There is a race plan a lot of runners carry to the start line without ever saying it out loud. Get ahead of goal pace early, while the legs are fresh and the crowd is loud, and put some time in the bank for the last 10K when things get hard. It feels like insurance. It is the most reliable way I know to run slower than you are capable of.
Aubrie has run more than ten marathons and a 2:59. The races where she ran her best were not the ones where she felt fastest at mile six. They were the ones where mile six felt almost boring.
The belief underneath the bank is that a marathon is an endurance problem you can get ahead of. It is not. It is a fuel problem you can only stay level with, and the arithmetic of the bank is worse than most runners have ever actually worked out.
The bank does not pay interest
Say you are chasing 3:00 to qualify for Boston. Goal pace is 6:52 per mile. You run the first six miles at 6:42, ten seconds quick, because it feels easy and the first six miles of a marathon always do. You have banked one minute.
Now price it. One minute spread across the remaining twenty miles is worth 2.3 seconds per mile. That is your entire cushion.
Against that, consider the other end. If going out fast makes you fade by even fifteen seconds per mile over the last six, you gave back ninety seconds to buy sixty. And fifteen is the gentle version of a marathon fade. The serious ones run a minute or more per mile, and they do not politely wait until mile 20.
That asymmetry is the whole argument. You deposit in a straight line and you withdraw on a curve. Everything below is an explanation of where the curve comes from.
What the extra ten seconds per mile actually costs
The intuition that fails here is that fuel burn scales with pace the way distance does. It does not.
George Brooks and Jacques Mercier laid out the mechanism in the Journal of Applied Physiology in 1994, and it is still the standard framework. They called it the crossover concept. At low intensity your muscles run largely on fat, which you carry in effectively unlimited supply. As intensity climbs the mix shifts, and past a crossover point somewhere near 65 percent of VO2max, carbohydrate takes over. Beyond that point, each further increase in intensity pulls a disproportionately larger share of energy from glycogen and a smaller share from fat.
Two things drive it. Harder running recruits more fast-twitch fibers, which are glycolytic by design and cannot run on fat as readily. And it raises sympathetic nervous system activity, which accelerates glycogen breakdown inside the muscle directly.
So those ten seconds per mile are not a ten-second problem. You have moved up a curve where the carbohydrate cost rises faster than the pace does, six miles into a race that will be decided by carbohydrate.
Benjamin Rapoport put numbers on that in PLoS Computational Biology in 2010, with a model that estimates, for an individual runner, the distance at which glycogen runs out at a chosen intensity. His framing of the paradox is the line worth keeping: even the leanest marathoner stores enough fat to power back-to-back marathons, and yet a comparatively tiny carbohydrate reserve is what ends the race. He also notes that more than two in five runners who race the marathon distance report hitting the wall. Not a small unlucky minority. Close to half.
One detail in that paper lands squarely on this audience. Rapoport's analysis runs all the way to the Boston qualifying standards and argues they sit close to the physiologic ceiling his model implies for well-trained runners at each age. Those times are not arbitrary. They are roughly what the fuel allows.
What 78,912 finishers actually did
The theory is tidy. The field data is blunter.
In July 2026, Qi Dong and Zhen Li published an analysis in PLoS One of split times from 78,912 Boston Marathon finishers across the 2015, 2016, and 2017 races, with checkpoints every 5K. Clustering the pacing profiles produced four groups: even pacing at 47.9 percent, mild positive splitting at 31.9 percent, strong positive splitting at 12.2 percent, and erratic, variable pacing at 8.0 percent.
Even pacing was associated with the fastest finish times at every age group and performance level they examined, averaging 217 minutes against 276 for the strong positive splitters.
Be careful with that comparison, because it runs both ways. Faster runners pace better, so some of that 59-minute spread is fitness rather than strategy. This is a large observational study, not a randomized trial, and nobody has run one of those over 26.2 miles. But the pattern inside the ability bands is hard to wave away: among runners finishing under three hours, 82.0 percent paced evenly. Among casual participants, 19.6 percent did. The people who have done this often enough to be good at it have nearly all landed on the same answer.
The most useful number may be the one about the erratic group. Variable pacing, meaning pace that surges and sags rather than simply slowing, produced the highest rate of hitting the wall, defined as a pace decline greater than 20 percent in the final segment, at 85.4 percent. Surging is expensive. Every surge takes you back up the crossover curve, and you do not get the glycogen back when you settle.
