Saturday, 6 April 2013

Pacing - Still a Mystery!


Should be a short post tonight, mainly responding to the many comments that have been left during the last week.

Firstly thanks to the many comments left, both the feedback from last weekend's TORQ day and in response to my blog post on hydration, nutrition and pacing.  With regards to the blog post comments, to summarise the comments, it seemed to be agree my hydration observations, pretty well agree with my nutrition/carbohydrate views, but really mixed reaction to my pacing ideas" Hence tonight's post title "Pacing Still a Mystery!.

Just before I move onto the pacing issue, I just want to mention one more bit of evidence from the scientific literature regarding carbohydrate and fatigue.  I have just come across an interesting 2013 article titled "Running Pace Decrease during a Marathon Is Positively Related to Blood Markers of Muscle Damage" which is a free available journal at this link.  I will come back to this article later in the post but one key finding that is reported is as follows:
During exercise, glucose supply for the active skeletal muscle comes from glycogen stores in the muscle and liver. However, if the exercise bout is of long duration (> 1 hour), muscle and liver glycogen stores deplete [17] and blood borne glucose has to be used to provide energy, threatening blood glucose homeostasis [38]. It has been found that hypoglycemia attenuates the activation of the CNS and hence produces reduced exercise performance [39]. For this reason, the reduction in blood glucose concentration has been proposed as a source of muscle fatigue during the marathon [40]. When blood glucose is maintained by ingesting carbohydrates during exercise, muscle force and CNS activation are better preserved [39]. Interestingly, participants in this investigation increased the blood glucose concentration by 0.6 + 0.6 mmol per litre, from pre-to-post exercise (Table 2), as has been previously found in other athletes participating in endurance events [41]. Although we did not record carbohydrate ingestion during the race, previous studies have found that marathoners have appropriate rates of carbohydrate intake [42]. According to our data, blood glucose concentration was well maintained during a marathon in a warm environment, reducing the influence of hypoglycemia as a source of fatigue during this race [43].
So therefore confirming what I have read previously in other journal articles, in that now carbohydrate gels / bars are commonly used DURING marathons, the issue of hypoglycemia (low blood sugar) and glycogen depletion now seldom occurs.  Therefore fatigue during marathons is due to other fatigues, which I believe is when Race Focus Energy is depleted (see my Race Focus Energy / Fatigue Post).  

Back to pacing, as mentioned above their was mixed reaction to my most recent blog post, where I totally 'slated' the concept of the negative split.  One reader commented that they had "never heard anyone say that their goal for a race was to run a negative split".  Well Richard obviously doesn't listen to MarathonTalk (unlike I believe around 15,000 other runners per week!)  If he did, he would then know how this negative split is totally 'raved' about!  There were many comments, including Tom from MarathonTalk, back in December 2011 when I first blogged about the fallacy of the negative split, who present evidence that even pace works for them, or for other people they know including elite runners.  But then others state that a fast start worked well for them.  So it appears simply from this small sample size that no one approach appears superior to the other, and it really gets down to trial and error, and what works for you.  

Although, that does appear to be the 'safe' conclusion, I am not really convinced, so I will continue to try to find evidence to dispel the negative split / even pace myth!  So before I finish tonight's post, a few more comments.   The key focus of the article mentioned above was related to muscle damage, and there finding tends to reinforce what I believe is one of the main cause of fatigue during marathons and ultra events, yes muscle damage.  Their key finding was "Running pace decline during a marathon was positively related with muscle breakdown blood markers.".  Therefore those runners with the most muscle damage, slowed down the most.  However, the mechanism in which the muscle damage slows the runner down I feel is not entirely directly related to the muscle damage, as the muscles are still able to generate more than enough force to run, but indirectly.  Yes, the fatigue will be partly related to the reduced efficiency of the muscles due to the damage, so it will take more energy to produce the same amount of force, so an increase in intensity for the same running pace.  But more important is that the discomfort resulting from the damaged muscles will require increased focus, increased mental effort to deal with this discomfort.  This will therefore increase the rate at which you consume 'race focus energy' (RFE), so you will become mentally tired as your RFE stores deplete, and will therefore be unable to maintain that mental drive, that push, focus to keep running at the hard/demanding pace you were able to manage earlier.  Hence you will slow down. 

