How to pace the perfect Marathon or Ultra-marathon

In search of Optimal Marathon and Ultramarathon Pacing.

The general rule for pacing that is preached by the best coaches and sports scientists is an even split or slightly negative is optimal, and this is prescribed for elite's and every day runners, but is this advice sound?  I'm not one to take things on face value, but as an inquisitive sort I've thought about the topic extensively and tried out things in my own racing.  The more I've looked into the topic the more convinced I have become that this approach is sound.  This view is based on analysis of what elite's use, and also my understanding of physiological demands that the runner is placing themselves under.

Previously I wrote about elite marathon pacing is my post  2013-London-Marathon-Elite demonstrate folly of Positive Split.  Another great discussion about elite marathon pacing can be found the Science of Sport website, relevant articles : Wilson Kipsang’s marathon world record: Pacing and splits, Geoffrey Mutai: 2:04:15, misses WR, Haile Gebrselassie World Record 2007.  There is also a convenient thread for all their marathon analysis.  If you real all these bits of analysis it becomes clear optimum pacing for elites is even pacing, with big penalties paid for times when athletes surge or push on too hard at any one stage.

Another hint that even/negative pacing is provided in a quote from an Jez Brag in an interview published in Ultra Tales:

"Jez Bragg: From my experience the best way to find the ‘magic’ ultra running gears is to be patient, and start steadily. Even pacing, or ideally running negative splits, is the ideal approach to achieve this. By ‘magic’ gears I mean when you settle into a strong pace that feels effortless and like you can go on forever. It does happen; it is possible. That’s how the really special ultra running performances are achieved. The bottom line is that it’s very individual and it’s a case of finding out what works best for you."

Not only does Jez say even pacing is best, but this is exactly what he does - his course record on Devil O'Highlands is almost perfectly evenly paced. Last year I did some analysis Devil O'Highland splits, but alas been too rushed off my feet to write it, the short story of this analysis is Jez' course record stands out clearly as the strongest second half ever run on the route.  The pacing that female Devil O'Highlands record holder, Lucy Colquhoun, was also very close to even pacing.  This pattern is repeated in analysis I've done of the Highland Fling as well. Whether it's marathon world records or local ultra marathon records, the theme is consistent:

The best of the best run with very close to even pacing.

Muscle Fatigue and Muscle Damage

All marathoner will suffer fatigue and portion of this will be down to muscle damage.  Damaged muscles are less efficient so require more energy to generate the required forces, and also results in greater perceived effort level to maintain force output and pace.  Not only does this chime in with my own experience but studies have also shown how muscle damage accumulates through longer races, I guess it's one of items that is common sense and common experience.

I believe the amount of muscle damage and general muscle fatigue that we accumulate is directly related to pace that we have already run at for the preceding miles.  Running at 7:00 min/mile pace for 13 miles leaves the body with much greater muscle damage and glycogen depletion than running at 7:30 min/mile for 13 miles.  I do not believe the relationship to be linear, running 5% faster won't lead to 5% more muscle damage and glycogen depletion, but will lead to more than 5%.  Exactly how much I don't know, but I know for sure from my experience training and racing it's not linear, going a little bit faster can create a lot more damage.

In my articles from last year on the effect of fat vs glycogen utilization and  pacing strategies on glycogen utilization I touched upon the non-linear relationship between glycogen and intensity, one of the graphs that I included show how rapidly glycogen rises as intensity rises (this is based on study data of well trained athletes), I'll reproduce here so you can see how small increases in intensity have a big effect on how much fat and glycogen(carbs in the graph) you'll need:

My guess us that rate of glycogen depletion and muscle damage are correlated, although a causal link might not be direct.  The bottom line is the faster you run the faster you deplete glycogen and the faster your damage.  The fact that the relationship is non-liner it's likely we'll pay a dis-proportionate amount for going too fast over any particular section.  This is something that can been with the elite marathoners - an surge too early or too hard can blow a field apart and often kill the chances of the one pushing the surge.

Could a Positive split optimal?

Positive split has some advocates, even if they are far outnumbered by the those advocating event splits. In the British ultra-marathon scene Stuart Mills is one such advocate of positive splits with his "run as fast as your can for as far as your can" philosophy of pacing.  Could this approach be view be optimal?

