Moving PT to 2pm: A zero-dollar human performance Army test

The Army Ran the Cheapest Human Performance Experiment in History. It Moved PT to 2 p.m. A zero-dollar intervention the Army keeps rediscovering, and what it would take to remember it.

Somewhere on the East Coast, an infantry unit spent a month doing what every infantry unit does, which is wake up in the dark and run. Then, with the command team's blessing, it spent the next month doing the same PT at two in the afternoon. Nothing else changed. Same soldiers, same programming, same first sergeant.

A team from the Walter Reed Army Institute of Research was watching. Daytime sleepiness went down. Insomnia symptoms went down. Napping dropped by about eleven minutes a day, which doesn't sound like much until you remember these are people who can fall asleep standing up in a chow line. Sleep quality scores improved. Morale went up. Every one of those was statistically significant, and the authors landed the plane about as plainly as a peer-reviewed journal allows: "a simple intervention of shifting PT timing from early morning to afternoon to allow for more opportunity to sleep, supported by leadership, can lead to improved sleep health and morale" (link).

The intervention cost nothing. The most sophisticated piece of human performance technology in the study was the minute hand.

I sell software to the military, so you'd be right to expect this is where I tell you the answer is software. Not yet. The study first, then a hundred years of history, then the part where I lose my patience a little.

What the study did and didn't find

Forty-seven soldiers enrolled and 37 finished. No control group. Morning PT always came first, so you can't fully rule out the season or the novelty. Most of the outcomes were surveys. The wrist monitors that were supposed to be the objective measure got worn about 64 percent of the time in the morning phase and 49 percent in the afternoon. The authors list every one of these limits themselves.

The finding I keep thinking about isn't in the headline. Total sleep didn't change. Soldiers got roughly 376 minutes a night under morning PT and 379 under afternoon, about 6.3 hours either way. (That number is its own scandal, and the Army has known about it for a decade.) What changed was the quality. The same six-and-some-change hours, moved to a point on the clock where a 26-year-old's body actually wanted them, produced less daytime sleepiness, fewer insomnia symptoms and less catch-up napping.

So: a strong signal, not proof. But I dare say it’s a signal everyone reading this knows is pointing at a truth.

Where 0630 came from

Nothing in Army doctrine requires PT at dawn. The 2010 physical readiness training manual says, in black and white, that leaders should "schedule and conduct PRT when it makes the most sense. PRT should not be reserved only for the early morning hours," and that they have "the latitude to adjust the PRT schedule" (link).

The dawn formation is a habit, and like most Army habits it has sensible roots. Heat, for one, if you're at Benning in July. Accountability, for another: first formation is how you find out who didn't come home. And there’s a strain of pure 1918 marketing in it. That year a magazine praised Walter Camp’s “Daily Dozen,” the twelve-exercise calisthenics routine the Army adopted for its World War I training camps, for giving recruits “a running jump start for the serious work day.” That is a Peloton ad with a doughboy in it.

People inside the Army have been poking at this for years, carefully. A captain writing in Military Review in 2009 put his finger on it: "The problem with morning PT isn't the program - it's the timing." He added that the Army "may accomplish more by 0900 than most people do all day, but this catchy former recruiting slogan fails to tell the whole story" (link). Thirteen years later the Sergeant Major of the Army said it with a lot less diplomacy. "I have to stick to a 0630 training schedule because I can't think past tomorrow," is poor training management, he told Military Times, (link).

So the doctrine allows it. The senior enlisted soldier in the Army has endorsed it, on the record, in a newspaper. And at roughly 0629 tomorrow morning, several hundred thousand people will be standing in a parking lot in the dark anyway.

If you're thinking everyone has already quietly moved on from this, I looked, and I don't think so. Nobody tracks how many units still run PT at dawn, which is its own kind of answer. But the Army's own sleep researchers described 0630 to 0730 as the typical schedule in a paper published three months ago, the brigade at Fort Wainwright was waking soldiers at 0522 before its pilot, and the researchers, the sergeant major and the officers writing for Military Review all treat 0630 as the thing to be argued out of, not a relic. Whatever is holding the schedule in place, it isn't policy.

The biology is not on the first sergeant's side

The soldiers in the WRAIR study averaged 26 years old. That turns out to matter.

A study of about 25,000 people found the body clock drifts later all through the teens and peaks around age 20 (19.5 for women, 21 for men), then slowly creeps earlier for the rest of your life (link). Which means the 20-year-old private and the 45-year-old command sergeant major who sets the training schedule are, biologically, living in different time zones. The sergeant major feels great at 0530. He is not imagining it. Neither is the private, who feels like he's been hit by a truck. Army sleep researchers, including WRAIR's own, have written that young adults need closer to nine hours and that their natural sleep window sits later (link). Teenagers are owls. Privates are owls with rifles.

Performance rides the same clock. Studies of athletes put the swing in physical performance across a day at up to 15 percent, found late chronotypes "significantly impaired during the morning hours in all measures," and put whole teams' peak windows somewhere between early afternoon and early evening (link). You will notice 0630 appears in none of those windows.

