I was going through past race results, trying to make a list of where my current PR’s stand.  I ended up putting it into an Excel spreadsheet and came up with two very interesting graphs.

They actually both contain the same data, just one shows pace and the other shows time (in minutes) for each race distance.

I was fascinated by the fact that, with two exceptions, my fastest races follow a very distinct path.  The two exceptions include 1) the Boilermaker 15K where I tore my calf muscle, and 2) the 2010 New Haven Road Race (20K).  Other than these two, everything else falls on the lines.

If I did not have my recent marathon PR time, I could have predicted it within a few minutes of the actual time I ran.

Even more interesting is that these race did not all occur this year, so that my race paces should have been consistent.  They happened over a couple years:

5K    Apr-09
10K    Nov-08
15K    Jul-09
20K    Nov-10
1/2 M    Apr-09
30K    Mar-06
M    Nov-10

The real kicker was that 30K (18.6 miles) in 2006.  My time was right on the curve for that one.

Looking at this graph, I have to wonder if this means that my maximum speed, for any given distance, is really no better than it was 4 years ago.

So what does this mean?  I think, I am seeing the maximum that I am capable of for a given race.  One would expect that over time, I would eventually hit my max.  I could run several races (say 5K’s), different times of the year, different conditions, and whatever.   After enough of them, I will have gone just about as fast as my body would ever be able to go.   If I were not at that point, I think the noise in that line would be quite large.  Later dates would show much faster times, and the trend line would not be there.

So what happens when I plot, say, all my 5K times I can find?

I think in general it might be showing a downward trend, but its a little hard to see.  It might be a stretch.  Looking at it more objectively, it looks like noise.

It makes me wonder if I really am improving, or if there just isn’t enough data (i.e. races) to give me a large enough sampling of race times to show my maximum speed.

It could be I am looking at this all wrong and I don’t understand really how statistics work.


  1. i have heard that runners improve for 10 years from when they start running. however, i doubt if you (or i) are maxed out. i definitely wasn’t training w/ direction for 10 years.

    training differently might yield different results. for example. more miles or, if you are focusing on shorter races, race-specific speed training, might yield faster times.

    it gives me a lot of hope and inspiration to see members of my running older than me (who have been running competitively since they were teens) still PRing in their mid-30s (those are the girls i look to b/c i see myself in a few years!) and the men in their mid-40s. i think there is still much room to grow as a runner.

  2. You can always screw around with statistics until it confuses the hell out of you!

    1) You’re definitely getting faster. Look at the 5K times. Calculate your average finishing time for each year (or each set of two years), and your most recent finishing times are faster. (Heck, your PR is last year. 3 of your 5 fastest times are in the last two years.)

    2) The curve on the second graph is calculated based on your times – after they’re already there! Your most recent times (new haven and the marathon) are faster than expected. take those two races out, recalculate the curve, and so how well they “predict” your finishing times in those two races. i think you’ll see that you smashed the “predicted” times.