The 20 Miler

I mentioned earlier how important last year's 20 miler
was (see previous post, below). Ok, I know the regularity with which I update this blog is a bit absurd since that post went up 7 hours ago, but I think this might be helpful for some of my teammates to know about before the big run.

To add to what I mentioned before, last year Coach Adam had said something to the effect that my 20 miler performance would be indicative of what I could do the first 20 miles of the actual marathon. Furthermore, despite the additional few weeks of training followed by 3 weeks of tapering, that I would not get faster between the 20 miler and the marathon. I took that to heart and decided to move up to the 9:30 group and give it my all that day, which resulted in a 2:54:28.

Today I recalled Adam's words and went back to check my marathon time to see what my split was for the first 20 miles of the race. Sure enough it was 2:57:06, pretty close to my 20 miler time, and certainly not faster (2:38 slower in fact). I then got to thinking whether what he said was universally true or not, so I looked on the 20 Miler Database in the 2005 Yahoo Group and found some other times. (Last year when people finished the 20 miler team members wrote their times on a big flip chart in magic marker and Coach Erica McNally painstakingly converted the times to paces for everyone). Not everybody did it, but 78 people did, which I figured would be enough data to see if there was some kind of relationship. So I then looked up all of those people's marathon times and added that to the spreadsheet.

Ok, math is not really my forte, nor is putting accents over French words, so let's all pretend that there is one over the 'e' in "forte" for the moment and move on from there. I emailed my friend Nathan, who is a good guy to know when you need help with something. When I told him I needed to run the 60K to qualify for the marathon, he emailed me back his completed registration form 5 minutes later (there is the picture of us with our trophies after finishing the 37.2 mile run). In additon to his loyalty and speediness he also has some serious math skills, so I figured he was the right guy to try and make some sense of all the numbers I had plopped down on this Excel schedule. Sure enough he emailed me back with an analysis in about 20 minutes. Ok don't be frightened when you see this, this is his job after all, it was no big deal.

Below is a chart. To the left of the left most vertical line (not the axis, the one labeled zero) you see only 5 data points. Those 5 are the only people who ran faster during the first 20 miles of the marathon than they did during the 20 mile training run. The other 73 dots represent everyone else who recorded their time after the 20 miler. They all ran the first 20 miles of the marathon slower than they did the 20 miler.

According to Nathan, some things you can take away from the chart and the underlying numbers:

"1) Expect to run a 13 minute 'slower' 20-mile leg when you actually run the marathon.

2) The 'slower' the pace group, the worse the 'lag' (the formulas are on the charts)."

In addition, some further delving into the numbers may shed some light onto the 5 backwards performances. Marla "Flash" Weinstein I believe was injured for the 20 miler with ITB issues, a problem she had substantially worked out with Dr. John Schneider by marathon day. So that explains her faster 20 mile split vs. 20 miler time. Our resident photographer Omar was another one of the 5 who sped up on marathon day. I checked his photos from the 20 miler last year and he took 81 pics while running, many of which he appeared to stop for several seconds to take. Some of them were even elaborately set up, i.e. people posing together on the George Washington Bridge, shots of Omar (which would have involved him stopping, giving the camera to someone else, etc.) Omar snapped 74 pics during the first 20 miles of the marathon, all of them apparently taken on the fly. I conservatively estimated that the difference in these approaches probably cost him 10 more minutes during the training run, which would turn his 2 minute gain into an 8 minute loss. Perhaps there are other explanations for the other 3 increases in speed as well.

The reason I wanted to explain away the speed increases is because Nathan says the correlation of the data goes up dramatically if the outliers can be ignored, which increases the predictive power of the model. I also wanted another excuse to look at Omar's marathon photos again. :o)

The plan for last year's 20 mile training run was 15 minutes commute and 3 minutes warm up, a 5 to 1 ratio. The first 20 of the marathon called for 16 miles of commute and 4 of warm up, a 4 to 1 ratio, so the plan was to go faster on marathon day.

The weather on the day of the 20 miler was similar to that on marathon day, perhaps a few degrees cooler, although factors that may counteract that are the presence of all the fans, and water & Gatorade stations every mile on the marathon course.

What the Hell does all this mean?

It means you want to do well on Sunday, because your performance at the 20 miler, plus about 13 minutes (less if you are in the faster pace groups, more if you are in the slower ones), is what you can expect to do for the first 20 miles of the marathon.

According to the plan we will commute for 17 miles and race the last 3. Adam advises taking your commute pace down a notch, 30 - 60 seconds, in order to have the energy to race at the end.

So everybody get a good night's sleep on Saturday and fuel up with food and Gatorade endurance formula (lemon lime) all day.