 I particularly like one of the comments left during the week, from Chris:
I think with the ultra pacing side of things it's really about finding the correct balance for you as an individual. If you are always going to slow to a certain speed then it makes sense to leg it whilst you can, but the ability to get data with a real true control group is impossible to draw a true conclusion; e.g., what if you had run 10 seconds per mile slower on average for the first 50, what would you have done in the second 50? would you have been more than 10 seconds per mile faster to give you a faster overall time? You'll never know with absolute certainty.  For me this leads to two conclusions (1) you have to experiment to find your right personal blend, preferably on the same course and conditions to give you the best control possible - does running faster to start with make you hit the slow down point significantly earlier? essentially experience counts for a lot and you have to be prepared to fail to get your personal data; (2) assess through experience what happens to you in the latter stages - mental state, concentration on maintaining speed and pushing on (rather than mentally drifting) if prone to mental drifting then you probably adopt a default "knackered pace" so the focus is then on getting as far as possible before you hit this and training to up your knackered pace.
Particularly the importance of "assess through experience what happens to you in the latter stages - mental state, concentration on maintaining speed and pushing on".  Yes, it is really important to reflect on your previous races and try to identify what caused you to slow down.  I did this extensively at the end of 2012, and identified that it was the muscle damage that caused me to slow down, hence my training this year is focused on trying to reduce the muscle damage I experience during the second half of ultra trail races.  Please not that I don't attribute my slowing down in races being due to starting fast, as Sp Lane states "I know it doesn't make much difference if I start slow or fast, I'm still going to die later on.  Hence, it's my best interest to get as far as I can as a short a time as possible."  Exactly, I couldn't have said it better!  

Anyway, I could go on and on, I feel I need more than personal opinion.  So time to get my analytical brain in action, and gather some data, and last weeks Anglo Celtic Plate 100km road race in Perth, Scotland, provided the data I needed.  I took quite a big interest in last Sunday's race as I have raced quite a few of the guys running.  Thanks to John Kynaston's tweets, I was able to keep up to date as the race progressed.  From John's tweets it looked like it was a two man race between running friend Dan Docherty from Ireland, but lives in Berkshire, and England's Thomas Payn, who I didn't know.  Dan's running has really come on in the last two years, so I was surprised when he wasn't leading during the first few hours, with Payn out in front.  John would tweet the leader board every hour or so, so I could see the race positions, but no time gaps.  Then shortly after the 5 hour mark, Dan takes the lead and goes on to win in a very respectable time of 7:05:23.  Yes, that magical 6:59 still eludes Dan!  

Back to some evidence regarding pacing strategy.  Well, the 100 km race consisted of 42 laps, each 2.38km in length, and each lap time was available on the web.  So I downloaded these and got to work!   First I looked only at Dan and Tom's splits.  I was also interested in how far ahead Tom got, and when did Dan hit the lead.  The following to graphs display this data.    

  So you can see both athletes slow down, but Tom's decline in pace was much more dramatic!  Tom started out faster, got up to a 5 min 48 sec lead, and then Dan pulled this back, and got in front on lap 32 or 42.   Next I decided to look at all 31 finishers to see how their pace decline over the 42 laps.    
  Two runners had some really slow laps, and two others had laps slower than 14 minute mile pace, so I deleted these four runners, and carried out further analysis on the 27 other runners, who never got slower than 13:45 min mile pace.   I think it is reasonably clear to see that the trend for the 31 runners is that they slow down during the second half of the race.  I then plotted the average , fastest and slowest min mile rate for each lap, so the average slow down would be clearer to see.    
  This graph illustrates some interesting points regarding slowing down. Firstly it confirms that the average pace slows down a reasonable amount from around half way.  But was is really interesting is that the slowest average lap was lap 37 with an average pace of 9:19, and then speeds up on each of the following last five laps, so the last lap average pace is 8:44.  Now this data clearly illustrates just how rubbish the old physiological model of fatigue was.  I challenge any physiologist to explain what has happened during the last part of the race physiologically!  Hence why a new model of fatigue was required, and I haven't seen any better than my Race Focus Energy (RFE) model.  Yes, a bit of bias there, but if any psycho-physiologist out there are aware of any better models, then please let me know.  

Why is the slowest lap six laps out from the finish.  I propose that the main reason is that as the race draws closer to the end, five laps is less that 12 km, so around 7 miles, the subconscious no longer needs to maintain it's reserve tank of RFE.  Up to this point it keeps stored safely away an emergency supply of RFE because of the unknown, the doubt of what could happen, a safety supply to deal with experiencing the un-encountered,  But with only 7 miles to go, the subconcious is well aware that 7 miles is easily manageable so therefore makes this emergency supply of RFE available, and hence the running pace can increase as there is more Race Focus Energy.  In addition, the negativity from the unknown, the doubt of possibly DNFing, struggling etc. is reduced during this late stage of the race, and this reduction in doubt / fear / negativity swings the RPE - RFE arrow downwards, so able to run quicker for the same rating of perceived exertion (RPE).  

Next for the 27 runners I wanted to get an even more clearer indication of the degree of the slowing down during the race.  To achieve this I decide to break the race into six sections of 7 laps each, so a little over 10 miles per section.  I then calculated the average decline in pace during the 2nd to 6yth section, i.e. each 10 mile split from 10 to 61 miles, in relation to the time split for the first 10 miles.  Here is the graph below.    
  Yes, some interesting percentage slow down data here, with the pace nearly being maintained until halfway, and then slowing down in relation to the first ten mile split, by 9, 17 and 19% for splits 30 - 40, 40 - 50, 50 - 61 miles respectivelyy.  

