Back-to-Backs, Average Load and Variance

The one thing seemingly agreed upon among ultramarathoners are back-to-back long runs; almost everyone does them, almost everyone says they're important. Admittedly, the rationale behind them has largely escaped me because there are so many different explanations that have been given. If I don't understand something, I tend to dismiss it and I have been far better at planning back-to-backs than actually doing them. I generally run the first too hard, making the second a necessarily shorter run; if I consciously hold back during the first one, I end up running so slowly that the sheer length of time on my feet makes the second one necessarily shorter. I need to know very specifically what I'm doing and why.

The most common rationale given is that one needs to learn to run when already tired. But if you run a 6 hour run, aren't you running on more-tired legs for longer than if you did back-to-back runs of three hours? It seemed to me that those doing two three hour runs back-to-back were doing them because of the schedule of the rest of their week: they run long Saturday and Sunday because those are the days available and it is possible to get up very early and run for three hours and still be able to do normal family activities at normal times. If you ask runners if they'd rather get in some six hour runs, they always say they would, but don't have the time. The eternal question is: how can you prepare for running 24 hours if you've only run 4 hours in training? This is where lead-up races of 50K, 50M and 100K get included - but wouldn't longer training runs still be better than back-to-backs? I was starting to think that the main advantage to back-to-backs was that runners were invariably doing them before sunrise and were thus better prepared for the night running that's almost always required in a trail 100.

It was a rather misguided alternative rationale that started me seeing the situation in a new light. This source said that the reason for back-to-backs was that one could shoehorn in a few extra long runs into a schedule that way, thus increasing the average load over the training season. I don't think that that is true in most cases (this is admittedly mostly a gut feeling based on experience). The "average load" did lead me to an idea:

A general model and framework for training

Most biological systems can be described by a series of bell curves, so one's trianing can be expected to fit such a curve if one had an appropriate measure of effort - most common measures, like mileage, do not adequately measure effort and attmpts to measure efforts (including my own) are only slightly better. That said, moving the center of the curve, the average, means improving fitness.

Races, however, are attempts to move as far as possible away from one's average. If all of one's training runs are about the same, it's difficult to do something far from that average. If you do a training run too far from your average, it becomes so stressful that one needs several easy days to recover and this can bring down the average. This is the problem with the six hour run - even at the easiest of paces, it is too far from the average for any typical runner, whereas two three hour runs may not be. Of course, if one ran six hours every day, a six hour run would be average, but this is too much work to be useful.

I think the ideal typical week for training for a trail 100 would look something like:

M 1 hour
T 1.5
W 1
Th 1.5
F 1
Sa 2
S 4-4.5

It's common practice to put the longer of the back-to-backs first, as it is the more important. What I'm finding top runners doing, however, is running a hard and fast run on Saturday, followed by a very long  very easy trail run on Sunday. The faster running makes the Saturday workout a hard one and the increased pace balances the slower pace on Sunday, so that the average pace run during the week does not drop. Because one has to run the longest run slowly - or it becomes too stressful - it tends to drag the average pace down just because it is such a large proportion (about a third) of the week's mileage. If one tries to run 4 hours on Saturday, running a fast 2 hours on Sunday will be impossible; one needs the easier runs before it. If one tries to place two hours of hard running in the 4 hour run, whether one does it on Saturday or Sunday, it becomes too stressful and affects the next week negatively.

Again, how can one expect to run, say, 24 hours, if one never runs more than 4 hours? From the week given above, the race is about 20 standard deviations from the mean - an anomaly bordering on impossible. I found that to make a 100 mile run not be too far from the average, one needs to run 100 miles per week: 12-13 on Sundays through Fridays, with 25-26 on Saturday, a 50 mile Saturday race in the 7th week and a 100 mile race in the 14th week. This would be a great mainenance program for a sub-2:30 marathoner, but too much for the less gifted. Additionally, the weekly variation is too small until the 50 miler - this could be adjusted by some runs being done faster than others - so a different method would be needed to reach the first 100.

To run 100 miles more often and make the weekly variability better, one would have to run even more, about 160 miles per week to have a monthly 100, and this without breaking into two-a-days. This too is best left to the most talented of runners.

This is why it's so hard to wrap my head around the numbers. No one ever said "the logical thing to do is to run 100 miles."

What hasn't been determined is: if one manages the appropriate average effort in training, with the proper amount of variation, does the back-to-back still matter? I think not.

I'm still going to do them, however.