Matt Fitzgerald on marathon pacing

[Joe Dwarf]

I don't think 10k time is a good use for marathon prediction unless your volume and training are indicative of marathon training, and even then. The 1/2 mary x2 + 10 min is somewhat more reliable.

1/2 mar X2 +10 would have predicted a 3:24 for me last year, which IIRC splits the difference between the McMillan and Daniels predictors based on the 1/2 mar time, all of which was off by 20 minutes or so from my actual. I started out pacing for 3:30, ended up with 3:44. I have no idea if I would have been any faster if I'd started out pacing for 3:45. I suspect not as it was only my 2nd time.

[MichaelMc]

Ultimately what people need to do is make their OWN calculator.

Absolutely. When I use VDOT tables, I typically subtract 2 levels for the jump to 42K. Body weight is another key variable for me, especially in the absence of a recent race result.

I know a runner with a high 17 minute 5k, high 36 minute 10k, 1:20 Half and 3:08 Full. Expressed in terms of 5k pace he runs 10k @ 97%, 21.1k @ 95% and a marathon @ 81%. My results are nearly identical until the marathon where I run 17 MINUTES faster. 10k @ 97%, 21.1k @ 95%, 42.2 @ 89%. He has run many more marathons than I have, too. He would have failed at 5x 10k pace, btw.

I smell a fink. That guy managed to blow up every. single. time. while running 400m track intervals with us one day a few years ago.

True,

I paced him to his 3:08 PR though: it wasn't a blow up (dead even pace to km 36, then sped up slightly).

[jamix]

HcCD I hope your reading this because you'd prove my guess that some people cannot train themselves to set a negative split (AND SIMULTANOUSLY RUN TO THEIR POTENTIAL); You've trained sufficiently for most of your marathons yes? Have any of them ever been a negative split?

[MichaelMc]

HcCD I hope your reading this because you'd prove my guess that some people cannot train themselves to set a negative split (AND SIMULTANOUSLY RUN TO THEIR POTENTIAL); You've trained sufficiently for most of your marathons yes? Have any of them ever been a negative split?

What is your evidence? First "Trained sufficiently" is subjective, but either way the only proof is establishing whether one runs faster while running even splits or faster by going out faster and slowing down on the same fitness. ANYONE can run a negetive split: simply start out slow enough that you don't slow down. The question is whether the result is better or not.

You'll note the majority of people who say it would be slower have never run one that way. It'd carry more weight with me if someone said "I ran a marathon with even splits the whole way, sped up slightly at the end but it was slower for me than going out faster and fading". If a lot of people tried it and proved to themselves that it was actually slower then I'd reconsider whether this is generally the best strategy. Instead I hear a whole lot of "I've never managed to do one that way, so it must not be right for me".

I've run BIG positive splits, small positive, barely positive, exactly even, slight negetive and a couple of minutes negetive: for myself I have a pretty good sample. I've trained with a lot of runners who've had a range of strategies and outcomes, and for a number who thought they couldn't run "nearly even" I've assisted them in doing so. The evidence I've seen fits nicely with the research and my personal experience>. If someone else has similar experiences with the opposite result, then I can understand the contrary position but simply "few run that way, so it must be wrong and might be impossible" doesn't carry much weight.

[ian]

You'll note the majority of people who say it would be slower have never run one that way. It'd carry more weight with me if someone said "I ran a marathon with even splits the whole way, sped up slightly at the end but it was slower for me than going out faster and fading".

May 22, 2013: Bluenose Marathon - Out in just under 1:23, blew up at 30K and finished at 2:53:30 for a disgusting +11 split.
May 29, 2013: Ottawa Marathon - Out in 1:26:mid, held everything together for a 2:53:32 finish for an even split.
The fitness was obviously equal, I have a track record for running well on the second race in consecutive weekends (more often than not faster than the first week), and the Ottawa race had a faster course with better running weather. Therefore I have "proof" that even splits don't always work. Well, not really. I think it's proof that it's unlikely that anyone will change their mind about a strongly held position based on someone else's anecdotes and that our own experiences carry a disproportionate weight.

I've run BIG positive splits, small positive, barely positive, exactly even, slight negetive and a couple of minutes negetive: for myself I have a pretty good sample.

