Monday, May 2, 2011

Towers/Marathon Correlation

In response to the interest shown in the Down Hill Mile/Towers Ascent scatter plot and linear correlation, Nick and I have been working on a similar plot of Towers PRs and marathon times. With the new data from the recent marathons in Rhode Island, Massachusetts and Colorado, we have 17 data points and a fairy good correlation.
click on image to see full size
Each point on the chart represents one runner from the list below with their marathon time on the x-axis and their Towers PR on the y-axis. The line represents the average correlation between these two sets of data. The r squared value for this line is 0.8967 which tells the degree of correlation. An r squared value of 1.0 would mean that all the data is perfectly correlated (it all falls on the line).

By this data, runners whose point falls above the line (Nick C., Ryan, Cat, Tina, me) have marathon times that are disproportionately faster than their Towers PR. Runner's whose point falls below the line (Sam, Jenn, Slusher, Kyle, Felix) have had faster runs up Towers than in a marathon by comparison to the mean.

Of course this comparison has many flaws (insufficient data points, marathons were run on different courses, data does not have a sufficient spread over the times, etc.) but there is enough information to make some statements or predictions about a runners' potential in one event when compared to the other. For instance, to run a 3 hour marathon, one should be able to run 33:40 on Towers. Likewise, someone who can run a 30 minute ascent on Towers should be able to run a 2:46 marathon.

If you have a recent marathon time and a Towers time and I have not included you in the plot, please let me know, I can always use more data. A 5K/Towers plot and a 1/2 marathon/Towers plot would be interesting as well, so if you have that data please send it my way. And next time you see Nick on Towers and he runs anything slower than 27:30, be sure and ask him, "What happened?"


  1. Nice work, Alex. I think I'll start talking to real mountain runners only and ignore posers like you and Nick. Oh, and I'll be at the track tomorrow AM.

  2. Evidently I have my work cut out for me if I ever plan on running a road marathon. That graph suggests my marathon time should be about 10-15 minutes faster than I have even thought about running (and 45 min faster than my 2 year old PR)

  3. I can add a point, although they were set quite a few years apart - towers: 29:26 (2016), marathon: 2:36:26 (Boston 2009). Haha, that is insanely close to Nick's!

    1. My bad - didn't realize this was posted so long ago, ha. Someone just shared it on fb.