McMillan's Running Calculator (Read 2818 times)

What has been your experience with the McMillan's Running Calculator? Does it provide an accurate prediction of your race times? I ran a 5 mile race today in 48:13. Using my 5K race time from two weeks ago, the McMillan's Running Calculator predicted a 5 mile finish time of 48:17 - an error of only 4 seconds or 0.14 %. Not bad!

Whoops, the race was actually an 8K (4.97 miles), so the calculator's prediction is 48:01, an error of -12 seconds or -0.25% (still pretty good!).

I think it gives you a good idea of what to expect. I always do better than the Macmillan Race preditor. The shorter the race, the less I beat it by...for instance, I did a 10k last month and it predicted me to finish in 48+ minutes and I came in a minute under. I came in 6minutes under it's predicted time for my marathon. It gives me a good range to think about.

These calculators are always too optimistic for me on the longer distances. Which tells me where there is room for improvement. I find the RunningAhead-calculator that Mikey pointed too the best and the easiest to use. Goes for the other calculators on this site too. bas

Thanks for pointing the RA calculator out. I didn't realize it was there. I like that it allows you to tweak the performance coefficient.

--------------WARNING!!! EXTREME GEEKERY TO FOLLOW!!--------------------- Yeah, I also like how you can tweak the "performance degradation coefficient" (as Eric calls it). With a little help from Microsoft Excel, you can actually calculate your own personal coefficient. I'm not going to get really detailed -- this isn't a tech forum -- so this assumes you have a working knowledge of graphing in Excel. 1) Put all of your PRs from 1500m on up into a spreadsheet. You'll want distance in the first column and time in the second column. (Races below 1500m are mostly anaerobic, and aren't good predictors of long distance performance). 2) For distance, use a consistent unit for all races (m, km, mi). I picked km since my old track races (1 mi, 2mi) were actually metric (1600m, 3200m) and since most road races are metric. 3) You can do time in whatever set of units you want. I chose decimal minutes (i.e. 1h5m30s becomes 65.5 min). This works easiest for what comes later. 4) Make a scatter (x-y) plot, but with the points not connected. You should see distance on the horizontal axis, and time on the vertical axis. The points will look like a straight line, but if you look closely, you'll see a slight upward curvature (which makes sense -- you get slower as distance increases, so time goes up). 5) Make both the X and Y axes logarithmic. You'll probably want to make the minimum x-axis value be 1000 (if you're doing meters). Your points will now be in an almost perfectly straight line. [BTW, the decimal minute time and km distance setup seems to produce the most readable graph]. 6) Right-click on one of the points and select "Add Trendline." Choose "Power", and under the options tab, select "Display Equation on Chart" and "Display R-Squared value on chart". 7) Voila! You'll have an equation of the form y=ax^b, and an R^2 value (hopefully) very close to 1.0. The exponent of the equation is your own personal performance coefficient. Like Eric said on the calculation page, this value will typically range from 1.06~1.10 (mine was 1.1012). You don't have to do log-log axes, but I find it make the graph more readable since there's such a wide spread in both time and distance. This exercise is serving the same purpose as entering two race times into Eric's calculator, but is presumably more accurate because you are including more data points. If your R^2 is much less than 0.99, you have a race or two that is anomalous (i.e. a really fast PR for a given distance, or a really slow one (maybe you've only raced 15k once in your life, and it wasn't what you'd call a "best effort")), and you might want to delete those points and see what you get. All running calculators are based roughly on this mathematical behavior and will give similar results. I think McMillan's calculator includes a tweak to make the sub-1500m distances more accurate, because I found them to be surprisingly close. Cheers, Jeff

Did someone say geekery? I put together a somewhat visual representation of my own performance degradation. The slope represents a degradation coefficient, although I'm not sure mine is scaled as Eric lays it out. The graphical depiction is interesting though. I also laid in world record paces to see how endurance degrades for the best in the world. The WR slope is more difficult to interpret the paces run are from different individuals. Still, an interesting comparison I think.

Thanks...and "had", as you said, is the operative word. Sadly, the wheels have fallen off this wagon.

Thanks for the compliments on the geekery and the speedery. Speedery is quite relative though, as we all know. The sub-15 sounds fast, and I'd be thrilled if I ever ran that. But, back when I was serious with things, I trained with guys that ran 3:57 (mile), 13:35 (5k) and 28:26 (10k), and one guy has subsequently gone 2:14 in the marathon. Simply put, I was out of my league when it came to anything over 3 laps on the track. Back to the geekery...I tossed out the 5k as an outlier when I dropped in the trendline, so I agree that the 5k could optimally be high 14's. This reinforces the importance of the race quality concept on the performance coefficient. I never did take anything over 2 miles seriously, and it clearly shows here, that I've never run 5k on up to my potential.