The Western States 100 Post

Heartbreak, Triumph, ... and Statistics!

I’m aware that lately this blog has become more of a trail running blog. I still have a love of the mountains, and I’ll be back on them again soon enough, both with the family, and on my own adventures. But hey, it’s June, and if you’re into trail running there’s nothing bigger right now than the Western States 100-Mile Endurance Run! Mountains, snow, heat, miles, and yes, puking… it’s all part of the magical WSER weekend. One of these years I might get a chance to run it. Until then, I’m merely a spectator in awe of the trail running greatness on display. With the race now behind us I wanted to offer a little post-race analysis, and a lot of stats. I’ve gathered 36 years worth of race data. In true stats nerd fashion, I made my predictions, in part based on the data, and in part based on the hype and what I hoped would happen. Ultimately I failed badly, so you get a chance to laugh at me! But maybe of more interest, have a look at some of the statistical facts and the insights they offer! Now, let us away!

A Super Brief Intro on Western States

You haven’t heard of Western States 100?? Wow, awesome, you’re about to discover something really cool… The Western States Endurance Run (“WSER”) is a 100 mile trail race in Northern California, that starts in Squaw Valley and finishes Auburn. It began in 1974 when a guy known as Gordy showed up for the Tevis Cup horse race… but forgot his horse back at home. He decided to run the 100 miles, just to show everyone how boss he was. It went viral on Twitter, all the equestrians realized how uncool horses were, and the rest is history. The 100 mile foot race was born! Something like that anyway. Every June, runners test themselves against the course to see if they can cover the 100 miles in under 30 hours, with special accolades for those who can do it in 24 hours or less. Elite runners have pushed finish times to under 15 hours. Crazy stuff. So trail.

Map provided by WSER foundation. Distance shown is approximate.
WSER Elevation Profile
WSER Approx Elevation Profile

2017 Race Results

By now, anybody who was interested knows what happened, but here’s a quick recap. Jim Walmsley was off to an early lead, he set out to crush the course record, and possibly make history with the first sub-14 hour finish time. It didn’t happen. His stomach turned on him and he dropped at the American River crossing. It was Ryan Sandes for the win with a solid effort, and Alex Nichols for second place after moving up consistently all day. On the women’s side of things, it was an exciting and pretty close battle for a good part of the day. Cat Bradley for the win, and Magdalena Boulet for second place (yay Magda!). iRunFar has a very satisfying recap, as expected.

Predictions that Didn’t Pan Out

I thought and hoped that Jim Walmsley would do the impossible, a new CR and possibly sub-14 on a very tough day. The course had a lot of snow on it, and the temperatures were roasting. Jim’s attitude indicated he would go all out, which made me really nervous for him given the conditions. I thought, maybe the conditions will knock him down to “just” an elite runner and he’ll run a good time for the day, maybe even a CR. Looking at the data from the last 5 years, the temperatures had been warm several of those years, the winning times had been fast, and 2012 was the last course record. By crunching some numbers I had the best outcome -for Jim- in the range of 14:42 to 15:17. This seemed totally improbable for anyone else, I wanted to believe though! But, what did the data actually point to?

Hot Hot Heat

There are innumerable factors that determine the outcome of a trail race of any distance, much less a long distance race such as the Western States 100. With that in mind, I looked only at a few obvious factors, the first being the weather. Given the time and resources I had to put this together, I had to settle for historical weather data for Auburn, which is of course the finish of the race. My logic was, a hot day there equated to a hot day on other parts of the course (“The Canyons” for instance will be an oven if Auburn is hot). The average high temp for Auburn in June is 91ºF. I decided to split all of the results into two groups using this information. The groups are (a) race days over 90º and (b) race days 90º and under.

This way we can look at the conditional probability that the winning time is in a certain bracket. For example, if the temperature is over 90º, what’s the likelihood that the winning time will be under 16 hours. This probability is worked out using the historic data. We are working with 36 years worth of data, so it’s a small but decent sample size to work with. The charts below show the conditional probability for various finish time brackets based on high temperature.

Win Brackets - Hotter Years

Win Brackets - Cooler Years

Yo Dawg, I heard you like Snow!

Snow is the other trouble maker! Last year we had a pretty limp snow pack. This year it was back with a vengeance. Snow pack is above normal, with a lot of water content. The runoff in the Sierra has been nuts for a few weeks now. WSER has had its years of big snow pack. 1980, 1983, 1993, and 1995 are some of the more notorious ones. In each of those years, finish times ranged from 16 to 18 hours. 2010, and 2011 were also fairly high snow years, but the winning times were faster in the mid-15 range. Those results are a bit of a mixed bag, and too few data to crunch numbers, but intuitively we should equate “snow” to “slow”. Wet shoes, wet feet, sun cups, mud, slippery trail, all that nonsense. With this factored in, on top of the heat, I was not finding a way to envision a sub-14 finish, even a new CR would take a herculean effort!

Type-I Error, Type-II Fun?

I made a Type-I error and rejected the truth as my stats professor would say! Dammit… I knew better too. I should have run a separate set of numbers for the typical (elite) runner. I put all my money on Jim, despite the data saying otherwise. Let’s go back for a second and relive the drama… Jim was about 26 minutes ahead of the CR (!) somewhere prior to the Rucky Chucky aid station at the American River crossing (mile 78). At 12:12 elapsed, Rucky Chucky was reporting that no runners had been through. The split time for the CR was 11:32 at Rucky Chucky. Wait, wut? Yes, 12:12 minus 11:32 = 40 minutes, behind CR pace – and no sign of Jim. I had a sinking feeling something had gone very wrong. Of course we later learned, it was stomach issues that began around mile 62. Such a bummer. The stomach can be a frustrating piece of machinery.

