Oceanside (CA) → Annapolis (MD) · 13–25 June 2023. What the watch recorded across 11 days, 20 hours and 1 minute — and what that data says the story told from the inside.
Source: Garmin Connect. Data collection has limits, which are flagged.
The watch recorded temperatures between 2 °C and 50 °C across the crossing. The high comes on Day 2, in the desert (California/Arizona, near sea level); the low on Day 4, at Wolf Creek Pass, Colorado, where the Garmin reads 3,309 m of altitude. Between the two, a little over 48 hours — confirming, with data, the line to the press on arrival: climbing to “mountain passes at 3,200 m, less than two days after crossing the desert, at zero metres, at 50 degrees”. And the measured figure still falls short of what was felt: in the desert the asphalt reached 60 °C and radiated heat.
One caveat: the Garmin measures temperature, not humidity. The difference between the dry desert heat and the muggy Midwest heat (Kansas/Missouri, Days 8–9, highs of 46 °C) has no direct record on the watch.
Summing the positive elevation of the 12 recorded blocks, the total is 36,630 m (about 120,200 ft). The route book and race press announced 175,000–179,000 ft (53,300–54,600 m). The highest point matches — Wolf Creek at 3,353 m official against 3,309 m on the watch, within a barometer’s tolerance. The divergence is only in the sum of the climbs.
It is not an error on either side: it is methodology. The route book adds section by section, with generous rounding; the barometer measures real elevation minute by minute, but with smoothing that “eats” micro-climbs. The truth is somewhere in between; the watch number is the conservative one.
Note: “Day 12” shows 8,020 m — 22% of the whole race’s climbing — but it is not a day: it is the merged final leg (530 km, ~30 h).
Across the 12 days, the watch recorded 4,742 km. The same race gives three very different average speeds, depending on what you measure:
| What is measured | Maths | km/h |
|---|---|---|
| Moving (bike rolling only) | 4,742 ÷ 190 h | 25.0 |
| Recording (incl. short stops) | 4,742 ÷ 224 h | 21.2 |
| Door-to-door (elapsed time) | 4,742 ÷ 285 h | 16.6 |
The ~25 km/h “moving” is the sustained base pace; the ~17 km/h door-to-door is the real race, with a third of the time spent off the bike. The split of time:
| Time | Hours | % of race |
|---|---|---|
| Moving | 190 | 66.7% |
| Short stops (recorded) | 35 | 12.3% |
| Long stops / rest | 60 | 21.1% |
In total, nearly 95 hours off the bike — two whole days spread across twelve. Average power was 109 W (144 W normalised) — low by definition in a race of management; only Day 1, with fresh taper legs, rises to 142 W average (174 W normalised). It was also the fastest day: 29.9 km/h moving average, with a continuous 30 km stretch above 40 km/h on the descent into the desert.
The pedals measure each leg’s work separately. And it is here that the knee injury — which the story tells from the inside — appears in the data, raw and without rhetoric.
At the start, the body pedalled almost balanced (Day 1: 52% left / 48% right). From there, the balance came apart always in the same direction: by Day 7, the left had fallen to 40.8% on average. And to be clear: it is not a healthy leg compensating for an injured one — both knees were suffering. The right retained the capacity to generate force, and the body leaned on the side that could still push.
And — most remarkable — it did not stay that way: by Day 12, near Annapolis, the balance was back to 53% / 47%. An unlikely recovery, achieved in-race. The pedal feedback gave me something rare in the middle of the pain — real-time awareness of what each leg was doing. The data was not only the record of the problem; it was the mirror that helped correct it.
And within Day 7 itself, the average softens the worst. Isolating the 5 km segments, the morning pushed the right leg to 60.6% of the work, and at 09:39 to 63.6% — nearly two thirds on a single leg. In that same segment, left-leg torque effectiveness dropped to 49.5% (against ~71% when fresh).
It was not just a feeling: on the morning of the hardest day, with both knees suffering, the body shifted nearly two thirds of the work onto the leg that could still push.
Between stages, total stopped time averaged ~5 hours (11 intervals, ~56 hours total). The watch detected real sleep on only one night (1h18, between Day 1 and 2); on the others it was mostly off the wrist and charging, or without enough sampling. The ~56 hours existed and were used — but as fragmented power naps, seated, not continuous sleep.
This ties directly to the RV: the crew ran with two vans, not a motorhome. A fixed, dark, vibration-free bed would have created conditions far closer to what the body (and the watch) recognise as structured sleep — with likely effects on the autonomic recovery in section 7.
My normal weight was around 88 kg. I did not weigh myself during the race. The first post-arrival readings show ~3% body fat (with caveats, being bioimpedance on a rehydrating body). My weight bottomed at 82.9 kg and was recovered by around 13 July.
As for fuel: the watch’s burn estimate adds up to about 9,000 kcal/day — and the body, under effort, absorbs at most ~6,000. That gap is the norm in ultra-cycling of this length. Estimated water intake totals ~129 litres — ~11.4 litres per day.
Before the start, heart-rate variability was stable and balanced (41–49 ms, nearly every day). The last nightly reading is 13 June — the start day. During the race there are almost no readings: the watch was mostly off the wrist, and HRV is read during sleep.
When the records resume, on 26 June, something revealing happens: the watch enters recalibrating mode — the race broke continuity so badly that the Garmin had to relearn the baseline from zero, as if I were a new user.
| Period | Nightly HRV (ms) | State |
|---|---|---|
| 1–13 Jun (pre-race) | 41–49 | balanced |
| 14–25 Jun (race) | — | no record |
| 26–28 Jun | 30–45 | recalibrating |
| 28 Jun — minimum | 30 | the lowest of the whole series |
| 29 Jun – 8 Jul | 33–49 | low / unbalanced |
| 9–15 Jul | 44–52 | back to balance |
Three days after the finish, HRV drops to 30 ms — the lowest of the whole series. And it only returns to balance on 9 July, about two weeks after arrival. “Feeling recovered” and being recovered, at the autonomic level, are different things — and the second takes far longer.
Resting HR tells the same story from the opposite side: as HRV fell, resting HR rose. From the usual ~44–50 bpm, it jumped to 62 bpm on 28 June — minimum HR that day was 59 —, exactly the day of the HRV bottom. Two independent signals, one verdict: three days after the finish the body was at peak stress, and took ~2 weeks to settle.
Crossing the themes, the picture is of a race that imposed a physiological cost far wider than a simple weight-loss reading would suggest: a real episode of pedalling asymmetry mid-race, resolved before the finish; rest made of unstructured power naps; a chronic caloric deficit; and — the clearest finding — an autonomic recovery that took about two weeks to normalise, long after the body already looked whole.
The three highest-return points: a dedicated RV to structure sleep; a fuel plan calibrated above the watch’s estimate; and post-race HRV tracking as the main signal of when the body is truly ready to train hard again.
Distance, elevation, power, cadence, temperature, balance, sleep, HRV and weight come from Garmin Connect. Official elevation and time come from the organisation: the official time was 11 days, 18 hours and 51 minutes; the elapsed time, door-to-door — the frame this analysis uses — was 11 days, 20 hours and 1 minute. Three limitations: the watch measures temperature, not humidity; bioimpedance right after extreme effort is unreliable; and there is no sleep or HRV record for almost the entire race (14–25 Jun).