Lastly I wanted to see what the relationship was between the percentage that the runners slowed down in terms of their first ten mile split and their final ten mile split, and their finishing time.  As it is commonly believed, as stated by Christof "Slowing down has a very profound effect on finishing time, so a near flat pace theoretically produces the fastest result."  If this myth actually true.  Or as I belief, the only reason anyone actually achieves a negative split, i/r/ doesn't slow down during the race, is that they have run the first section so very slow, that the time they have lost during this section is never regained!  What does the data show us.  In order to try to relate directly to elite ultra runners, I decided to delete from my 27 runners, any that did not break 10 hours, so four more runners were deleted (sorry if you were one of these four runners!), and the slowest finishing time from my sample of now 23 'elite' runners is now 9 hours 37 minutes.    
  The graph above clearly suggests that Christof's statement is incorrect (sorry Christof, but you are most welcome to reply and counter my evidence), with a correlation of only r = 0,23.  And if one looks at the coefficient of determination (R squared), which is thought to indicate the amount of commonality between the two variables, it would suggest that only five percent of the factors that contribute to ones finishing time can be attributed to the percentage one slows down during the race.   So five percent is a very low percentage, barely above zero, but the correlation is a positive correlation, which indicates that the more you slow down during the race, the slower your finishing time.  Which is rather disappointing for me in providing evidence that "Run as fast as you can, while you can" is the best way to go.  I would have need a negative relationship between the two variable in the above graph for my approach to be supported by the data. 

However, time to cherry pick.  If we look at the winner, Dan's ranking in terms of slowing down during the race, he ranks in only 16th place out of 23 runners under 10 hours,  in terms of slowing down the least.  He has a massive slowing down percentage of over 17%, but yet quite easily wins the race!  One runner, Antonia Johnson, 3rd placed women, actually run her fastest split of the race during the last 10 miles.  Second place male Craig Holgate also displays some, in my view, less than ideal data.  How much quicker, and higher up would have these two runners finished if they had adopted a less 'strange' pacing strategy???  None know, hence why pacing is "Still a Mystery!"  

Well enough for tonight, it wasn't as brief as I thought.  They never are!  Time to sign off, again from a quote from Tom from Marathon Talk.  

"I'd rather know I was wrong than think I was right"  Tom Williams, 2011.  

Will we ever know in terms of pacing, what is the wrong or right strategy?  


PS  You may recall that within my recent Endurancelife Sussex Coastal Trail Marathon posts I mentioned Scott Forbes, who finished in third place after he started earlier with ultra runners and the slower marathon runners.  Well a bizarre thing happened the other day.  I was sorting out a few filing cabinets / folders in my office and came across an old 220 Triathlon magazine from 1996.  
  The reason that I had kept this specific issue of 220 was that there was  photo of me and my wife Frances on our wedding say in it.   
  Anyway, I started flicking through the pages and got reading the article on the French Triathlon Tour that was a pretty unique event that had taken place in 1996, which involved the top triathletes in Europe, racing a triathlon every day for over a week!  I then noticed that  Scott Forbes from the UK finished in 50th place.  
  Now this got me thinking, could this Scott Forbes from the Sussex Coastal Trail marathon, be the same Scott Forbes, the elite triathlete from the nineties.  Chatting to Scott following the race, he mentioned that he was just getting into ultra running after previously being a mountain biker, but he had suffered a horrendous bike accident, breaking his neck along with many other injuries, and had decided to try ultra running.  I then remembered that back in 1996 while I was studying for my Masters in Sports Science at Loughborough University, every Wednesday at 1:00pm I used to go out with the triathletes and cyclists from the University.  There used to be around 20 or so riders, and we would go for around 3 - 4 hours, staying in a group for the first 2 hours, and then it was flat out coming back.  It was awesome fun.   

Anyway, I began to remember that on one of the rides I got chatting to one of the young students who mentioned he had raced the French Triathlon Tour, and that he was an elite triathlete, which he clearly would demonstrate on the ride back, as the pace got substantially quicker and quicker.  Could this guy I used to ride with most Wednesday afternoons nearly seventeen years ago, be the same guy I raced last month?  Well a bit of searching on the internet, and yes.  I discovered a few  interesting articles on Scott, and it is the same guy.  Amazing, what a small world it is.  And Scott, if you are reading this, hi again from 1996.  You may remember me, I was the ancient 33 year old guy with a faint NZ accent on the Wednesday that used to somehow keep up with the front group all the way back to Loughborough, even though it looked like I was really suffering and dieing.  I'm sure you will remember me with that description!   

