click the picture to enlarge

day 382: this data point is not quite real. Yasso 800’s were on the trainings plan, however had to stop after 3 and a half, the left side of the hip hurt too much. However the first three runs were really promising, 2:32, 2:32, 2:33min. Applying an equal degradation in times as in previous Yasso sessions, it would result in a 2:35min on average for the whole series.

day 332: It’s been a while, after the spring’s disappointments. But now we’re back on the road and on the track. Yasso 800’s provided a first data point on the home stretch towards the Berlin marathon. 2:38,7min on average. That would indicate a 2:38h race time. But let’s not take this at face value, with athletes below the 3h marathon mark, one should add up to five minutes. Let’s add 2,5 min since it was a lot of endurance training lately. 2:41:20h for this data point.

day 259: the Helsinki City Run, a half marathon, even worse than the Rotterdam marathon.

day 225: the first real marathon as data point, quite disappointing to see such a hit on the progress. A brief analysis of the race here. There’s no other option than going forward the way it was planned with a focus on 10k and half marathons during the summer and then to take it one step up towards the Berlin Marathon.

day 203: the sixth 10k test race, same place as in the previous test races, was a bit of a desaster. With 35:56min it was a step backwards instead of a forwards. The km amounts and intensity of the marathon preparations seem to show its negative effects, race comments here. Well, I don’t believe that the marathon running capabilities have generally decreased through the training, but rather the training required this much energy that the current race result wasn’t that good. The progress curve of course gets a bad bumper from this.

day 147: the fourth 10k test race, again hakunila, as race 2 and 3, this is good for comparing results. this time a 35:34min, below the self set target of 35:45min, good stuff! reasons here. the marathon time predictions based on this 10k time range from 2:41 to 2:47, average of 2:44:30. the actual progress curve has an alarming tendency to level out faster than the target progress (blue curve)…

day 112: the third 10k test race, same place as the second test race, it’s part of a 5 race winter running series. the result was a time of 36:14min, a new personal best in this project but clearly below the target of 35:50 - 36:10min. the marathon race time predictions based on this latest 10k time range from 2:44 to 2:50 with an average of 2:47:30. reflections on the 3rd test race.

day 91: the second 10k test race this time in vantaa, north of helsinki, resulted in a time of 36:40min. exactly 5 sec faster per km than in the first test. the marathon race time predictions based on this latest 10k time range from 2:47 to 2:52, the calculated average 2:49:11. well, let’s just say 2:50, that’s a round number.

day 71: the kuivannon kymppi (10k) result of 37:30min was fed into several prediction counters and formulas. the results for the full marathon distance were surprisingly close to each other. they ranged from 2:55:51 at the higher to 2:51:40 at the lower end. the calculated average 2:53:30. intuition says that no error margin adjustments needed this time. 10k results are a better input to a series of short runs, such as yasso 800s. for more on marathon race time predictions see here.

day 52: ran yasso 800s today, at pretty good speed 2:47 on average. added for the data point a 10 min “error margin”. with all the track training lately, i seem to get used to it and these shorter interval runs indicate a much faster marathon time, that it actually would be.

day 49: two data points were added, the first from “race time prediction” based on a current 1000m personal best, i guessed even that one, so very high error probability, the second based on the yasso 800s that were done around day 40. i feel that with those two data points the initial improvement would be too steep and think that in reality we are closer to the target progress curve right now.

measuring progress while training is not quite trivial. yes, there are indications of improvement from one week to the next or over the span of several weeks, interval runs are faster or more repetitions at the same pace. also the heart rate is showing lower averages at similar efforts. but neither is the actual question answered whether this improvement is good enough to reach the sub-2:30 by the end of september 09, nor is there sufficient time to run several marathons before to see how we’re doing. Hence a target progress curve is needed for the whole period against which data points from various test/training sessions + calculations can be compared.

the best guess for the target process was to base it on an ln(x) function as the majority of natural/organic development processes show this behavior. the actual equation is

y =a * ln(x+b) + c

where y is the improvement in seconds over the initial at day 1 assumed marathon time (3:30h). the x-axis is the number of days going from 1 to 400. with the end goal - 2:30h after 400 days of training - and one more taken data point from current training, the yasso 800s after 40 days, there are three equations with which the three variables a, b and c can be calculated. with more data points being collected those variables, and thereby the curve of the function might still be fine tuned. more explanations on the target progress curve here.

Leave a Reply

You must be logged in to post a comment.