16:38 - The Magic Number

It turned out that 13:59:59 was not the magic number of the day, it was actually 16:38:16. In 36 years of data, 21 race days have been over 90 degrees. The mean winning time for those days is 16:38:16. Constructing a 95% confidence interval around this, we end up with a window of 16:08 to 17:08 for the predicted finish time. It’s a pretty wide range due to the smallish sample of data, but it turned out to be accurate. Ryan Sandes broke the tape in 16:19:39, about 19 minutes faster than the average for hot race days, and really damn good for a day with all the snow and slop early on. He was with the majority (61%) of winners on hot years, in the sub-17 bracket. In the cooler years, every race except one has been won in sub-17 or better, which should tell us something about how tough it is out there when the heat is on.

Outro…

That’s a wrap for now. There’s a lot more I’d love to do with this data, and there’s more data out there that I’d love to have access to, so if anyone with a connect from WSER Foundation is reading, please drop me a message! Beyond the data, I have a lot of respect for Jim Walmsley and what he set out to accomplish. What an amazing talent he has, no doubt it will come together for him at some point. Much respect to all the runners out there. I find it pretty amazing that people are able to run 100 miles. What a mental and physical accomplishment that is. I’m still searching for the motivation to give it a shot. I’ve qualified a couple of times for WSER and will put my name into the lottery this year to start the waiting process. I guess when it’s my turn, that will be the motivation! It was exciting to follow all of the action this year from a distance via iRunFar’s Twitter Feed. A handful of local runners and friends were running in both the men’s and women’s side, and it was awesome to see them do well, or give their best effort trying. The 100 mile distance is one that takes so much support; It really brings people together, and that may be one of the best reasons to do it.

Other WSER Stats 1980 – 2016

Here are a few more fun stats of the WSER, enjoy! (and if you spot any mistakes, please yell at me) Hike It. Like It.

Weather

  • Race Day Average High (Auburn): 91º F
  • Race Day Average Low (Auburn): 60º F
  • Most “Mild” Weather Years: 1997, 2005
Results of High Temps
Crazy line diagram!!
Temp vs Finish Rate
No correlation between high temps and finish rates, ok fine.
Temp vs Sub-24
No correlation between high temps and sub-24 finish rates… interesting.

Winning Times

  • Fastest Winning Time (CR): 14:46:44 (Tim Olson. 2012)
  • Slowest Winning Time: 18:35:42 (Mike Caitlin, 1980)
  • Average Winning Time: 16:26:39
  • Average Winning Time Days Over 90º: 16:38:16
  • Average Winning Time Days 90º and Under: 16:10:24

Finish Times

  • Average Finish Time: 25:37:39
  • Average Finish Time Days Over 90º: 25:45:51
  • Average Finish Time Days 90º and Under: 25:26:11
  • Standard Deviation of Finish Times (avg all years): 3:18:04

Finish Rate

  • Average Finish Rate: 66%
  • Highest Finish Rate: 83% (2011, 2012)
  • Lowest Finish Rate: 49% (1980)

Sub 24 Hour Finishes

  • Average Sub-24 Rate: 39%
  • Highest Sub-24 Rate: 63% (1984)
  • Lowest Sub-24 Rate: 18% (1995)

Gender

  • Average % of Finishers Male: 84%
  • Average % of Finishers Female: 16%
  • Highest % Male Finishers: 92% (1987)
  • Highest % Female Finishers: 23% (2005)
  • Average M:F Ratio: 5.6:1
  • Highest M:F Ratio: 11.1:1 (1987)
  • Lowest M:F Ratio: 3.4:1 (2005)
An un-related stat - M:F Ratio shown over the Years
The ratio of male to female finishers is declining over time. Way to go ladies!
MF vs Finish Rate
No correlation between M:F ratio and finish rates.
MF vs Sub-24
Very weak correlation between M:F ratio and sub-24 finish rate.
Finish Time Stats Compared
Finish Time Stats Compared. Exact Middle of Field takes into account the size of the starting field.

(No Hands Bridge photo credit: Nick Ares)

Jacob D Written by:

Jacob is the head honcho, wearer of many hats, and modern day berserker here at Hike It. Like It. When he's not out hiking or running the trails you'll find him operating in full capacity as a Super Dad and chipping away at a degree in Kinesiology. This guy likes to stay busy. Follow on Strava

4 Comments

  1. Andrew Skurka
    June 27, 2017
    Reply

    Can relate to the shift in focus.

    Thanks for the analysis.

    • June 27, 2017
      Reply

      Hi Andrew. Thanks for stopping by.

  2. Jen
    June 27, 2017
    Reply

    I’d be curious to see gender specific finishing rate. I assume the highest Male finisher rate (92%) was based on both men and women? But breaking it out separate… as in if 20 females finished out of 30 females total that year? I wonder what the overall average would be?? And if male vs female would be higher. I’ve heard leadville’s female only finisher rate is shockingly high despite an overall very low finisher rate.

    • June 27, 2017
      Reply

      Hi Jen. Ok, a couple of things, as it’s a little confusing (maybe I will add some notes to the post above). I did not have access to data that showed what the composition of the starting field was… I would love to know how it looked in terms of number of men, and number of women who started, vs. how many finished.

      The 92% stat is NOT a reference to finish rate, it just points out what percentage of finishers were male. I hope that makes sense.

      I have a lot of interest in the gender related statistics, if I cannot get ahold of the data needed from WSER, I may look for another large 100-miler that publishes the full start and finish data to run up some analysis on for fun. Having a few different races to sample from would be ideal.

Leave a Reply

Your email address will not be published. Required fields are marked *