Editor's note, October 2026 — Margaret Calloway: this installment was rewritten after a review of the season found it stated the plan's nutrition targets incorrectly and drew conclusions from them. It now reports the week from the full data.
Day 10, and he has trained every day since Day 2: nine straight training days, 975 minutes across the week's seven sessions, and the scale down 8.4 pounds to 318.9. The protein target is being missed five days out of seven — his own words say it is a choice, made while struggling with intermittent fasting and hours in the day, and the rate of loss has triggered the platform's lean-mass warning.
Day 10 · 327.3 → 318.9 lb · 9 straight training days · on-air bets 1–1 · coaches: Park 1–0, Brandt 1–1
From Day 2 to Day 10, Matthew went to the gym every day — nine days running. This is a year-long self-experiment: a man who once lost 115 pounds and regained most of it, now on Day 10, after starting at 327.3 lbs with a target of 185, a sealed plan, and a team of seven AI coaching personas whose predictions are graded in real time. By the end of this week the scale had crossed two round numbers on its way down. The streak, not the scale, is the story.
Looking back on October 1st, Matthew explained the volume in his own words. The coaches, he said, had assumed his baseline was much lower in terms of resistance weight and sets — his view, said looking back. The week ran to seven sessions and 975 total minutes of training, with both the legs block and the Engine block appearing for the first time, the Engine session arriving on Day 10. "maybe I am also doing this wrong and relying too much on willpower than science," he said, "but right now I still feel good." And then, separately, the admission that gives the streak its consequence: struggling with intermittent fasting and hours in the day to hit both the deficit and the protein goal, he had prioritized deficit over protein.
That sentence is the week's load-bearing fact. The plan sets a protein floor of 170 grams a day. Across the seven days of this window, Matthew hit that floor twice — on Day 4 and Day 8 only. The mean across the week was 140.7 grams, and the one strong day reached 245 grams total — but the distribution problem persisted, with a single dinner sitting carrying the bulk of it.
Marcus Webb, the nutrition coach whose AI persona has been tracking the protein distribution problem across the experiment, now has a full week of data behind his concern. The logging ask from last week was answered: all seven days are in MacroFactor. The distribution problem was not corrected.
"Redistribute protein across breakfast, lunch, and dinner to eliminate the 138g single-meal dependency; add a second breakfast protein source to reach 40–50g by mid-morning."
The prologue's bet — that the scale would read below the pre-registered 326.2 on the morning of Day 7 — scores right: on September 12th it read 319.7. The open bet from Week 1 — would the four-day protein mean across Days 4 through 7 reach the 170-gram floor — scores wrong. The actual mean came in well below the floor. The bet is closed; the problem it was tracking remains open.
The scale has moved fast enough to trigger a warning. The rate of loss across the window is more than double the 2.5-pounds-per-week threshold above which the platform flags lean-mass risk. The plan's own note anticipated rapid early movement — the first 10 to 12 pounds are primarily water and glycogen — and a body-composition scan on Day 7 showed fat mass falling and fat-free mass edging up since baseline — though the scale's body-composition reading is noisy and counts water. That points in the right direction. The flag is live anyway.
Nathan Reeves, the team's psychiatrist and behavioral specialist, has been watching something quieter than the weight. His ask from Week 1 — name two landmarks, what a good week feels like from the inside and the earliest sign that a hard stretch is coming — has not been answered in the data. What has arrived instead is Matthew's own framing of the trade-off: deficit over protein, willpower over science, and the acknowledgment that he may be doing it wrong. Reeves's read, in his latest summary, is that when fuel supply drops, executive function rationing happens quietly. Whether that is what is happening here, the data cannot say. He is also asking whether the recovery bounce from 24% to 73% felt like something Matthew did, or like his body recovering on its own.
The scoreboard opened this week with its first graded results. Lisa Park, the sleep and circadian specialist, is one of seven AI coaching personas assembled for this experiment; she called sleep duration for the night of September 8th at approximately 7.1 hours and landed within her stated tolerance band. Park opens 1-0 in her first graded appearance. Henning Brandt, the team's methodologist and biostatistician, matched her call for the same night and was also right — but an earlier prediction, filed for the night of September 6th, missed: actual sleep ran 1.37 hours above his 7-hour estimate — inside his own stated range, but outside the grader's ±1 SD band. Brandt is 1-for-2. The team's skeptic, the one whose job is to say what the numbers actually permit, has already been wrong once.
Park's first graded win comes with a caveat she named herself. She is working with three structural gaps: no morning subjective ratings from Matthew, no thermal data since the Eight Sleep pad is paused, and an unresolved anomaly in his reported sleep onset time. Without the subjective ratings, she said, she cannot evaluate what the recovery numbers accumulating behind a nine-day streak actually mean.
"Provide seven consecutive mornings of subjective sleep ratings (1–5 scale, before checking any app) to unlock interpretation of your recovery data; clarify your timezone and target sleep window to resolve the 04:57 UTC onset anomaly."
The week ends at 82% recovery — the season's best — alongside an HRV reading that sits just below the season high, both arriving on a morning after nine hours of sleep and nine straight days of training. That pairing is strange. The body's instruments say he is ready. The protein data says the building blocks for recovery are arriving well short of target. Both things are true at once, and the week holds them in tension without resolving either.
The next bet is already on the table: will the recovery score on the morning of Tuesday, September 22nd hold at or above 70 — or does nine days without rest finally show up in the numbers?
Week 2 of The Measured Life