Fast Starts = Slow Finishes (Chicago Edition)

What the data from the Chicago Marathon tells us about the dangers of starting too fast.

TLDR;

• Starting too fast too early may wreck your marathon by as much as 60-minutes based on data from more than 400,000 Chicago Marathon finishers.
• Going out too fast also makes you much more likely to hit the dreaded wall later in the race; over 50% of Chicago Marathoners who hit the wall go out too fast in the first 5k.
• And yet lots of people continue to go out too fast; almost 40% of Chicago Marathoners have the first 5k as their fastest race segment. Women and older runners are slightly more likely to do this but ability and experience play a major role.
• Starting more conservatively can improve your finish time; about 10% of Chicago Marathoners run the first 5k as their slowest of race and they gain a whopping 50 minutes on those who’s first 5k is their fastest. And, according the data, none of these slow-starters hit the wall!
• This is part of a series of posts that I am writing as part of an ongoing analysis of marathon data. So if you are interested in this sort of thing then you will find a growing number of articles here.

Seriously. Don’t Go Out Too Fast in a Marathon

One of most common pieces of advice for marathoners — the number one race time if you will — is: don’t go out too fast. If you do you’ll suffer later, or so goes the conventional wisdom, and yet every year a significant percentage of partipants throw caution to the wind and fly off the starting line. Some simply get caught up in the excitement of the start and neglect to monitor their pacing before realising their mistake. Others plan to bank some time early on while they are still fresh. Yet others adopt a “go out hard and hang on for dear life” strategy. These are all a bad idea according to the lore of running, but what does the data say?

Our data is drawn from the last 12 years of the Chicago Marathon (2005–2016). This data provides a wealth of rich data including 5k split-times for each runner; that provides us with 8 x 5k segments (5k, 10k, 15k, 20k, 25k, 30k, 35k, 40k) plus a final 2.2k…