A positive split strategy deliberately has the first run faster, but this does mean you are creating more more muscle damage and greater glycogen depletion in the first half of the race than an even or negative split strategy.  With more muscle damage you physically have to expend more energy per mile to cover the second half.  With greater glycogen depletion you also have to draw that greater energy contribution from your fat reserves, which in turn means greater oxygen load as well as greater perceived effort level to maintain pace.  Carrying a lower efficiency over the second half of the race will mean that the total energy requirement for a positive split will be greater than that of even or negative split.

Now Stuart's hypothesis is that in response to this muscle damage will enviably slow down and should plan for it.  Stuart's uses his own analysis of London marathon results that show the vast majority of runners including elites positive split, and my own analysis of ultra-marathon also support the observation most athletes positive split.  What Stuart then claims that this means positive split is optimal.  I can't figure out why he believes this to be sound conclusion.  To me it's akin to saying because on average the UK population is getting more overweight we should all therefore should aim to gain weight. To me all Stuart's analysis proves is that on average runners slow down in marathons, however, it simply isn't at all useful at telling us what is optimum - you have to use other types of analysis for this.

From a theoretical standpoint I can't see how running a positive split could be beneficial.  It requires more energy due to having to carry greater muscle damage through the second half of the race.  It requires you to run faster at the start of race before your body has fully warmed up and moved from anaerobic to aerobic metabolism wasting precious glycogen.  There is also a physiological price to pay - slowing down is demoralizing, as is being passed by runners who are looking stronger and fresher, you also have to carry yourself on fatigued legs that with every step eat into your soul and will to maintain the pace.  Going out fast and then hanging on is really tough mentally as well as physically.

Having followed Stuart's blog for a few years I do wonder if this last observation may be a key to understanding why he advocates a "Going out as fast as you can for as far as you can" - could it be this pacing approach makes you feel like you've run the absolute hardest and put in the most effort you could have at all points in the race.  It makes you feel that you put your all into all, you suffered for your performance so it must be the fastest you could have run.

This logic ignores the fact that such pacing causes greater muscle damage, dehydration and glycogen depletion and erodes your mental state, so while it might feel that you are running as fast as you can at any point in reality have to slow down more in the second half than you save in time by running fast in the first half.

Stuart's performance can be a bit hit and miss.  Some races he runs well and has won some huge races, but these all balanced with other races where things don't go right, or where it seems like after one third or half way he simply gives up racing completely.  Stuart often puts this less successful races down to lack of mental as much as physical preparation, but perhaps the pacing strategy is high risk and contributes to these days that don't go so well.  For these races that don't go well perhaps "Run as fast as your dare, or as far as you care" might be a more suitable signature.

I haven't yet seen a compelling argument why positive split might advantageous, and I've yet seen any good real world data to illustrate that it is appropriate for any one, let alone the average runner who is mostly likely to suffer serious muscle damage and fatigue from pushing pace early in long race.  I can't help feel a bit frustrated with Stuart.  He is massively talented runner that has so much potential, but for me I feel it rather squanders it on a less than rational pacing strategy.

What might be optimal pacing look like in practice?

So what should the runner do?  Even or slight negative split looks best from a theoretical standpoint (looking at muscle damage + glycogen depletion, heat stress and dehydration), and is backed by what pacing strategies that world top long distance athletes use when they run their best races.

In practice  one should run the first 15 minutes slower than your average pace to allow you muscles to warm up to temperature that it operates mostly efficiently at - start too fast an you are far more anaerobic and burn precious glycogen and increase muscle PH - which in turn leads to more fatigue and muscle damage.  Once past this easy start and your are fully warmed up you then steadily approach your average pace/intensity that you then aim to maintain for the majority of the race.  If the race has been run well then in the final miles it's then possible to pick up the pace a little and run a strong finish with what glycogen stores are remaining.

While I believe this to be an optimal strategy is hard to achieve and most runners aren't taught or practice pacing enough to do it.  Running the first few miles at slower than your target pace is especially hard as adrenalin is pumping and everyone else around is tearing off.  Adrenalin keeps hiding the true cost of the damage you are doing to your muscles and your prospects of running a good marathon - adrenalin prompts the the release glucose from your liver so you feel a buzz of energy, and reduces perception of pain and fatigue, which is great for a 10k, but not really what you want for a marathon.

