Splits Strategy: Even, Negative, and Positive — What the Data Shows

The most common marathon mistake is free to make and expensive to fix: going out too fast. Study after study of mass-participation marathon data shows the same pattern — the average recreational runner runs the second half 6–8% slower than the first. Among first-time marathoners, the figure is often 10–15%. The physiological consequences are severe and irreversible mid-race.

This guide explains the science behind splits strategy, what elite race data actually shows, and how to build a race plan using the splits planner.

What the data shows: elite splits

The most instructive starting point is looking at what the fastest marathon times actually look like in terms of splits.

Eliud Kipchoge, Berlin 2022 (world record 2:01:09):

• First half: 1:01:02
• Second half: 1:00:07
• Split difference: second half 55 seconds faster (0.75% negative split)

Paula Radcliffe, London 2003 (women's world record 2:15:25):

• First half: 1:07:52
• Second half: 1:07:33
• Split difference: second half 19 seconds faster (0.24% negative split)

Both world records are essentially even splits — within less than 1% between halves. This is not coincidence. Elite runners run precise even or very slightly negative splits because it is the physiologically optimal strategy for maximising marathon performance.

What does the typical recreational runner do?

Allen and Hopkins (2020) analysed half-marathon and marathon records from over 1.6 million US road race finishes. Their key finding: the median runner slowed 6.4% in the second half relative to the first. Among runners finishing between 3:30 and 4:00, the figure was approximately 7–8%. Among runners finishing over 4:30, it exceeded 10%.

The implication: almost every recreational runner is leaving time on the road by going out too fast.

The physiology of pacing

Understanding why positive splitting is so costly requires understanding glycogen.

Glycogen: the fuel that runs out

Your body stores carbohydrate as glycogen in muscles and the liver — typically 400–500 grams in a trained runner, providing roughly 1,600–2,000 kcal of energy. At marathon race pace, the body uses a mix of glycogen and fat. The faster the pace relative to your aerobic threshold, the higher the fraction that comes from glycogen.

At aggressive marathon pace (above aerobic threshold), glycogen burns significantly faster than at conservative pace. For a 3:30 runner going out at 3:20 pace for the first half: glycogen depletion arrives earlier, often around 28–32 km instead of 32–38 km. When glycogen is critically low, the body is forced to rely almost entirely on fat oxidation — which produces energy more slowly and cannot sustain marathon pace.

The result is the wall: a sharp slowdown in km 30–35 that no amount of mental fortitude can overcome once it's set in. The wall is not a mental event; it is a metabolic one.

The quadratic cost of pace

Running energy expenditure scales non-linearly with speed. At marathon race pace, running 5% faster doesn't just require 5% more energy per unit time — it requires approximately 10% more energy per unit time due to the speed-dependent aerobic and mechanical costs. Running the first 21 km at 5% above target pace depletes glycogen reserves at roughly 10% higher rate, then you must run the second 21 km depleted.

The energy arithmetic is unforgiving. Saving time by going out fast in the first half is illusory — you lose more time in the second half than you gained.

The calculator: building your split plan

Enter your target finish time and race distance to generate per-kilometre splits. Choose between:

• Even splits: identical pace every kilometre
• Negative split: first half slower by a specified percentage
• Positive split: first half faster (for modelling what happens if you go out too fast)

For most recreational marathon runners targeting a finish time, a 1–2% negative split is a practical and realistic target. This means starting the first 21 km at approximately 30–45 seconds per km slower than the second 21 km — not a dramatic difference, but a meaningful physiological buffer.

Worked examples of split strategies

Example: targeting a 4:00:00 marathon

Target finish: exactly 4:00:00. Required average pace: 5:41 min/km.

Even split plan:

• Every km at 5:41
• Half marathon checkpoints: 1:59:58 at 21.0975 km

Negative split plan (2%):

• First half average: 5:46 min/km
• Second half average: 5:35 min/km
• 21.0975 km checkpoint: 2:01:54
• Finish: 4:00:03 (2-second rounding)

The negative split requires running the first half approximately 2 minutes "slower" than even splits. In a race environment where you feel fresh and the crowd is energetic, holding back to 5:46 in the first few kilometres takes discipline. That discipline pays off around km 28–30.

What a 10% positive split looks like

For the same 4:00 target runner who goes out too fast:

• First half at 5:27 min/km → first half time: 1:55:04
• Available second half time: 2:04:56 at even splits — requires 5:54/km
• With the physiology of glycogen depletion, actually running: more like 6:15–6:30/km
• Likely finish: 4:17–4:25

Going out 5% fast in the first half produces a 6–10% deficit in the second half and costs 17–25 minutes on the finish time. The aggressive early pace was purchased entirely at the expense of the finish.

Checkpoint-based race execution

The most effective race-day pacing uses split checkpoints rather than pace displays. Print a pace band (a strip of target times for each 5K interval) and check it at the kilometre markers.

For a 4-hour marathon targeting even splits:

• 5K: 0:28:23
• 10K: 0:56:45
• 15K: 1:25:08
• 20K: 1:53:30
• Half: 1:59:58
• 25K: 2:21:53
• 30K: 2:50:15
• 35K: 3:18:38
• 40K: 3:47:00
• Finish: ~4:00

Check your watch at each checkpoint. If you're ahead at the 10K, slow down. If you're behind by more than 90 seconds at the half, assess — is it a good slow (intentional negative split strategy), or did you start conservatively and feel strong?

When positive splits are intentional

There are scenarios where planned positive splits make sense:

Point-to-point courses with net downhill: A course like Boston (net downhill in the first half, uphill at Newton Hills) or a trail race with a climactic early descent may require faster splits early to match the terrain, not the physiology.

Age group tactics: Racing to secure a qualifying time that requires a specific first-half split may justify going out at target pace even if conditions suggest a conservative approach.

Very short races: At 5K, the glycogen constraint barely applies — the race is over in 15–30 minutes. Even splits or slight positive splits are common and acceptable at 5K.

For marathon, the research unambiguously favours even or slightly negative splits for performance maximisation. The only question is how much negative.