Metrics Explained
Skate Intelligence derives a lot from your raw Strava data — power, heart rate, wind, weather, and surface analysis you won't find in Strava itself. This page explains what each one means, how it's calculated, and where to find it in the app.
Numbers Strava itself computes and hands over as-is — not derived by this app, but worth knowing what they mean.
Strava's own heart-rate-based effort score, built from time spent in each HR zone weighted by exertion — conceptually similar to TRIMP below, but computed by Strava using its own zone model, not by this app. It only shows up on activities that had heart rate data recorded. Used here purely as a fallback: if an activity has no average heart rate to compute TRIMP from and isn't a power-meter ride, its training-load contribution falls back to this number.
Total mechanical work done during a power-meter ride, in kilojoules — the area under your power curve (watts × seconds ÷ 1000). A rough proxy for calories burned, since your body is roughly 20–25% efficient at turning food energy into pedaling work.
Power output divided by your body weight — the standard way to compare power across riders of different sizes, since a bigger rider can produce more raw watts without necessarily being fitter. Shown next to raw watts wherever your Strava-synced weight is available.
Computed from your activity's full second-by-second streams (position, speed, grade, cadence).
Your speed converted to what it would have been on flat ground, given the terrain you actually covered. Climbs cost extra energy per mile, descents give speed back, so GAP lets you compare a hilly route to a flat one on equal footing. Shown alongside your raw average speed on the activity page.
Stopped time is seconds where you weren't moving at all (lights, breaks). Coasting is time spent moving at speed while doing no work — detected from a cadence or power sensor reading zero while you're still rolling. Without a cadence or power sensor, coasting can't be measured and is left blank rather than guessed.
Mean stride/pedal cadence across moving samples, ignoring zero-cadence coasting periods so it reflects your actual working rhythm, not an average dragged down by freewheeling.
For rides with a power meter, or estimated proxies for skate sessions where noted.
A better measure of true physiological cost than plain average watts. It's a 30-second rolling average of your power, then the 4th-root of the mean of that rolling average raised to the 4th power — a formula that weights hard surges more heavily than easy spinning, because your body does. A steady effort has NP close to average power; a punchy, surge-y ride has NP noticeably higher.
Normalized Power divided by average power. Close to 1.0 means a smooth, steady effort; above about 1.1 means a punchy or interval-heavy ride with lots of surges.
Normalized Power divided by your FTP (Functional Threshold Power). An IF of 1.0 means you rode right at threshold on average; 0.7 is an easy endurance pace, 1.1+ is a hard race effort.
A single number combining how long and how hard a ride was, scaled so a one-hour effort right at your FTP scores 100. Calculated as duration × NP × IF ÷ (FTP × 3600) × 100. Used to track weekly training load alongside HR-based load for non-power activities.
A two-parameter model of your power ceiling: Critical Power (CP) is the power you could theoretically sustain indefinitely, and W′ ("W-prime", in kilojoules) is a fixed anaerobic energy reserve you draw down above CP and must recover before you can tap it again. Fitted from your best 3-, 5-, 12-, and 20-minute efforts across your ride history — shown on the Power Curve.
The standard Coggan 7-zone model (Active Recovery through Neuromuscular), each a percentage band of your FTP. Used to show time-in-zone for a ride and classify how hard it was overall.
How hard a session was on your cardiovascular system, and how that accumulates over weeks.
Training Impulse — a heart-rate-based load estimate for activities without a power meter (most skate sessions). It weights your heart-rate reserve (how far above resting HR you were, as a fraction of your HR range) through an exponential curve, so a hard effort counts disproportionately more than an easy one of the same length. Rescaled onto the same ~100-per-hard-hour scale as TSS so skate and ride sessions can be compared and summed.
Compares average speed per heartbeat across two equal halves after trimming the first 10 minutes for heart-rate lag. Lower is better; under about 5% on a steady long skate suggests your aerobic efficiency held up. The card includes every qualifying 40+ minute skate, so intervals or a progressive speed practice can move the score simply because the pace changed. Compare similar, steady session types for the clearest trend.
A rolling model of training load over time, borrowed from endurance coaching. Fitness (CTL) is a 42-day exponentially weighted average of daily load — your long-term training base. Fatigue (ATL) is the same idea over 7 days — your short-term tiredness. Form (TSB) is Fitness minus Fatigue: positive means you're fresh and race-ready, very negative means you're digging a hole that needs recovery to climb out of.
Normalizing performance for conditions outside your control, so trends reflect fitness, not weather or terrain.
Your average speed adjusted for how much headwind or tailwind you actually fought that session. The app fits a line relating your speed to the distance-weighted headwind component of each route (computed from wind direction/speed and your route's compass bearings), then uses that relationship to show what your speed would have been in dead-calm air.
The same idea, for heat: your speed is regressed against dew point (the best single predictor of heat strain — it captures the humidity-driven ceiling on evaporative cooling that dry temperature alone misses), so a summer slowdown from heat stress doesn't get mistaken for a fitness dip. Falls back to feels-like temperature when too few sessions have a recorded dew point yet.
The US EPA's standard air-quality scale for the activity's time and location, alongside PM2.5 and ozone readings: 0–50 is good, 51–100 moderate, above 100 starts affecting sensitive groups and eventually everyone. Sourced from Open-Meteo's historical air-quality data, not Strava.
Each route's surface mix (asphalt, concrete, gravel, etc.) and a 1–5 smoothness rating, matched against OpenStreetMap tagging along your actual path. Useful context for inline skating, where surface quality affects rolling resistance and speed as much as wind or grade does.
Longer-horizon views of progress across many activities.
Your best-ever pace over standard benchmark distances (mile, 5K, 10K, half, marathon). Each one is the fastest continuous stretch of that distance found anywhere inside your skates — a rolling window over the activity's GPS trace, not just skates or laps that happened to be exactly that long — so a fast mile buried in a 15-mile session still counts. A long stop ends an effort rather than being deducted from it, and wind context flags a record set on a tailwind day.
A progression score built from segments you've skated repeatedly and over a long enough span (at least 8 efforts across 90+ days) that repeat times actually say something about fitness rather than one-off route exploration. Each effort is scored against your all-time median pace on that segment — 100 means you matched your median, and every 1 point above or below is 1% faster or slower — and each period rolls its segments up once so a big interval day on one loop can't swamp it. An Average / Best toggle sets what each month or year shows: Average uses your typical effort that period, while Best uses your fastest effort, so the year-over-year view compares your best June this year against your best June last year. The median baseline stays fixed either way, which is what keeps periods comparable. Tracks whether you're getting faster on the routes you return to most.
A companion chart that compares your latest normalized effort across your most-tested benchmark segments. Every bar uses the same personal-median anchor of 100, making it easy to see whether an apparent fitness change agrees across several routes or comes from one unusually strong or weak effort. It adds a current cross-benchmark snapshot without replacing the smoothed Segment Fitness Index trend.
Several of these (full movement metrics, Critical Power, weather/air quality, surface quality) are computed by a backfill pass over your existing activities. If a metric looks empty on older activities, run the matching backfill from Settings → Data Enrichment. New activities pick everything up automatically as they sync.