one energy system depletes and the next takes over. These transition points are physiologically real, individually variable, and they shift as the rider develops.
The precise location of each threshold is not just a fingerprint. It is a second axis of development. Traditional power measurement asks one question: did the watts go up at this duration? PTA asks two: did the watts go up, and did the threshold itself shift?
Consider a rider whose PT-A (aerobic capacity) sits at 278 seconds producing 310 watts. Three months later, that same rider's PT-A has shifted to 294 seconds, still producing 310 watts. By a fixed-window measurement at 300 seconds, nothing changed. By PTA, that rider's aerobic system is now sustaining dominance 16 seconds longer before the next energy system becomes the limiter. That is a real physiological adaptation, invisible to conventional measurement.
The same power, sustained for longer, is progress. More power at the same offset is also progress. The most complete development shows both: the threshold shifts outward and the power at the threshold increases. Two independent signals from each of the seven PTA points gives fourteen dimensions of development from a single profile, versus one number from FTP.
PTA operates on the mean-max power curve: the best average power at every duration from a collection of rides. The curve is computed from 1Hz power data at variable-density resolution, with dense sampling in the bioenergetic transition zones where curve shape is most information-rich, and coarser sampling in the post-transition regions where the curve is smooth.
Transition points are identified as inflection points where the rate of power decline changes significantly, indicating one energy system's depletion and the next system's dominance. The detection algorithm locates each individual's specific offsets within the seven bioenergetic zones, producing a unique physiological fingerprint.
PTA value is the unweighted average watt change across all seven thresholds from Q1 best to Q4 best within a measurement period. It answers a question that no single metric can: across the full power-duration spectrum, did this rider get stronger or weaker?
A rider with a PTA value of +3.2W improved by an average of 3.2 watts at each of the seven transition points. A rider at -18.0W declined by an average of 18 watts at each. The composite captures whole-curve movement that FTP, CP, or any single-duration metric misses entirely.
PTA value measures the power dimension of development. Combined with the positional dimension (how each threshold's time offset is shifting), the full PTA profile produces up to fourteen independent development signals from a single assessment, compared to one from FTP.
Each rider's PT-N through PT-T positions and powers form a unique signature. Two riders at the same FTP may have completely different PTA profiles, reflecting different strengths, different energy system development, and different responses to training stimuli.
PTA detects whether a rider is expanding (improving at long-duration thresholds), sharpening (improving at short-duration thresholds), or developing uniformly across the curve. This informs training prescription with specificity that zone-based systems cannot provide.
When multiple power sources record the same effort, PTA curve shape comparison can detect measurement inconsistencies and establish trust scoring for data quality. The shape of the PTA curve is a more reliable indicator of consistency than absolute power values.
The analysis in The Racing Effect applies the PTA framework to population-level event participation data. The seven thresholds are approximated from the best available discrete power peaks per event. This simplified application is sufficient to detect macro patterns (cohort-level development rates, seasonal erosion, dose-response relationships) while the full-resolution PTA methodology from 1Hz data enables individual-level precision.
For clarity and consistency across RaceProof publications:
PTA refers to the methodology: Power Threshold Array.
PTA model refers to the analytical framework and detection algorithm.
PTA curve refers to the visualization of the seven threshold points and the power at each.
PTA value refers to the composite development metric (average watt change across all seven thresholds).
PT-N through PT-T refer to individual Power Thresholds. When referencing a single threshold, use PT (not PTA). The array is the collection; the individual point is a Power Threshold.