As I mentioned after the race.  Come down to the Classic Quarter 44 mile race in June, down in Cornwall, and we can have a good battle there, assuming you don't start at the wrong time!  Sorry, about that dig.  I'm just excited about the prospect of 'pay back day' for the suffering I endured on those Wednesday afternoon bike rides in 1996/97!  Yes, as I mentioned earlier, those rides were great fun!  The interesting article on Scott an airline pilot, is at this link  and a bit more about him as a mountain biker here.  Keep an eye on his name, I'm sure it will crop up in future ultra trail results, hopefully at this June's Classic Quarter!  

PPS I got a phone call yesterday from a company involved with marketing Asics kit.  They had been to the new Trail Running Sussex website, as well as to this UltraStu blog.  They thought that visitors to both the Trail Running Sussex website and my blog might appreciate the latest promo video they have just released on Austrian Ultra Trail Runner CHRISTIAN SCHIESTER.  So I took a look at the video, and it is pretty inspiring, with some pretty good footage, so here is the link.  

But then also today I came across the Guardian Article on Jez Bragg's amazing New Zealand adventure, and if you haven't seen this article, follow this link.  Within the Guardian article there is a link to some fantastic footage of New Zealand.  So two great videos to watch.  

PPPS I have attempted to put a link to my 100km analysis excel file, but it keeps saying file corrupt when downloading!  If anyone would like a copy of the excel file simply zap me an e-mail at or my new e-mail address  

PPPPS Finally time for bed!


  1. My analysis of a couple of years of Highland Fling splits last year showed that faster runners on average ran slowed less in the second half. There was a great deal of variation though, just as you found with your analysis.

    Curiously your splits were pretty even compared to the average, so even with your strategy of going out fast you still slowed less than the average runner. However, compared to other elites your splits suggested more than typical slowing. If you look at your blog article on your previous Flings your HR trace showed that your physical effort level in the second half was less than your first half, and your faster time had you slow less in the second half and had a more even HR split too. Taking this evidence together it points to you going out to fast and unable maintain the intensity level.

    To tease apart the mystery of pacing I think it would be useful for you to look more deeply at the plethora of reasons that can cause a runner to slow excessively in the second half. The range of things that can go astray during an ultra are probably why the split variations can be so large.

    Getting everything right for the whole race is incredibly hard, perhaps part of being a good ultra runner is coping well when things don't go quite right.

    As for the drop in pace prior to a surge in pace in the last few miles of the results you have, my guess is that it's a lull before the second wind kicks in/last ditch effort commences. Whether this is conscious tactic or subconscious manifestation of the central governor at work I can't say but wouldn't be surprised if it's a bit of both. For instance stopping/slowing for a last feed/drink before the final push will be something that most runners do.

  2. Sorry to harp on about the negative split thing, I should have been clearer in my first comment. I meant that I hadn't met anyone who had stated a goal of running a negative split in a marathon. It seems an odd thing to aim to do. "I'd like to run the second half of a race more quickly than the first." Really? No, really really?

    I've heard lots of Marathon Talkers really pleased with doing it and I'm pleased myself to finish any race strongly. Mainly people have a time goal whether that's 4:59:59 or sub-2:30 and they state it in those terms. Even pacing seems sensible in flat races at least even though I quite like the idea of "go as fast as you can for as long as you can" heroism. I'll try it once and see what happens. Dying on your arse towards the end of a race is a familiar feeling for all of us at some point. Might as well do it good and proper.

  3. Nice one Millsy - enjoyed your analysis. Mike V

  4. It has occurred to me that a substantial flaw in trying to reading too much about pacing w.r.t once race and individual runners is that it's a race, and with an ultra often the winner might be decided by critical moves in the second half that change the dynamic.

    Once a leader has shown that they are untouchable challengers may back off because they feel it's not worth chasing.

    Also the leader may back off if they feel there lead is safe and either back off because they don't feel threatened, or that rather than risk blowing up and keep on pushing hard they ease back and keep something in reserve.

    This dynamic of chasing/pushing ahead/easing off due to position relative to other runners dramatically effects the pacing, and may go along way to explaining the variations between laps and competitors.

    Runners that are chasing PB's or course records are more likely to run a more even race and push right to the end. In these cases I'd expect a faster finish.

    If you want to analyse relative merits of pacing strategies from races then I believe it's crucial to look for those chasing PB or course records rather than those running tactically. Even within this group you'd still see affects on pace if the runners knows that they are going to miss their target, or comfortably exceed it.

    The aspect of pacing and race tactics is also an interesting area, this provides quite different motivation for pacing.

    So... I would suggest having another look at the data, look to spot the runners that ran more evenly and less affected by the pace of others i.e. their were running their own race and likely with specific finishing time/PB in mind, these runners would likely provide a better base for your analysis of best pacing strategies.