Your Wristband Says You're Recovered. Nobody Asked Your Body.

I was a reductionist for most of my career. Watts, heart rate, weight, every number I could get my hands on. So I know exactly what the people behind that screen want, and it isn't stupid. A number about one part is useful. I still use trackers with my riders, in moments, for one question at a time.

The problem is what the number replaced. It replaced the foundation: feeling how you slept, feeling when the training is too much, feeling how much to eat and when to stop, how much to rest, what mood you're in, what you're carrying. That was your equipment before the strap. Now the strap tells you, and you stop asking. You outsourced the one instrument you can't buy.

Here's what happens to a body that stops asking. A muscle that's always on stops being reported; the brain files the constant signal as noise. Do that with the whole body and the whole body goes silent. The rider becomes a machine. And a machine progresses fast, because it doesn't feel the cost: it trains more, hurts less, follows the number, recovers on schedule. For a few years that's a massive advantage. It takes him very close to the top.

Then the same thing turns on him. He can't feel the slow slide into fatigue, because the signal that would tell him is off. He can't read the bunch. He can't come down at night, because coming down is a feeling, and he doesn't have those anymore. And then we expect him to get off the bike and be a person again, warm, social, glad to see his kids. Impossible. The body that was numb on the bike is numb at the dinner table. We built a robot and we're surprised it doesn't hug back.

A tracker measures one part. The race, the recovery and the life at home are decided by how the parts work together, where the body holds its balance and how far it can go and still come back. No sensor reads that. The only instrument that ever did is the one the strap switched off.

Sleep. In a 2025 laboratory study, WHOOP 4.0 correctly identified 93.58% of sleep intervals and 40.13% of awake intervals. It correctly classified 61.99% of light sleep, 69.63% of deep sleep and 61.99% of REM intervals. Sleep-stage agreement was κ = 0.37, a chance-corrected statistic, not 37% accuracy. Finding sleep and identifying its stages are different jobs. The study included 40 successful WHOOP 4.0 recordings, one laboratory night per participant.

WHOOP updated its sleep algorithm in 2025. The company reports over 7% better sleep-stage classification and over 3% better sleep-wake detection. That is progress. Its announcement does not provide a new absolute accuracy figure for WHOOP 5.0. These improvement figures cannot be added to the results of an earlier study.

Heart rate and HRV. The strap's best work, and I'll give it that. In a 2025 ECG comparison with 13 adults, WHOOP 4.0 had a mean absolute error of 1.78 beats per minute for nocturnal resting heart rate and 4.17 milliseconds for HRV. The mean absolute percentage errors were 3.00% and 8.17%, respectively. The WHOOP analysis covered 288 nights for heart rate and 289 for HRV. These are average errors, not a guaranteed error limit for each reading. That is a test of those measurements, on that device. It is not a test of your readiness to race.

Recovery. This is the one people train by, so listen to how it's made. Whoop takes HRV, resting heart rate, breathing and sleep, and compares them to your recent personal baseline. Green is the algorithm's assessment of readiness, built partly on your recent personal baseline. It is a score from those inputs. It is not a direct measurement of your whole body's recovery.

So take a rider sliding into fatigue. He digs himself in over three weeks. His recent baseline moves with him. This is the blind spot I look for: the score compares him with a recent history that already contains fatigue. The reference itself has moved. In a study of elite collegiate swimmers during a heavy training phase, the recovery score had no relationship with their reported stress or recovery. HRV was associated with stress. In the men, the recovery score was also associated with one metabolic measure. Those are different findings, and they matter.

The sport ran the same experiment this week, with no strap in it.

A rider crashed on the last lap of a championship. She kept racing. She finished on the podium. Every instrument she had, including her own felt sense, said continue, and all of them were right about her state and wrong about her body, because her state was the most mobilized version of that body it will reach all month. The cast appeared the next morning.

The test was right. The pass became permission. You have your own version and you know it: the session finished on will, the week pushed through because the bloods were normal, the block completed because the watch said recovered.

Tirreno-Adriatico, 2009. I crashed and hit my head. The doctor watched me through the night. The next morning I was on the bike. Six weeks to recover from that one.

China, 2017. I told the doctor I had a concussion. Then I kept racing. I have no memory of the last thirteen kilometers. I was riding for a contract.

Both times I was cleared. Both times the body wasn't. I'm not going to say that's normal. It is common. I'm not telling you this as a victim. I'm telling you because I know how a cleared body feels from the inside, and it feels fine. That's the problem.

Watch him in the last twenty kilometers of a race he wins. How his body sits in the bunch. How fast it answers a surprise. How he reads a move before it's a move, and how he comes down afterwards as if nothing happened. That's organization, everything in him working as one thing, and that's what makes him the greatest of all time. Not the sleep score. The way the whole system holds together under a load that breaks everyone else. He said himself this week that he never thought tests were a real indicator of potential, and that he's probably not the one who trains the most. The people with the best numbers are the least impressed by them. The people most impressed by the numbers are usually the ones selling them.

Keep the strap. It answers one question, "how did the numbers look last night", and it answers it well. The questions that decide a season it can't hear.

Five readings, ten seconds each, no device. In a heavy block they'll all say tired, and tired is expected; you chose the block. What they tell you that the strap can't is whether the tired body still answers. If it does, keep going. If the breath stops ending, the movement goes heavy and the mood doesn't come back by evening, the block has become a slide, and the strap is still green, because the score is comparing you with your recent baseline.

One limit, because it's real. Use the numbers, and use the bloods. They count well. They just can't count this.

What you get back is a body you can read before the phone can, and a season where you know the difference between hard and too much.

JB