PSS, PI and familiar power metrics

By EffortSignal Editorial Team · Reviewed September 20, 2026

If you know TSS®, IF® and NP® from TrainingPeaks, you already know the core calculations behind PSS, Power Intensity and Weighted Power in EffortSignal. The names are different; the formulas and what these power metrics describe are the same. With the same calculation inputs, they give the same results.

Why we use different names

TrainingPeaks lists Training Stress Score® (TSS®), Intensity Factor® (IF®) and Normalized Power® (NP®) among its registered trademarks. That is why EffortSignal uses its own descriptive names for these metrics. The naming choice does not introduce a different scoring scale or require you to relearn the underlying concepts. TrainingPeaks trademark statement

We use the TrainingPeaks names here to explain the correspondence. This comparison does not imply affiliation with or endorsement by TrainingPeaks.

The names side by side

This comparison covers power-based cycling metrics. Power Variability is included because it belongs to the same calculation family; the trademark statement above concerns NP®, IF® and TSS®.

You do not need a conversion factor between the corresponding metrics. The equivalence follows directly from their definitions.

Weighted Power and NP®: the same weighting

Both calculations start with a 30-second rolling average of power. Each rolling average is raised to the fourth power; those values are averaged, and the fourth root gives the result in watts. This is the procedure described in TrainingPeaks’ explanation of Normalized Power®.

If M represents each complete 30-second mean, the common formula is:

Weighted Power = (mean(M⁴))^(1/4)
NP®            = (mean(M⁴))^(1/4)

When both calculations use the same sequence of rolling means, Weighted Power = NP®. The fourth-power weighting makes surges contribute more strongly than they do to average watts. Both metrics serve that same purpose; neither is a direct measurement of metabolic energy expenditure.

EffortSignal’s Weighted Power guide explains which samples and windows enter its calculation.

PI and IF®: the same relative intensity

EffortSignal divides Weighted Power by the applicable FTP. TrainingPeaks defines IF® as NP® divided by threshold power. TrainingPeaks’ definition of Intensity Factor®

PI = Weighted Power / FTP
IF® = NP® / FTP

With the same weighted power and FTP, PI = IF®. Both are unitless ratios: 200 W divided by an FTP of 250 W is 0.80, or 80%. This is weighted power relative to FTP, not average power relative to FTP or the share of a ride spent in a training zone.

PSS and TSS®: algebraically identical

TrainingPeaks publishes the following power-based TSS® equation, where t is duration in seconds. TrainingPeaks’ TSS® formula

TSS® = (t × NP® × IF®) / (FTP × 3600) × 100

Since NP® / FTP = IF®, this simplifies to:

TSS® = (t / 3600) × IF®² × 100
PSS = (t / 3600) × PI² × 100

That second line is EffortSignal’s PSS formula. Given the same duration, weighted power and FTP, PSS = TSS®. This is an algebraic identity, not an approximate conversion.

One hour at threshold intensity gives 100 points in either formula. One hour at 0.80 gives 64 points; half an hour at 0.80 gives 32 points. At a fixed intensity, doubling duration doubles the score. At a fixed duration, intensity contributes squared. These are the same units and the same interpretation of power-based load.

Power Variability and VI: the same ratio

Both divide weighted power by average power. TrainingPeaks calls this ratio Variability Index. TrainingPeaks’ VI definition

PV = Weighted Power / average power
VI = NP® / average power

With matching inputs, PV = VI. A steady constant-power recording gives 1.00. A larger ratio describes a greater difference between weighted and average power. FTP is absent from this calculation, so changing only FTP does not change it.

One ride, both sets of names

Consider an uninterrupted one-hour ride at a constant 200 W, with complete power data and FTP set to 250 W in both calculations. Every 30-second mean is 200 W.

There is no rescaling. The table applies both sets of equations to the same inputs; it is a worked example, not a comparison of uploads to two services.

Why an uploaded ride can still show different numbers

Identical formulas produce identical results when their inputs match. A file must first be turned into a power sequence and a duration. Differences at that stage can change the answer even when the final equation is identical.

When comparing a ride, check these inputs:

• FTP: use the same threshold applicable to that activity, including its date and indoor/outdoor setting. At 200 W for one hour, FTP 250 W gives 64 points; FTP 200 W gives 100.
• Duration and pauses: EffortSignal forms its rolling windows on timer-running time, removing pauses. PSS uses the positive recorded timer/moving duration, with projected timer duration as fallback. Counting elapsed time instead changes the score.
• Zeros and missing samples: EffortSignal keeps recorded zero watts as coasting. It holds power forward for at most ten seconds, leaves longer gaps missing, and excludes incomplete rolling windows. Activity analysis requires at least 80% power coverage for Weighted Power.
• Rounding: compare underlying values where available. Dividing rounded display values can give a slightly different result.

These are EffortSignal’s documented rules, not claims about TrainingPeaks’ handling of every file. A different displayed score does not, by itself, mean the core formula is different.

Where this equivalence ends

The correspondence here is to power-based cycling TSS®. It does not equate PSS with heart-rate-based hrTSS, pace-based running rTSS, or swimming sTSS, which use other inputs. EffortSignal can also calculate PSS from running power with a running-specific threshold; that is not a claim of equivalence to pace-based rTSS. TrainingPeaks explains its different training stress methods.

EffortSignal’s CTL and ATL history uses TRIMP, derived from heart rate, rather than PSS. The power-metric equivalence therefore does not imply that fitness, fatigue or readiness charts will match another platform.

Use the familiar interpretation when reading PSS, PI, Weighted Power and PV. For the sampling rules, data requirements and limits behind each value, follow the linked calculation guides above.

Content checked September 20, 2026 · EffortSignal release 0.1.30

All explanations · All features