Me too. This got me thinking to look at the correlation between my splits and finish times.
Fastest 10 finishes (2:39-2:50): +1, -2, -3, +5, -3, +2, +1, 0, -1, +2
Next 10 finishes (2:51-2:54): +1, +5, +1, 0, -1, -4, +3, +3, +11, 0
Third 10 finishes (2:54-3:00): +6, +4, +5, +3, +3, +15, 0, -2, +2, -7
Fourth 10 finishes (3:01-3:13): -1, +11, +13, +15, +6, +13, 0, 0, +8, +15
Last 12 finishes (3:16-5:00): +10, +19, 0, 0, +28, -1, +19, +13, +8, 0, +20, +60

Every race has its own story, but the coarse trend is clear that the fastest times are usually associated with the most even splits. I see at least two generalizations that can be made:
(1) As I improved my training and gained experience, it became easier to race more evenly, but the finishing times were improving even with poorly executed races.
(2) The cutoff between "slight fade that probably didn't cost more than a minute" and "big fade that really changed the outcome" feels like about 5 minutes. Along the lines of the adage that the perfect is the enemy of the good, I would advise a goal-driven marathoner to plan to run as evenly as possible but to consider it a success if the fade is only a few minutes.

[Jwolf]

Ian's data is interesting and quite telling. He's run 52 marathons so it's a really good sample size. Yes, does show a trend toward better-raced marathons the closer to even splits. However, there are definitely the outliers. For someone that has only a few marathons it's impossible to see such a trend or form my own "calculator" unfortunately. I just have to go by my limited experience and hopefully traning progress.

[MichaelMc]

(2) The cutoff between "slight fade that probably didn't cost more than a minute" and "big fade that really changed the outcome" feels like about 5 minutes. Along the lines of the adage that the perfect is the enemy of the good, I would advise a goal-driven marathoner to plan to run as evenly as possible but to consider it a success if the fade is only a few minutes.

Or if you only end up negetive by a few minutes?

My last 5 marathons were: -1:18, -0:04, +0:50, +1:40, -0:01

The plus 1:40 I ran with you, and you'll recall I knew the whole way I'd end up postive as we split the difference in our goal paces for the first 30k.

I actually agree with your premise, but I can't count how many times I've heard people stress about not running too slowly in the first half because they wouldn't want to leave anything out there. Once you reach a point where you KNOW you can pick it up you certainly should. If you do that and push to the end then you WON'T leave anything out there (or lose much time).

Nearly even is nearly perfect, and lets face it, the only time conditions won't affect an outcome is on a treadmill: dead even is a fluke. Calgary for me (-ve 1 second) I didn't look at my splits for the final 8k, I just raced it. There were hills on course, turns, a little warmth... anything could have thrown it 30 seconds or more either way. I simply think PLANNING to run even is the best strategy, and I'm good at estimating my fitness. I don't pace to end up negetive, I plan to end up even but if I'm having a good day I can sometimes squeeze a little more.

[MichaelMc]

Ian's data is interesting and quite telling. He's run 52 marathons so it's a really good sample size. Yes, does show a trend toward better-raced marathons the closer to even splits. However, there are definitely the outliers. For someone that has only a few marathons it's impossible to see such a trend or form my own "calculator" unfortunately. I just have to go by my limited experience and hopefully traning progress.

Your calculator should simply be that you have realtively better speed than endurance: until results show different, based on shorter races everybody will project better marathon times than you can achieve. Believe history, not the "experts". That probably sounds like bad news, but it works in both directions: you OVER ACHIEVE on short races given your marathon results!

If you're game, start your next marathon at the average pace of your last one (or just slightly faster if you're more fit): see if you fade or run even.

[Jwolf]

Ian's data is interesting and quite telling. He's run 52 marathons so it's a really good sample size. Yes, does show a trend toward better-raced marathons the closer to even splits. However, there are definitely the outliers. For someone that has only a few marathons it's impossible to see such a trend or form my own "calculator" unfortunately. I just have to go by my limited experience and hopefully traning progress.

Your calculator should simply be that you have realtively better speed than endurance: until results show different, based on shorter races everybody will project better marathon times than you can achieve. Believe history, not the "experts". That probably sounds like bad news, but it works in both directions: you OVER ACHIEVE on short races given your marathon results!

If you're game, start your next marathon at the average pace of your last one (or just slightly faster if you're more fit): see if you fade or run even.

I first have to get my current fitness back up to where it was before my last marathon. Not there yet-- and it will have been so long that I'm essentially starting the experiment over. But you're right- I can use my previous experience to gauge and plan.

[Dstew]

The first time I qualified for Boston, that was a negative split. I had read somewhere that for the first kilometer or even mile you should run at least 10 - 30 seconds slower than your goal pace and despite every bone in my body wanting to ignore that, I was slower. The theory is you are excited and easy to go out too fast and never recover. In fact did that for a nice 4 - 5 K warm up, sped up a little and was close to where I wanted to be within my margin of error at the halfway mark. At 25 K I knew I had it and so with extra energy from that, slowly sped up and when I started to pass other runners who were very encouraging, it snowballed in a good way.