For a marathon you need to make sure your glycogen stores in your liver don't empty too quickly, but are called upon steadily through the race.  If the body spots that liver glycogen levels are getting too low it actually creates the emotion of fatigue and slows us down, from their on it's a battle to stay focused and keep the adrenalin pumping because if you don't your body starts shutting down.  Once the adrenalin starts subsiding and the true state of damage and glycogen depletion is revealed it can be rather difficult to keep going.  I believe this is the wall that so many marathoners hit - it's not actually a hard limit of glycogen depletion but the central governor shutting things down as things have gone too far out of homo-stasis at too fast a rate.  These are the key reasons way it's so hard to run a marathon well.

How might an average runner achieve optimal pacing?

So how does an average runner run the optimal pace?  Being realistic, practice, and using tools.

Being realistic is really down to judging just what time you are likely to be able to achieve.  Marathon time calculators like the McMillan are based on what elite runners can do rather than average runners.  Average runners rarely have the aerobic, structural and metabolic resilience to be able to convert their 10k times to great marathon times like the elites.  This average lower reliance than the elites means you need to be more conservative about your time goals.

Practice is about running the target race pace regularly, getting a feel for the breathing, cadence and general feel for that pace.   As you approach your peak training and go into taper you need to speed more and more time at this pace to dial into it.  Dialing into the intensity/breathing is important is you need to run any hills at the same intensity and not judge them by pace.

Tools, here is something that may well be the crucial bit to helping elevate an inexperienced runner into a pacing god.  Heart Rate (HR) monitors are great for telling us just how hard our bodies are working.  While adrenalin may hide our actual effort level making fast pace seem easy and HR monitor will tell us exactly what is going on.  It's the face pace that does the damage so if the HR monitor values are high is indication that are likely incurring more damage than we are aware of.

HR monitors aren't flawless  - they can mis-read at times, and the thanks to natural HR drift the HR trace for optimal even pacing leads to a steady increase in HR from start to finish.  The mis-reading is typical down to poor/compromised contact, wetting a HR strap properly helps avoid this.  With the HR drift just what HR to aim for each stage is something that is likely to be personal, different runners experience different amounts of HR drift, and adrenalin can also elevate the HR for a given pace so if it goes up and down through the race what to aim for can change.

As a starting place one can use the HR guide suggested by the MARCO calculator:

    http://feelrace.com/fr.pl?th=MARCO

From the Loch Katrine marathon that I ran 10 days ago I had a bash at using the MARCO HR guidelines. In a practice run 6 days before the marathon I ran a half marathon route on similar undulating terrain using the HR profile suggested by the MARCO calculator.  What I found is that my first mile was 15 seconds per mile too slow compared to the MARCO calculator, but in the second half on my run I had to run 30 seconds per mile faster than my planned pace.  This trail run told me that my starting HR needed to be higher and my HR drift was slower so I didn't need to target increasing my HR through the race at quite the same rate.

Race day is where it counts, does pacing by HR work?

Come Loch Katrine race day I used my HR monitor and the general principle of increasing HR through the race suggested by the MARCO calculator with tweaks based on my experience with my HR trail run, and also tweaking it on the day.  My aim for the day was to use the marathon as a training run so not push on too hard, and if possible set myself a PB with a target time of 3:30.

The Loch Katrine marathon is extremely hilly so each mile I either got ahead or behind my target average pace of 8min/mile, so I had to trust that pacing would even out.  I hit the turn around point in 1:45:09 which is far more luck than judgement to get that close, but getting so near to this just by using HR and tweaking it little bits each mile to how things were progressing is what got me in the right ballpark.  The tweaking I was doing was to lower my average HR I was targeting for each mile as my HR drift was less than I expected even compared to my half marathon trail run.

Thanks to less hills in the miles after the half marathon point by mule 16 I got ahead of my 8min/mile target without needing to up the intensity, but once I hit the hills between mile 17 and mile 22 my average speed dropped and I down a couple of minutes my the top, but as the downhill miles came I gradually got back on target.  All the while I was using my HR monitor to judge whether to slow down or speed up.

