Strava · 10 rides with power + HR · 9 Jul – 24 Aug 2026

Where the watts go

Watts per heartbeat falls across every ride — but not for the reason you'd expect. Your hard rides hold up better than your easy ones, because on an easy ride you open at tempo power and spend the next ten miles settling back down.

Progress by segment

For a given kind of ride, are the mile blocks improving — and is that watts going up, or heart rate coming down? The chart carries the shape; every number is in the matrix beneath it.

Window	Miles 0–5	Miles 5–10	Miles 10–15

Two stages, separated

Power and heart rate, each indexed to its own segment 1, so you can see which one gives way first. A ride counts toward every segment it finishes — ride 6 miles and you are in 0–5 but not 5–10 — so the ride count falls as the segments get deeper. Watch the n values.

Average power Average heart rate

Every segment

Every 5-mile block of every ride. The two rolling columns hold each segment against the trailing 30 days of that same segment, so you can see whether a ride beat its own recent baseline.

Date	Ride	Type	Segment	Avg W	Avg HR	W/bpm	vs seg 1	Rolling 30d	30d vs seg 1

Method & caveats

Metric: Mean power divided by mean heart rate over each 5-mile block of road, from Strava's per-activity streams. Higher is better: more watts per beat.
Ride type: Share of ride time with heart rate above Zone 2 (above 147 bpm, your manual zones): Endurance under 35%, Tempo 35–70%, Threshold above 70%. The split is naturally clean — two rides at 19% and 20%, everything else 49% and up, nothing in between.
Sample: The binding limit. Ten rides carry both power and heart rate, all between 9 Jul and 24 Aug 2026. Only six have two complete segments, giving n = 2 per ride type. Treat the by-type result as a direction to test, not a measurement. 2025 holds one outdoor ride over 5 miles.
Confound: Intensity correlates with the segment-1 ratio itself (r = −0.64), so easy rides start from a flattering number and have further to fall. On the clean test — intensity against segment 2 alone, sharing no term with segment 1 — it weakens to r = +0.48. Direction survives, magnitude does not.
Terrain: Ruled out. Climbing averages 28, 29 and 31 ft/mi across segments 1, 2 and 3 — later miles are marginally harder, not easier.
Warm-up: Ruled out. Dropping the first 1.5 miles, where heart rate is still climbing, moves the segment 1 → 2 decline from −7.5% to −7.7%.
Windows: “Most recent” is the latest ride of the selected type — otherwise it empties out whenever you filter. Last 30 days is the rolling window, which is why it marches; the exact dates for every window are shown in the legend. The custom range is off until you add it and defaults to the trailing 12 months, which given the power meter's history is every ride you have — so read it as the all-time baseline the other two sit against, not as an independent sample.
Incomplete: Each ride's final partial segment is usually the coast home and is excluded from every average. Aug 24's last 2.9 miles lose 58 ft at only 5 ft/mi of climbing.