The Elusive Negative Split

Every runner knows that the correct way to pace a marathon is the negative split, a conservative approach intended to ensure the athlete reaches the latter stages of the race in suitable condition to finish strongly. I can distinctly remember reaching the halfway point of my first marathon, glancing at my watch and resolving to run the next 13.1 miles faster. I didn’t. Despite best intentions I crossed the line with a positive split having gradually slowed over the final 10 miles. I was not alone. For all the discussion of the negative split, remarkably few actual manage it on race day.

The Elusive Negative Split at the 2013 Virgin London Marathon

As the results from the 2013 Virgin London Marathon show hardly any athletes negative split, a mere 4% of all finishers ran the second half of their race more quickly than the first. In fact the largest group of finishers are those that ran between 5 and 10 minutes slower in the second half of the marathon, that’s somewhere between 20 and 45 seconds slower per mile. The pattern if we look at average pacing over the course of the marathon is a clear positive split as the graph below demonstrates.

The Average Pace During the 2013 Virgin London Marathon

Start quickly, at first slow gradually, then, after the halfway point, slow more rapidly, until the sight of the finish and a final surge of energy: this is the positive split. Despite all the advice to hold back in the first half of the race few of us manage it, not even the elites.

Even the Elites Struggle with the Negative Split

The dynamics of the event may be different but perhaps the fact that 93.5% of the elite field also failed to negative split will be of some consolation to the amateur runners. Treading the line between going too fast early on and not holding back too much is difficult; especially when you’re racing for the win.

What should runners do? Certainly not give up on the idea of negative splitting. Practice makes perfect, negative splitting some training sessions will help develop the skill. If nothing else mentally adopting a strategy that holds back early in the race will help curb the instinct to go out too fast. Attempting to work harder in the second half of a race may not deliver the elusive negative split itself, but it might help to prevent a significant positive split.

[Edit 25/4/2013: updated figures and graphs due to an error in my calculations. Original, incorrect versions of the graphs can be found here: “The Elusive Negative Split at the 2013 Virgin London Marathon” and “Even the Elites Struggle with the Negative Split“.]

All Ironman Results and Statistics

A growing collection of results and statistics for the whole Ironman race calendar.

Find out what it takes to place in your age group or to qualify for the Ironman Worlds Championships in Kona.


  • Rob Knell

    Is there any reason to think that a negative split is the best way to run a marathon? If you’re running for a time and not for a place then a slight positive split might well be the best option, I’d have thought. If you can run the second half faster than the first doesn’t that mean you didn’t run the first half fast enough?

  • Good question Rob,

    I tried not to be overly committed to the negative split in the post because I’m not certain it is the best way. At the very least, on a practical level, I suspect that for most of us to achieve a negative split we would tend to run the first half too slowly to be compatible with a goal of our fastest possible time. So the main value I see in the negative split comes from thinking in terms of effort and from the principle that you don’t over-commit at the start.

    There’s been a few discussions on twitter about this and some interesting points and articles come up from that. This piece at Competitor covers a lot of good ground: The Art and Science of Marathon Pacing

    I liked the idea from Ross Tucker that we should take the fact that most of the fastest marathon times at a world class level are negative splits as an indication that this is the strategy that delivers the fastest times. Perhaps for those pushing the boundaries it is most likely to happen when their pacing sets them up to be able to finish that strongly and negative split. Of course it’s hard to know if all other aspects equal they could have achieved the same or faster time through a different strategy.

    That article talks about even splits and also thinking in terms of effort. It seems negative splits are rarely large and I’d suggest positive splits from a well executed race aren’t either. Maybe +/- 2 minutes would equate to a close-to-optimal race, perhaps the negative split version being more likely to deliver close to fastest potential time.

    Another suggestion from twitter that I agree with is that the positive split strategy inherently comes with more risk. If you mis-pace your race and your chosen positive split is too aggressive to start your second half will be much slower than planned and more likely to lead to an overall slower time. In this instance the negative split strategy would hold fewer risks as the chosen initial pace is less likely to put you in a situation where you slow significantly. Of course it could and this is where it’s worth saying that regardless of pacing strategy what’s probably more significant is an excellent judgement of ability and pace.

