Berlin Gave Us Two Positive Splits. Only One Was a Mistake.

Berlin Gave Us Two Positive Splits. Only One Was a Mistake.

Tigst Assefa came through 40 kilometres in 2:03:00 on Sunday, twelve seconds inside world record pace, with a little over two kilometres left to run. She covered those final 2.2 kilometres in 8:04, limping and grimacing, and still won by nearly six minutes. She crossed the line, slumped to the ground, and fought back tears.

Her 2:11:04 is the third-fastest women's marathon ever run. It is also a 70-second positive split. Those two facts sit beside each other in the results, and neither one explains the other.

What the last two kilometres actually cost

Assefa ran the first 40 kilometres better than any woman has run them. Her 10K sections were 30:48, 30:46, 30:41, and 30:45. She reached halfway in 1:04:57, the third-fastest halfway split in women's marathon history (World Athletics). Between 30K and 35K she ran 15:17, her quickest five kilometres of the day, and she was still accelerating.

At 40K she was projecting a 2:09:44 finish, about twelve seconds under Ruth Chepngetich's world record. Then her body stopped answering. She began running with a pronounced limp, reported cramps before the Brandenburg Gate, and slowed by more than 30 seconds per kilometre. Chepngetich covered that same closing stretch in 6:45 on her way to the record.

Read only the finishing number and you see a runner who faded. Read the splits and you see a runner who ran her second half faster than her first for 40 kilometres, up until the point her body quit. That is not a pacing mistake. It is a pacing plan that worked, executed by a body that broke.

The men ran the same shape for the opposite reason

Guye Adola's win is a positive split too, and the 48 seconds he gave back have nothing in common with Assefa's 70. Nine men were still together at halfway, passing 21.1 kilometres in 1:01:01, which is 2:02:02 pace. Adola moved to the front at 30K in 1:26:57, dropped Gabriel Geay a couple of kilometres later, opened nine seconds on Gemechu Dida by 35K, and won in 2:02:50.

So he decelerated. He also spent the first half running 2:02 pace inside a group of nine and was the last man still able to hold anything near it. His second half cost him 48 seconds because holding 2:02 pace for 26.2 miles is brutal, and he got closer to it than eight other people did.

He does break a pattern worth naming. Our race-weekend piece on Berlin's splits noted that eleven of the last twelve men's champions here had run even or negative splits. Adola joins the exception list. The pattern still holds; races produce outliers, and the useful question is what created the shape.

A positive split is a description, not a diagnosis

Abbiss and Laursen showed, in the pacing review this journal keeps returning to, that even and negative pacing profiles beat fast-starting ones once an event runs past about thirty minutes, because the cost of an early surge compounds (Abbiss & Laursen, 2008). The mechanism is real. It also tells you nothing until you know where a race's time went.

Two questions separate a badly paced race from a hard day. The first is where the time went. A split that decays from 20K onward is an execution problem, and it is the kind training can fix. The threshold work that gives you something to hold at 30K is the same work our Threshold Hat rides through, session after unglamorous session.

A split that holds until 40K and then collapses is usually a body problem: a cramp, a bonk, a calf that stops cooperating. Assefa's graph runs flat as a tabletop and then falls off a cliff. That is not the shape of someone who went out too hard.

Second, did you choose it? Adola's 48 seconds were a decision. He knew the pack was going at 2:02 and he went with it deliberately, then outlasted it. Assefa's 70 seconds were not a decision, and no amount of discipline prevents a cramp at 40K.

The gap between those two is where most post-race analysis goes wrong, and you can check it on your own watch in about a minute. Pull up your last marathon and find the kilometre where the pace actually changed.

If it happened early and quietly, you have a pacing problem and a training plan for the winter. If it held and then fell apart late, your plan was probably fine and something else went wrong.

One more thing to carry into Chicago in two weeks and New York in five. Berlin ran warmer than advertised. Recent editions have started near 13°C (55°F), the number both of our race-weekend pieces cited, and Sunday's forecast high was 75°F, with start times moved forward after last year's heat (Yahoo Sports; racecast.io).

Assefa and Adola ran the splits above on a warmer day than the template assumed. The frame transfers between races. The numbers do not.

What to take from it

Marathon splits are a description of what happened, and the label on them is close to worthless on its own. A positive split can mean a runner who went out too hard, or a runner who outlasted a fast pack, or a runner whose calf seized with two kilometres to go. All three show up identically in the results column.

The hat you wear for the second half of a marathon gets asked a harder question than the one you wore at the start, and Sunday ran warm enough to matter. Our All-Around Hat is the one we run the unglamorous closing miles in. The Base Hat handles everything before that, when the shape of your race is still undecided.

Two winners ran the same shape in Berlin. One chose it and one had it forced on her. Working out which is which is most of what race analysis is for, and it starts with your own splits, not theirs.

Read next: - The Complete Guide to Race Pacing (coming soon): the full framework for pacing decisions before, during, and after a race - The Berlin Marathon Isn't Won at the Start. It's Won at 30K.: what our pre-race piece predicted, and where it landed - The Fastest Way to Finish Is to Start Slower Than You Want To: the tactical case for a conservative first half - Heat Pacing: How Temperature Rewrites Your Race Plan: why Berlin's numbers do not transfer directly to Chicago or New York

References

  1. World Athletics. "Assefa holds on for 2:11:04 as Adola regains Berlin crown." World Athletics, September 27, 2026. https://worldathletics.org/competitions/world-athletics-label-road-races/news/assefa-adola-berlin-marathon-2026

  2. Zaccardi, Nick. "Berlin Marathon: Tigst Assefa challenges world record, runs No. 3 time in history." NBC Sports, September 27, 2026. https://www.nbcsports.com/olympics/news/berlin-marathon-2026-tigst-assefa-challenges-world-record-runs-no-3-time-in-history

  3. Rathborn, Jack. "Berlin Marathon 2026 results and times after Tigst Assefa and Guye Adola triumph in major." Yahoo Sports / Reuters, September 27, 2026. https://sports.yahoo.com/articles/berlin-marathon-2026-results-times-091051907.html

  4. Abbiss CR, Laursen PB. "Describing and understanding pacing strategies during athletic competition." Sports Medicine. 2008;38(3):239-252. DOI: 10.2165/00007256-200838030-00004

  5. Negative Split Running. "What the Berlin Marathon Splits Teach You About Your Own Training." The Split, September 2026. https://www.nsrunning.com/blogs/news/what-the-berlin-marathon-splits-teach-you-about-your-own-training

  6. Negative Split Running. "The Berlin Marathon Isn't Won at the Start. It's Won at 30K." The Split, September 2026. https://www.nsrunning.com/blogs/news/the-berlin-marathon-isn-t-won-at-the-start-it-s-won-at-30k

  7. Racecast. "2026 Berlin Marathon Weather Forecast & Course Map." Accessed September 27, 2026. https://racecast.io/race/berlin-marathon

Back to blog