The reasonable objection is that afternoon exercise might wreck the night's sleep instead. The evidence is pretty settled: it doesn't. A 2019 meta-analysis of 23 studies found evening exercise did no harm to sleep and nudged deep sleep up a little, with one exception, hard exercise within an hour of bed (link). A 2025 look at four million nights of wearable data drew the safe line at four hours before sleep (link). The Army's own sleep guidance says finish PT three hours before lights out (link). A 1400 run clears all three by a wide margin. Nobody is proposing PT at 2100. Well, somebody probably is. Not me.

Everyone else learned this the hard way

The Army is hardly the only institution to discover that the clock is a performance variable. It's just arriving a little late, and with more people in formation.

Seattle pushed its high school start time from 7:50 to 8:45 in 2016, and researchers strapped wrist monitors on students before and after. Median sleep went up 34 minutes. Grades went up about four and a half percent. Attendance improved. The kids did not simply stay up later and erase the gain, and the authors explained why in a sentence that could be stenciled on the wall of every battalion headquarters: "the time at which teens generally fall asleep is biologically determined - but the time at which they wake up is socially determined" (link). Swap "teens" for "privates" and "school" for "first formation" and you've written the WRAIR abstract.

If 34 minutes sounds trivial, consider the mines. A study of 576,000 mining injuries across 23 years found that on the Monday after the spring clock change, when workers lose roughly 40 minutes of sleep, injuries rose 5.7 percent and days lost to injury rose by two-thirds. The fall change, which hands the hour back, did nothing at all (link).

Medicine found out with a randomized trial. Interns on the traditional 24-hour-plus shift made 36 percent more serious medical errors than the same interns on shifts capped at 16 hours (link). 

The Navy found out the worst way. The 2017 review of the Fitzgerald and McCain collisions, which killed 17 sailors, named fatigue as a contributing factor and ordered the fleet to "codify a circadian ship and watch rotation routine" (link). Four years on, GAO found the policy "inconsistently implemented." An estimated 14 percent of surface warfare officers got seven hours of sleep on their last deployment; 67 percent got five or fewer. And the Navy, GAO wrote, was not "measuring and managing fatigue in a timely manner" (link). Read that last clause again. The watchdog's first two recommendations weren't about sleep. They were about data.

The Army has run this movie before

This is the part that made me want to write this piece. The 2026 WRAIR study isn't the first time the Army has tried later PT. By my count it's at least the sixth.

In 2009 the 75th Fires Brigade at Fort Sill moved PT to 1530 and pushed the duty day to 0800. The brigade surgeon figured soldiers were picking up as much as 90 minutes of extra sleep. University of Maryland sociologists came out to survey everyone. Junior soldiers loved it; senior NCOs, less so. In April the brigade went back to mornings because of the Oklahoma heat, and the survey results never saw print (link). 

In 2014 Fort Carson went to reverse-cycle PT at the end of the duty day. The Army's first Health of the Force report singled it out for praise and noted that younger soldiers benefited most (link). No numbers were ever released that I could find.

Around 2016 a brigade in the 1st Infantry Division ran reverse PT and sleep banking for a week before a field problem. Sleep went from five or six hours a night to more than eight. Gunnery scores improved, along with morale and motivation. Then came the most honest sentence in any of these write-ups: leaders said the rest time "conflicted with other work duties and meetings" (link). Meetings. The same Army Public Health Center article mentions, almost in passing, that fatigue was a factor in 628 Army accidents and 32 deaths in a single year.

And this past January, on an AUSA panel, a former Old Guard leader described how wearables showed the regiment's leadership team was sleeping badly, so they moved first formation from 0630 to 0800. "Morale improved, sleep improved, cognitive decisions improved." On the same panel an H2F cognitive performance specialist talked about using self-report and wearable data "to better inform our training times" (link). Full disclosure: the system the Old Guard used to line up sleep and recovery against the demands of standing very still at Arlington was ours. I'd love to tell you we engineered the schedule change. We didn't. The regiment's leaders looked at their own numbers and made a call, which, as you'll see, is the entire point.

The best of the bunch happened in the dark. From November to December 2024, Charger Company, 1-5 Infantry, part of the 11th Airborne's 1st Brigade at Fort Wainwright, Alaska, ran an "offset PRT" pilot: PT moved later in the duty day, and soldiers woke up later. The brigade's H2F team surveyed the company before and after and put the results on a slide deck. Wake time moved from 0522 to 0641. Bedtime didn't move at all, 2222 before and 2228 after, which is the Seattle finding in camouflage. Self-reported sleep went from 6.04 hours to 6.55. Sleep satisfaction went from 49 percent to 79. Soldiers who never ate before PT dropped from 70 percent to 7. ACFT scores, minus the two-mile run, rose from 410 to 431, with gains on the deadlift, the throw, the push-ups and the plank. Soldiers who said their work had meaning rose from 62 to 80 percent. When asked for the biggest drawback, the most common answer was "none" (link). One company, eight weeks, surveys and a scale, no control group. Still the most complete dataset of the six, and it exists because a brigade H2F director decided to count.

