The road back to 84.5% age grade.

In 2021, 392 W was my 5km best. Now I hold my old half-marathon power, about 370 W, for 5km.

Target band 380–385 W shaded. A 5km runs at about 104–106% of critical power, a half at about 96%. By those ratios, Saturday's 5km and the GNR both point to about 350–355 W now; Tuesday's threshold reps and the runzones LT2 estimate point to about 355 W.

Like for like, 1.05 now against 1.03 in 2021.

Economy is what each kilometre costs me. ECOR (energy cost of running) is kilojoules per kilogram per kilometre: 1.03 means 1.03 kJ for every kg of me, every km. Lower is better: the same speed for less fuel.

ECOR = power ÷ speed ÷ 73 kg. Races use official distance. My older pod's files show almost no air power, while the Stryd 5.0 adds it into total power, so the rings are the fair comparison with 2021–22. The June 5 miler, on the old pod, is the outlier at 1.10.

Heart rate at easy pace, on the route I've always run.

Three runs on the same out-and-back this week, in three different shoes. The best reading came on the fastest run.

HRV drops the morning after a hard session and is back within about 36–48 hours. That's the recovery rhythm the plan is built around: one easy day after each hard one. Each reading is taken on waking, so a session or rest day shows up in the next morning's number: watch the point after each green band for the bounce.

Heart rate from a Polar Verity Sense throughout 2026. The 2021 range comes from runs recorded with the same kind of sensor; files using the watch's wrist sensor are left out.

At the same speed, my 2026 stride is 3–4 cm shorter.

Each arrow is one race, from its opening kilometres to its closing ones. The two lines are fitted through the 2021–22 and 2026 race data. Squares are Tuesday's reps; triangles are strides, where extra speed comes from cadence, so they naturally sit below the race lines.

Contact in milliseconds, leg spring stiffness (LSS) in kN/m. At the same speed, +4 steps per minute adds only about 0.1 kN/m, so most of the lost stiffness is in how each foot strike happens, not how many there are. Vertical ratio is oscillation ÷ stride: in 2026 I bounce less but stay on the ground longer, a flatter stride rather than a more efficient one.

+20% peak power in three days. The ceiling is my top speed.

Thursday's best stride: 586 W for 10 s, 65 m covered at 2:34/km. Over 10–20 s, Stryd power rises roughly with speed, so a realistic top speed of about 8 m/s caps it near 620–700 W. The payoff isn't the peak itself: it's a bigger reserve above race pace, and stiffer legs at race pace.

Shape of the curve. 5km power ÷ critical power: 1.04 in 2021 (392 W ÷ 378 W), about 1.04 now (371 W ÷ ~355 W). The shape hasn't changed; the whole curve has dropped. Lift the engine and the top end comes with it.

My leg stiffness is 50–80% less consistent than in 2021.

Coefficient of variation of each metric while running at race or threshold pace. Pods and recording rates differ between 2021 and 2026, so read this as a strong hint, not proof. It matches how the stride feels: unstable underneath. Single-leg strength, pogo hops and quiet-contact drills target it.

The line I'm heading back to is 84.5%.

2025 WMA / USATF road standards, as used in the Runner's Arc. Raw grade per race, so distances differ.

Power is what the engine can hold for 10 miles. Watts per m/s is what shoes, wind and stride cost. The course drops overall, which this model ignores.

In November 2019 I ran 17:15 at Riverside parkrun, then 55:37 at Brampton. That's a ratio of 3.22, built on my legs, these courses and the same time of year. Riverside's layout has changed over the years, so this is provisional until the GPS tracks are compared.