    We could look at the numbers and judge those falling close to an even split as those most likely to have correctly chosen their race pace(s). Those looking at more than 5-10 minutes positive or 5-10 minutes negative have very likely left something on the race course.

    Final thought, from some tweets by David Tilbury-Davis around this Physfarm article: Understanding Work Above Threshold. Applying those same principles to running at least helps to explain the dynamics. A critical pace for a marathon (affected by course and conditions) and a limited capacity to work above that pace; any strategy needs to pace close to the critical pace and use periods above that in limited quantity, spend too much time above critical pace and effectively you eliminate any further capacity to work above that pace and potentially reduce your critical pace for the remainder of the race. Exact optimal pacing strategy may then depend on course/conditions, but again I see that as coming back to the risk if you put too much of that above critical pace work in early on.

    Which all leads to my thoughts that for most the negative split is almost an accidental side effect of good pacing rather than something we can deliberately target to deliver our fastest race. Probably we need a pacing strategy in mind that has some idea of our best potential marathon pace (critical pace if you like) and some idea of how much we might be able to push within that. We probably need to be mentally planning that to coincide with increased RPE later, inevitable with fatigue, and more likely an even split which if we’ve judged our pace well, has a small chance of negative splitting in the end through a final push, and some potential of a slight positive split as we fade.

    Phew, that’s almost as much as the original post!


  • Dan McCormack

    Nice stats Russ.

    I wonder whether a -ve split is something to strive for.

    Based purely on HR when I have run an even paced marathon (or long run) my HR has been c5-10 BPM higher in the second half. The effort to maintain pace has been high. Conversely, when I stuck to a steady BPM at VLM 2011 (the only time ive capped myself – due to having an IM two months after and wanting to be ok for that) my pace dropped off from iirc 1:26 to 1:30.

    I don’t really have a fixed opinion on this but I wonder whether even effort level would reap the best results – which, for me, would certainly involve some kind of negative split.

    More data needed I guess. I’ll have cracked it in 20 years…

  • Dan,

    Much harder to gather data for that, you’d need a lot of marathon HRM files from a single course for consistency. But I would suspect that the practicalities of pacing are you need to be prepared to even or negatively split effort as you say in order to deliver an optimum marathon performance. If the effort levels were correctly chosen I’d expect that to get somewhere from slight positive through to a negative split.

    I had an athlete who ran London for training, took the first half steady and lifted pace in second half for a negative split and an increased HR towards the end.

  • Rob Knell

    Is it not likely that you’ll see an increased HR anyway because of cardiac creep? As I understand it as you lose fluid your blood volume goes down and your HR will increase even if you keep effort the same? You might not see this if you’re drinking a lot but if that’s the case you’re probably overdoing the fluid intake for something like a marathon.

  • You’d see cardiac drift – condition and pace dependent (along with fluid intake) – so an even effort strategy should show up with a higher HR in the second half as would a negative split of effort. To an extent any effort/HR based strategy is going to be inherently harder in the second half due to the nature of the event.

    I do think its possible if suitably trained to make some distinction, however subtle, between a deliberate increase of effort to improve finishing leg and an inevitable sense of increased effort due to increased strain on the body. The good close to even splits and negative splits are likely from successfully raising effort even if it only overcomes the increasing strain.


  • K

    With reference to HR, I ran a poorly paced race, Whe I started I kept an eye on HR and kept the avg at around what is probably my peak HR, (according to Hadd, which seems to fit with my own experience). However the cap should have been lower and I probably had 2 many peaks in the first half and faded badly after 30kish. However HR stayed the same roughly the same throughtout, near the max that I could theoretically manage. RPE however went through the roof after 30k and pace came down. Fitting in with what Dan is saying, I expect if I had used a lower cap in the first half I would have probably paced a more even race, even though HR would haev risen in the 2nd half. FWIW I have gone under 3 hours twice, once with a -ve split and this time, with a 7 minute +ve split (still a pb however).