The brigade commander, his command sergeant major and that H2F director then went on Drew Hammond and Alex Morrow's MOPs & MOEs podcast to talk it through, and the hour is worth your time. The commander had done the same thing earlier at 1st Ranger Battalion. The sergeant major's concern was keeping NCOs in charge of training, which is the right concern. And they were careful to call it a brigade initiative with H2F in support, not an H2F program, which is also right (link). If a brigade at the coldest post in the Army can run this experiment and publish the deck, the heat excuse is going to be a harder sell at Fort Sill.

Six experiments. Six good results. One slide deck. Not one dataset the next brigade over could pick up, drop its own soldiers into, and compare. Every unit learned something true about itself, and then summer came, or the commander changed, or the calendar filled up, and the knowledge went wherever knowledge goes when the person who had it PCSes.

It's a measurement problem

Let me be clear about what I'm not saying. I'm not saying every unit should run PT at 1400. Fort Sill's heat was real. The Big Red One's meeting conflicts were real. Fort Wainwright in December and Fort Bliss in July are not going to land on the same answer, and anyone with a kid in on-post daycare probably has a vote too.

What I am saying is that a battalion commander ought to be able to change one variable, PT time, and see his own unit's answer within 30 days. Sleep, sleepiness, morale, sick call, test scores, before and after. And the data should show up as a byproduct of how the unit already runs, without a WRAIR team, a grant, and a sociologist from Maryland.

That is, more or less word for word, what the H2F Management System is supposed to do. The Army's own description says the sleep domain "leverages biometric and self-reported data to monitor sleep quality, trigger rest and recovery recommendations, and automate appointments with specialists" (link). FM 7-22 makes sleep one of the five H2F domains and says, without hedging, that "leadership behaviors that target sleep can improve the sleep habits of unit members and the unit's overall sleep culture" (link).

Doctrine, meet data. In an early H2FMS pilot with the 82nd Airborne, the brief back to industry recorded that seven soldiers had reported sleep data (link). Seven. In the 82nd Airborne. I don't say that as a knock; it's an early-pilot lesson, and the WRAIR team learned the same one when half their subjects quietly stopped wearing the watch. If measuring sleep takes effort from the soldier, you will get exactly as much measurement as the soldier feels like giving you, which is to say seven.

It can be done, and the Army has done it. In 2020 the DEVCOM Soldier Center, the Army's medical research command and the 10th Mountain Division wired up an entire infantry battalion, 560 soldiers, with Oura rings, Polar watches and a one-minute daily check-in. The goal was catching COVID before it tore through the ranks. The system was running in 30 days. Soldiers spent a minute or two a day on it. Platoon leaders could see names; everyone above them saw trends. It found what it was looking for, and then it found something else. A platoon leader noticed one of his soldiers trending the wrong way on both sleep and mood, pulled him aside, and learned the kid was carrying trouble from home through a deployment to the middle of nowhere. The project lead called it a reminder that "the Human is the #1 capability the Army should invest in" (link). That was our platform too, back when it was called Smartabase. The team's own lessons-learned said the next version had to vary how often it bugged soldiers and make the prompts personal, because daily surveys die of neglect. Six years later that is still the hard part.

So here is what I think the H2FMS has to be if the seventh later-PT experiment is going to outlive the people who ran it.

It has to take sleep data from whatever a soldier already wears, with no logging. That sounds obvious. It isn't. A European national Olympic committee we work with tracks athlete sleep across six different wearable brands at the same time, because elite athletes wear what they like and nobody is going to tell a shot putter which ring to put on. Soldiers are the same, with more opinions. It has to roll all that up into a unit view a commander can absorb in a minute, with no names on it. It has to let him mark the day he changed the schedule and show him what happened on either side of the line. And it has to do all of that for a battalion in a month, not for a brigade in a study.

The sports world has been doing this with athlete management systems for a decade. The reason a college strength coach can tell you whether moving practice from six in the morning to three in the afternoon helped is not that he's sharper than a first sergeant. It's that the data showed up on its own and the answer was one question away. Whoever builds the Army's version has a harder job, because the data is more sensitive and the 450,000 users didn't sign up to be studied. But the shape of the job is the same.

The clock is free. The answer costs a little more.

Moving PT by seven and a half hours did more for one infantry unit's sleep than any supplement, app or mattress on the market, and it cost nothing. The Army has now learned this six times.

The seventh time, the lesson should survive the PCS cycle. And the first sergeant who moves his formation to 1400 should get to find out, within a month, whether the science was right about his soldiers, so that when the sergeant major asks him what in God's name he's doing, he can answer with a chart instead of a feeling.

He'll still have to give up the 0630 formation, but at least Teamworks can help him with the chart.