I was on my way for a second positive split the 2nd BQ but saw road kill from a hot day and slowed down. Almost too much.

As far as predicting what pace, my long runs will give me a very good idea. That is go out of a long run and do what feels to be comfortably hard and then see what the pace was instead of running a pre-set pace. I agree with the thought that no matter how disciplined, it is hard not to "cheat" on the various shorter tests and/or some people may be good as Yasso 800s but that does not translate into an accurate marathon pace predictor. Those for example, were never close.

[Jogger Barbie]

Interesting article (now that I've finally read it, instead of just the comments here). I agree with much of what the authour is saying but also strongly identify with the comments here about tailoring training and racing to one's own fitness level, experience, and personal quirks.

On a personal level, my own "experiment of one", I've run (I think) three negative split marathons and had a negative split at the Niagara 50 km. Unfortunately (in a way), only one of those negative splits (Hamilton Road to Hope) would qualify as being the outcome of a true race. And that negative split was helped along by the huge downhill in the second half of the race.

The others were: Ottawa 2010 when I probably had a faster marathon in me but didn't know it, lined up with Canal Runner's 3:45 pace group, and went ahead around the 32 km mark because it felt like I could. Don't remember the numbers but I am almost 100% sure that I ended up with a small negative split because we were on pace but my final time was 3:43ish. Then Scotiabank 2011, which I was running as a training run for New York, and again started with the 3:45 bunny with a plan to stay with him. Then got impatient and felt too good and finished with 3:37ish.

In Niagara, I had no idea what to expect, ran the first half very conservatively by a) doing the first couple km with a friend whose average pace is slower than mine (otherwise I would have gone out too fast), and b) keeping someone ahead of me in sight but never speeding up enough to catch up or pass him. The negative split came mostly from running the first few km of the second half at a pretty quick pace, to get away from the crowds of tourists. I recall the last few km as being absolutely brutal - sunny and hot and l-o-n-g...

In general, I probably do get better finishing times overall by going out a little faster and allowing for some fade in the second half. And that's largely because of two of my own particular weaknesses: 1) a tendency to melt in sun and/or temps that other people don't find too bad, and 2) a way-too-touchy GI system, a.k.a. "mushy guts" that typically don't cause any problems until the second half. A several minutes long porta potty break (or a prolonged slowdown while fighting the need for that break) is a guaranteed way to end up with a significant positive split, no matter how well the first half was paced, or how strong one runs afterwards. For me, anyway, the ability to make up that much time just isn't there. If I have a race day goal, it almost always includes banking a little time in the first half to allow for known issues and lived experiences in the second half.

Having said all that, I've run 15 marathons and with more experience at the distance comes (one hopes) more wisdom. Certainly I agree with the article's point that the body does adapt and that running a given distance does help prepare the body to run that distance again (assuming relatively comparable training). I was thinking about this very point today, during my not-so-great 34 km run. It was wonderful to have it end, but I could have gone another 8.2 km. It wouldn't have been pretty, there'd have been walk breaks here and there, I'd have wanted more water and would have consumed more calories along the way, but it would have been do-able. I would not have felt nearly so confident saying that when my experience was limited to two or three marathons, and a run like today would have left me questioning my ability to ever go that far again.

Ultimately, each race result is its own unique combination of course, conditions, fitness and luck. And I'm using the term "luck" to refer to the things that are beyond my control and that I can only hope work in my favour. More experience has taught me more about what my GI system will do under a given nutrition regime both before and during a race, but I've had spectacular failures (Midsummer Run 2011, I'm looking at you) despite doing everything "right". I've gone into races expecting standard complaints from various body parts, because they've happened on every long run, and they've mysteriously not appeared during that race. Bonus!

My final thought (although this is already long), is that the race day test I always invoke is the statement from our own MichaelMc: at the halfway point of a marathon, you should feel like you could have gone just a little bit faster. It doesn't mean the second half won't be derailed by something else (and it has been), but at least it gives me the confidence that my pace is okay. And I send up a little prayer about the rest and just keep running.

[Jogger Barbie]

The difference with me is that I tend not to believe my overly optimistic friends who seem to think I could or should be faster than I am.

And who would those people be?

[ashleyco]

Its hard to predict because most of our training is general.

[NoQuit]

Nothing to add to this conversation since I've never run a marathon (doing my planning /research now), but thanks to everyone in this thread - it's fascinating and very informative!