Once I got to mile 23 I was only just behind my target pace, and although it was only meant to be a training run I found myself picking up the pace, really just for fun.  My energy levels had been rock solid all through the race, my legs were quite fatigued from all the downhill miles but I overall I felt great so just pushed on.  The nearer I got to the finish the more I took the breaks off and ended up finishing the last mile in 6:10, and the last quarter at 5 min/mile pace.  Within the crazy finish I would have hit my target of 3:30 quite comfortably, but with the fun last few miles I did it in 3:26:50, nearly a 6 min PB on much hillier course than the Edinburgh marathon course where my previous PB of 3:32:26 was set.

Now I am capable of running a faster marathon, as I wasn't racing it, so I ran the first 23 mile at a slower pace than what I likely capable of.  Running such a big the negative split I did thanks the overly fast last mile isn't optimal in any way shape or form - this occurred because I was playing, not racing.  If I were to race a marathon I'd run the whole thing close to same intensity and pace.  I would certainly struggle more in the last few miles if I were racing simply because of the accumulated fatigue associated with running faster, but that's racing.  My target for sure would be even intensity and even splits.

When scaling up to ultra-marathon I have also found running by HR to be really beneficial.  In my three ultras last Autumn I raced them all far better than my previous ultra record had suggested likely.  One part of this improvement was changes in my diet to higher fat/modest carb diet so I have become better are burning fats and sparing carbs, but a big part of it is getting my pacing right - in my case by running to a constant HR. Targeting a constant HR lead me to run the first halves comfortable and then finish strong.  In my analysis of Devil O'Highland splits my own first half/second half pacing had a small positive split, but 0.7% different pacing than the winner, who just so happened to also run the strongest second half of all the people last year, and the closest to the pace that Jez and Lucy used.

Could it be coincidence that I paced in a very similar way to the best of the best or is it an indication that they run with an even pace and that running with a near constant HR is a good proxy for this intensity and pace?  Personally I can't think of more compelling indicator of just how effective pacing by HR can be.

Probabilistic pacing

Since even and negative splits are common mantra's for runners, me advocating the same really isn't novel.  My positive experience and advocacy of pacing by HR aren't unique either.  Before I wrap I'd like to put forward an idea/why of thinking about pacing that I haven't see discussed before - Probabilistic Pacing.

What is Probabilistic Pacing?  My idea is that we can view our race statistically, where each mile we run and how we run it effects the probability of achieving a particular goal be it time based, place based or just ability to finish.  What you do in each mile can improve your chances of succeeding in your goal or diminish it.   I believe we can simply view this as at high level of conscious thought - i.e if I push on too hard up this hill I increase the risk of cramp on the other side, or depleting my glycogen reserves too much, or to decide to hold back on descent to avoid going over an ankle, this high mental level of Probabilistic Pacing strategy could be partly thought as our Tactical brain.

I don't believe this idea of Probabilistic Pacing is just high level associated with conscious thought processes, I believe the Central Governor Theory suggest that at a sub-conscious level our brain/central nervous system is making calculations all the from external and internal factors to judge whether we are pushing our bodies too far out of homo-stasis as to be safe. If the Central Governor judges what we are doing as potentially damaging or dangerous then passes on it's assessment as the feeling as pain and fatigue to the conscious part of brain.  The further out of homo-stasis we get, or look likely to get, the stronger the Central Governor will dampen our ability to push on, hitting the wall/boning is likely more a manifestation of Central Governor taking charge than actual complete exhaustion.  Conversely if Central Governor judges what we are doing as safe it leaves us alone to run a good pace.

The high level tactical side to Probabilistic Pacing is really about us consciously setting a particular pace that will most likely get us to finish in one piece and at our goal pace/position, as well as working out what our body needs right now to maintain it's workload and also what we'll need next.  In an ultra-marathon this type of mental process has to include thoughts about navigation - do I need a check the map, is the route clear/familiar enough for me not need to check?  Do I need to eat now or later, do I need to drink?  Do I need to pick up some more clothing at the next check point?  Will I need my head touch?  In all these questions we are weighing up the probability that different help or hind our progress.

Probabilistic pacing relates to how fast or slow to run each section in that if we run faster we increase the probability that we'll deplete glycogen more, increase heat stress, sweat more, be able to digest less so absorb less calories and less fluid - running faster reduces can even reduce the probability of finishing at all.  If we run slower we help our bodies stay closer to the safe homo-stasis that it loves, and increase our chances finishing significantly.  However, if we want to achieve a personal best then we have to make a judgement and work out just how much risk are we willing to take with crashing and burning before the finish.