Why it feels completely fine at the time
Here is the part that makes this hard to fix with resolve alone. At mile six, running ten seconds per mile fast does not feel like a mistake. It feels like a good day.
Tim Noakes and colleagues have spent years arguing that pace is regulated in anticipation of the finish rather than in response to present damage. Your brain reads fuel status, core temperature, heart rate, and the distance left, then sets your sense of effort to whatever will get you there intact. Twenty miles out, with full glycogen and a cool core, the brake is off. The signal that you are overspending is not available yet, because the cost has not been incurred yet.
The model is still argued about and you do not have to accept every version of it. The practical implication holds regardless: early-race feel is not evidence. It is the point in the race where your perception has the least information, which is exactly why the decision has to be made on numbers you set before the gun.
How to actually run it
Four segments, and only the last one is about courage.
- Start to 10K. Goal pace or a touch slower, and it should feel like you are making a mistake. Give back five to ten seconds in the first mile to the crowd and the congestion, and stop grieving them. Effort at a fixed pace rises through a marathon even when the pace does not, because of fluid loss, rising core temperature, and accumulating fatigue. If mile three already feels like the right effort, it is the wrong pace.
- 10K to halfway. Settle onto goal pace and stop doing arithmetic. This is where surging costs the most and buys the least, so let packs go and let packs come back. Aim to cross halfway within about thirty seconds of half your goal time, on the slow side of it.
- Halfway to 20. The honest test. Holding goal pace here should be work, not a fight. If it is already a fight at 16, back off by five seconds per mile now and save the race. If it still feels controlled at 20, you have earned the last part.
- 20 to the finish. Now you spend what is left. Every second you did not waste at mile five is available here, and there is no fuel calculation left to protect, only distance.
Two rules that override all of the above. Run the first half on effort if it is warm, because pace targets set in cool weather are fiction at 70 degrees and rising. And never chase a split you dropped. Lost time is a sunk cost, and trying to win it back in one mile is how a 30-second problem becomes a 10-minute one.
Fueling is pacing by another name
You cannot pace your way around an empty tank, and you cannot fuel your way out of a pace that was always too fast. They are the same problem.
Asker Jeukendrup's work on carbohydrate absorption sets the useful ceiling. A single carbohydrate source saturates its intestinal transporter and cannot be oxidized faster than roughly 60 grams per hour, no matter how many gels you swallow. Combine transportable carbohydrates, glucose and fructose in something near a 2:1 ratio, and you open a second pathway. For events of 2.5 hours or longer, which is every marathon this newsletter is written for, the guidance lands around 90 grams per hour.
Timing matters as much as the number. Start at mile four, while your gut still has blood flow and your stomach still empties normally, and keep the interval regular. Waiting until you feel like you need it asks a compromised digestive system to do its hardest work at its worst moment. And rehearse every gram of it in long runs first. Race day is a poor place to discover that your gut disagrees with your plan.
Where the course changes the answer
Even pacing means even effort, not identical splits, and a few courses make the distinction expensive.
Boston punishes this hardest. The first four miles drop sharply, the downhill flatters your watch, and the quadriceps damage you take on for free early is collected with interest in the Newton hills and again on Beacon Street. Run the descent by effort, and let the early splits be quick only because gravity gave them to you.
New York does the opposite. The Verrazzano climb and the Queensboro Bridge at 15 are effort spikes that never show up as fast splits, so runners chasing an even watch overspend on both. Chicago and Berlin are flat enough that even splits and even effort are nearly the same thing, which is part of why the world records land there.
Where NXT RUN comes in
Most bad race-day pacing starts weeks earlier, with a goal pace nobody ever justified. A number from a race two years ago, a number a friend is chasing, or a round number that sounds good.
That is what Best Efforts & Achievements is for. Every run gets scanned for your fastest times across standard distances, tracked three ways: all-time, this year, and this training plan. So when you sit down four weeks out to commit to a number, you are reading what you have actually run recently. A 10K best effort buried inside a workout three weeks ago is a far better input to your marathon goal than a PR from a different body.
From there, PACE Coach will work through the split plan with you, and afterward PACE Analysis shows where effort and pace came apart, which is usually earlier than it felt.
This article is general training information, not medical advice. Racing a marathon is a significant physiological stress. If you have a cardiac or metabolic condition, or you are unsure whether racing is appropriate for you, talk to a healthcare professional first.