Thursday, May 4, 2017

Is Pike's Peek legit?

The Pike's Peek 10K in Rockville Maryland has been historically mocked on, that bastion of intellectual exchange (full disclosure - I participate on that forum).  Why?  Because of the net downhill.

In the past, I agreed.  Several years ago, I ran this race and bested by a full minute my then 10K PR, which I had set a month earlier in good weather on a fast course (39:16 versus 40:18).  And that was despite blowing up badly in the final stretch and even tripping right before the finish.  A year later, I returned to run Pike's Peek as my rustbuster, right after a break.  And I ran 39:15.  A time utterly inconsistent with any other time I had ever run at any distance.

Yup - the course was definitely assisted.  I noted the time for kicks, with an asterisk, but didn't claim it as a PR.

But then, a few years ago, the course was rerouted slightly.  While most of the course stayed the same, the start and finish were each moved back a quarter mile, adding a solid hill to the start and removing the sharp drop at the end.

This changed the total drop of the course, but it still had a drop.  And so, I reasoned, it still wasn't a course one could count as a PR.  (of course, all personal records are personal, so technically anyone can count anything they want....)

But now, I'm engaging in the pixel equivalent of eating my words.

In my opinion, the new Pike's Peek 10K course - the course that's been used since 2015 - is legit. This conclusion is based on both my own experience running the race this year, and some number crunching I've done below.

That analysis is below.  But first, I want to acknowledge the obvious.  This entire blog post is ridiculous.  Utterly ridiculous.

That being said, I'm probably not the only person interested in this topic, as over-analytical as it is.   So keep reading if you want - it's a guilty pleasure type thing.  Nobody has to know.  Promise.


In evaluating the legitimacy of the current Pikes' Peek course for PR purposes, I considered two points: how the net elevation drop compares to other races, and how runners performed at Pike's Peek this year, as compared to what their recent performances would predict for 10K.

Elevation drop

USATF maintains a database of certified courses on its website, searchable by location, distance, name, and other variables.  Each course certification includes a statement as to the net drop of the course, if any, calculated in meters per kilometer.
All information taken from
the USATF website, except for
the Main Street Mile,
which I calculated myself

How do various races measure up?  I've posted a selection to the right.

To be record eligible for USATF purposes, a course must have a net drop of less than 1.0 m/km and a separation between start and finish lines of 50% or less of the total distance of the course.'re not setting a US record on point-to-point courses like Grandma's Marathon or Boston, regardless of net drop.

However, it is possible to qualify for the Olympic Marathon Trials on a course that is point-to-point, as long at the net drop is no more than 3.25 meters per kilometer.    My hunch is that the 3.25 is not random, but was chosen specifically to include the Boston Marathon (3.23 m/km drop) as a qualifier.

Looking at my chart to the right, it's interesting to see how the net drops of the old and new Pike's Peek courses compare to Boston, as well as some other popular courses.  The old Pike's Peek is well above the 3.25 line, while the new course is well below.    Heck, the new Pike's Peek has less than half the net drop of the previous version.

It's also worth noting that the new Pike's Peek course has the same net drop as the Big Sur International Marathon.  Big Sur is a notoriously slow and difficult course - evidence that a net drop doesn't necessarily imply an assisted course.

So...based on net drop alone, there's some basis for considering a Pike's Peek PR (try saying that 5 times fast) legit.  Put another way, there's a solid argument against ruling it out from PR contention based on net drop.  But that's only part of the analysis.

Performance Comparison

I've found the McMillan Guide to be pretty good at predicting equivalent performances for me.  So when I ran 1:26:34 at Shamrock (a legitimate, USATF record eligible course), I entered the time into McMillan, and got a predicted 10K time of 38:50.  I was pretty close, but ever so slightly slower, with my Pike's Peek time (38:56).

So...I got curious, and decided to run a similar comparison for a few other runners.   I selected a group of runners who had run both Cherry Blossom 10 Miler 2017 and Pike's Peek, so I could compare performances between the two races.  Why did I pick Cherry Blossom as my baseline?  It's a USATF record eligible course with no net drop and no separation - very fast, but undeniably fair.  This year, Cherry Blossom also had nearly identical weather to Pike's Peek, and the two races were also separated by less than a month, reducing the possibility of profound fitness gains between the two.

 I limited my sample to people I knew, whom I understood to have had tapered and run both races all out, with no confounding factors (illness, running Boston in the gap between Cherry Blossom and Pike's Peek, etc).

With those criteria, I ended up with a sample of 8 runners.  Certainly not a large sample that would yield statistically sound conclusions, but still interesting.  Below are my results.

A comparison of 10K times, as predicted by performance at Cherry Blossom 2017,
versus performance at Pike's Peek 2017.
I also noted myself, based on my predicted time from Shamrock,
though I didn't include myself in the analysis.
What is striking is that a slight majority - 5 of 8 - ran slower at Pike's Peek than predicted by their Cherry Blossom performance.  Some ran significantly slower.  Of course, others ran significantly faster than predicted.

On average, these 8 runners ran 4.25 seconds SLOWER at Pike's Peek than one would have predicted, based on their Cherry Blossom times.

Of course, this is (again) a small sample size.  And I'll note for the statistics geeks that the standard deviation here is 26 seconds and change, making that "slower-by-4.25 seconds" claim essentially meaningless, from a statistical POV.  I haven't proven that Pike's Peek is equal to or slower than the legit Cherry Blossom course (and thus legit-by-insinuation).

But, it is interesting to look at.  And it's frankly stronger evidence than my previous dismissal of the old course based on the single data point of my own experience.


So based on the two points above (plus the fact that I broke 39 and really want to call it a PR), I'm inclined to claim the current Pike's Peek course as legit.    No eyerolling here, no asterisk.

Again, all PRs are by definition personal.  But, personally, I'm a numbers geek.  And what I've written above is what has convinced me, personally.


  1. One of these days I'll write a script to do a large-scale comparison between Cherry Blossom and Pike's Peek...

  2. are nothing if not thorough! I admit, though, I ran Boston (once), and even though it was dreadful weather I felt like I was kind of hurtling downhill for the majority of the race. It really felt downhill to me. Now, I do always run on flats, so it probably stood out more to me, but I still think the downhill is significant.

  3. Excellent article...a little beyond the pail...but I LIKE it! In fact, I will reference this article for anyone who questions my 10 mi PR set at GW Parkway (on your list above for net down hill...and also point to point - although we had a headwind the whole time). Is it actually more possible that courses are fast because we "think" they are fast and just run harder because we expect to be faster? Regardless - good work!