Top Tips for your Best Marathon

What a large-scale analysis of marathon data says about how to run your best race.


  1. Don’t go out too fast; you’ll finish more slowly and probably hit the wall.
  2. Steady = Fast; even pacing leads to better finish-times.
  3. Run with pacers; they’ll help you run an evenly paced race.
  4. Don’t plan for a fast finish; you’ll finish slower with too much in the tank.
  5. Practice makes perfect; the more marathons you do the better you race.


I’ve been working on a large-scale analysis of marathon data and as peak Marathon season is upon us I thought it timely to share a summary of my key findings, and what they have to say about how we should race for this season’s personal-best.

All of these ‘tips’, such as they are, are supported by real data involving hundreds of thousands of runners and they are based on a growing series of articles published at Running with Data. Where approrpiate I will provide links to the detailed analysis supporting each of these tips.

Tip #1 — Don’t go out too fast.

Image courtesy of

Don’t start too fast is one of the most common pieces of advice we hear when it comes to running the marathon. Does the data support it. Oh yes!

  • Starting too fast can wreck your finish-time; for Dublin and Chicago marathons the data suggests that going out too fast can add 40 minutes or more to your finish-time, and the faster you go, the worse it gets.
  • A majority of people who hit the wall will have started too fast. For instance, about 50% of Chicago marathoners who’s first 5k is their fastest go on to hit the wall. In Dublin, over 70% of those who’s first 10k is their fastest hit the wall later.
  • Despite this one-third of all runners (male and female) go out too fast. But mostly it’s first-time marathoners who do this (70% of first-timers start too fast), and as we run more races we are far less likely to repeat this mistake; only 20% of people running their second race go out too fast.

For all the gory details on this tip check out my analysis of fast-starters at Dublin and Chicago marathons.

Tip #2 — Steady = Fast. Even pacing wins the race.

Image courtesy of

There has been much debate about the best pacing strategy for the marathon. Should I run at a more or less even pace througout the race? Or should I do a positive-split, running faster in the first-half and hanging on as best I can for the second-half? Or should I try a negative-split, running more conservatively in the first-half and then speeding-up for the rest of the race?

The data suggests that more even pacing (an even-split) is likely to produce the best finish-times, so aim for an even-split if you can:

  • While even-splits are rare (approx 5% of all splits) they are more common among faster runners; interestingly females prove to be significantly more disciplined than men for all finish-times, running more even-splits and a more even pace in general.
  • In the Dublin marathon, those running even-splits finished about 15 minutes faster than those running negative-splits, and 20 minutes faster than those running positive-splits.
  • In Chicago the finish-time data were similar for the difference between even and negative-splits, but in Chicago positive splitters were almost 40 minutes slower on average than even-splitters.

Further details and analysis on this topic can be found in Running with Data for Dublin and Chicago marathons.

Tip #3 — Run with pacers, at least initially.

Image courtesy of

Of course knowing that running a more evenly paced race is desirable and running an more evenly paced race are two entirely different things. That being said, one way to improve your chances of running an even-split and obtaining a strong finish-time is to run with a pace group. Here’s what the data from Dublin says:

  • Pace-groups matter: there are large spikes in the number of runners finishing at paced finish-times compared to unpaced finish-times. In Dublin, 60–100% more runners finished just before paced times than for any other nearby times.
  • Changing the paced times, changes the finisher spikes; in other words the finisher spikes track the pacers not the times!
  • Running with a pace group makes you up to 5 times more likely to achieve an even-split.

So, if you have a target finish-time in mind and it matches a paced time then running with that pace-group is probably smart. If there is no pace-group for your precise target time then start out with a pace-group that’s a few minutes slower — this will keep you evenly paced early on — and try to pull away later in the race for a slight negative-split. This is likely to be a better strategy than starting with a faster pace-group and slowing later for a positive-split.

Tip #4 — Don’t plan for a fast finish.

Image courtesy of's_400_metres#/media/File:Warinerbeijing.jpg

On the face of it it may seem that a fast finish must surely lead to a better finish-time but the data does not support this. Here’s what we have found:

  • Runners who complete the last 2.2 km (1.36 miles) of their race at or near their average race-pace have faster finish-times than those who finish faster or slower than their average race-pace.
  • Fast finishes are especially injurious to your finish-time. For example, based on data from Chicago, those who finish 10% faster than their average race-pace have finish-times that are about 15 minutes slower than those who finish at their average race-pace.
  • Slow finishes also degrade performance but less dramatically. Again, based on data from Chicago, we find that runners who finish 10% more slowly than their average race-pace tend to add about 3 minutes to their average finish-time (compared to those finishing at their race-pace).

The likely explanation for this is that our ability to produce a fast finish, or our inability to maintan our pace, is a proxy for how we have run our race overall. Runners who manage a fast-finish do so because they have something “left in the tank” and, given that their finish-times are slower, chances are they have too much left in the tank because they have paced their race too conservatively. The opposite is true for slow finishers: they can’t even maintain their race-pace at the end because — they have nothing left in the tank — probably because they ran too fast too soon earlier in the race.

Note: I have not yet published the details behind this particular study of finishing paces. I plan to do so in the coming weeks. For now the above summary captures the key findings and I will provide a supporting link in due course.

Tip #5 — Practice makes perfect.

Image courtesy of

In the words of Kate Bush and Peter Gabriel: “Don’t give up.” The thing about marathons is that we generally get better with practice. Here’s what the data has to say:

  • You should register for more races in the future. First-time marathoners struggle compared to those who have at least one marathon under their belt. Compared to repeat marathoners, first-timers are slower, their pacing is more erratic, and they hit the wall more often, in fact about twice as often as repeat marathoners.
  • Each additional marathon you run can knock 5–10 minutes off your finish-time, based on data from Dublin and Chicago, at least up to a point. As we get older we do tend to slow-down but marathon experience is a great way to roll-back the years. Repeat marathons can take years off their marathon finish-times. For example, Chicago marathoners in their 50’s can produce finish-times that are faster that first-timers who are 10 years younger or more.
  • Even with just one marathon behind us we are about 33% less likely to hit the wall and the more races we run the less and less likely we are to hit the wall.

So, if this is your first marathon and it has gone well, then congratulations! Your next one will be even better. If it didnt go so well then don;t worry, the experience will benefit you, and your next race will be a lot stronger.

Enjoy the Race!

I hope these tips and insights are useful. I’ll be putting what I preach into practice this weekend at the Dublin Marathon and will report back on how I get on.

If you enjoyed this short article then there is a lot more on this topic over at Running with Data. Please leave a comment, get in touch (@barrysmyth), or otherwise share with friends.