Now the Central Governor is also making these computations and if it decides to take control and cause massive fatigue then it doesn't matter how much me want to finish in certain time, if it feels that you are endangering yourself then it'll call a day and we can very rapidly feel massively fatigued and unwell. The Central Governor does like suddenly changes, non linear events are very hard to judge so it's natural for our safety systems to be conservative.  However, if we keep things smooth and linear our Central Governor is able to trust in what we are doing more and less likely to shut things down. An analogy that springs to mind is that boiling a frog, if you toss it into boiling water it'll jump out, while if you lay it cool water and let it slowly warm up it won't notice that things are getting hot.

Even pacing has the advantage that our bodies systems are steadily taxed and in way that is linear and predictable.  If the Central Governor sees that we are steadily depleting glycogen but at the rate we are burning it we will still have a safe amount left in our liver by the finish then it'll be happy and let us continue.  If we start fast or do a surge we inject a rapid and non-linear increase in rate of glycogen depletion, the heat stress on our bodies rapidly rises, our reserves of fluids get more rapidly depleted, again it's non linear, a modest increase in speed and we have to sweat much more, all of this is noticed and taken into account of by the Central Governor and makes it all more like to decide to create the fatigue that slows us down.

Listen to your heart

This probably all sounds rather complicated.  How are we supposed to use this in our next marathon or ultra marathon?  My advice is simple, listen to your heart.  I don't mean in the happy clappy, spiritual way, I mean quite literary, your heart is one of the best measures of physiological and physiological stress. The fantastic thing is we now have HR monitors that allow us to track ever beat.  We can now look down to our wrist and immediate now just how hard we are working.

With pacing by HR the absolutely simplest way is to run to constant HR.  Due to HR drift this will lead to your slowing through the race, but as long as you pick as sensible target HR this slowing shouldn't be too dramatic and just lead to a modest positive split, so it might not be optimal but it will be not far off.

A refinement on just running against a constant HR is to factor in HR drift so that your pacing goes from a small positive split back to a even or slight negative split.  The MARCO online calculator can be used as a guide for marathons, but as I've found out the HR drift it assumes may not be appropriate for you.  Using your HR monitor in training and seeing how pace and HR drift are associated can give you an idea what is appropriate for yourself.

The nice thing about using HR as guide to pacing is that it works for ascents and descent telling you exactly how hard it's safe to push on up and hill or how fast you can safely run the descent.  It also tells you when heat stress is affecting you, if it's a hot day your pace for a given HR will be lower, but as we sweat more and use more glycogen when we are hot we can't race as fast when it's hot anyway so having the HR reading guide us to slowing down is exactly what we want.

If you are using your HR monitor as a guide during a race and find that you a feeling better than expected, or worse then your can up you target HR slightly to adjust things to speed up or slow down. I used this strategy when I did my three ultra's in the Autumn of 2013, each time I was hit with cramp I dialled by the intensity, reducing my target HR by 5 bpm till any further hints of cramping had subsided then I let the HR target rise back up.

One can also look at the average HR you set as a target in terms of probabilities - which brings us back to Probabilistic Pacing concept.  Targeting a higher average HR will lead to running faster, but increase the risk being unable to maintain the pace or worse failing completely, targeting a lower average HR will lead to slower time but much better chance of finishing.  Judging the right HR to use on the day is difficult, but if you've got a HR record of previous events then you can use the average HR as a basis. If you don't feel as strong as the previous race then it is probably sensible to target a lower HR and if you feel good in the second half up the HR value.  If you feel stronger then a higher average HR might be attainable, but... if you are fitting your HR for a given pace will be lower.   Through tracking average HR and drift from training and other events look for where you feel might be appropriate come race day.  Then on race day itself adjust as necessary for how you feel.

By using your heart as guide what you do is provide a nice stable physiological load on your body that enables your Central Governor to work with nice linear changes in your homo-stasis, it doesn't get any sudden shock and doesn't panic such as making like you feel like you've hit a wall.   By working so closely with your body it will allow you to achieve much more.

Listen to your heart and your body and mind will be strong.