It is time again to consider where I have managed to drag my fitness to and consider what level of performance might be possible at the Frankfurt Marathon. With one week to go I think all of the 'training' is now done and I should have enough data to produce a reasonable guess.
My marathon performance has been gradually improving over the last 4 years, I started 'competing' in marathons in 2011 about two years after I started running (mostly 5Ks). Here are my PBs for each of the years:
2011 Prague Marathon 3:29:33 (HR 146)
2012 Prague Marathon 3:07:52 (HR 146)
2013 Berlin Marathon 3:05:39 (I did a track marathon faster, but it doesn't count...) (HR 155)
2014 Autumn Shakespeare Marathon 2:57:00 (HR 152)
2015 Virgin Money London Marathon 2:51:34 (HR 150)
Whilst there is a gradual improvement in times, as one might expect with continued training, there were a few strategy changes that I think have had an effect. In 2012 at Prague I tried using lightweight trainers for the first time (Asic Hyperspeeds) and I think they leveraged a 5-7 min advantage. In 2013 I spent most of my training doing stair reps which I think did improve my leg strength allowing me to pace my nephew Josh through his first marathon and get a PB despite my relative lack of cardiovascular fitness (average heart rate 155 - the highest I have yet achieved over a marathon). In 2014 I ran the Raceway to get my first sub-3 using a zero in-race fuelling strategy. By then I had realized that taking gels, fruit jellies and water was not necessary and it was easier to just run. I was also shifting my training towards training layered-up, wearing heavy shoes and a light rucksack. I then began to seriously engage with glycogen depleting on a daily basis (by not eating at lunch) on top of a high mileage diet of running. Those adaptations brought me down to 2:51:34 (on what was a day of perfect conditions). My weight has also gradually declined to an average of about 59 kg (the precise value depends upon the state of my glycogen loading - I can easily swing by 3 kg over the course of a week) and for those of you concerned about my health...my BMI is ~20.5.
Distance over the last 8 weeks 1,380 km (~173 km / 108 miles per week) at 4:55 per km
Two rest days - either side of a 20 km race effort.
Peak week: ~227 km (142 miles, 5th-11th October) at 4:51 per km
My Tanda (2011) marathon prediction (a published correlative formula) based upon the distance covered in the last 8 weeks and the time taken is ~2:46 but is heavily biased by an increase in my training over the last 4 weeks. Whether that is good or bad I don't know. Figure 1 shows how the Tanda prediction varies depending upon how far back you look in my training diary. All of those calculations are based on looking back at the training I have done up until today. So -3 weeks is the Tanda prediction using the past 3 weeks worth of data and it suggests a marathon performance of ~2:36. Obviously this is incorrect since it is unlikely one can expect that 3 weeks of a particular training intensity would bring the same results as the full 8 weeks of training that the formula is based upon.
 |
| Figure 1 Tanda predictions of marathon performance using different time periods. The published formula uses the full 8 weeks and predicts a time close to 2:45, however, it is biased by an intense training block over the last 5 weeks. |
So, the question is whether the recent tough short block (5 to 1 week ago) of lots of running but relatively slow, is going to pay-off. There are good reasons for thinking it could go either way. But, this represents a test of Tanda in two dimensions - the formula never claimed to be causative and I am operating some way beyond the dataset on which it was based (extrapolation). I have tested the extrapolative validity of the Tanda equation on a few runners, some of them considerably faster than me, and under the right conditions it seems to do a reasonable job. But, these were runners doing a standard marathon training plan. My plan was not a standard one. I have done no fast running over the past 8 weeks - I have only done one run faster than marathon pace and that was a parkrun today (19 min 5 km) and only two runs that came close to marathon pace (the 20 km effort, see above, and a 5 km pacing for a friend on the 10th September) - neither was easy. I have done no tempo runs and no speed sessions. I also have done no long runs. I don't 'train' for a marathon I just run to work and back and typically do a 10 km run instead of lunch. My longest single run was the 20 km effort (see above), otherwise my longest continuous run was 12 km (although I do a fragmented run on a Saturday to the parkrun and back that amounts to 25-30 km with no drinking or fuelling. My strategy has been to select pace and distance to optimize the amount of 'intensity' that I could put into my running over the 8 week block. Fast running causes me too many injuries and limits the amount I can do - running too slowly chews through too much time. So, I have run to and from work and added as much training stress as I could (rucksack, heavy trainers, coat and glycogen depletion) to maximize the benefit of my 'optimal-Tanda' plan although only distance and pace is used by the Tanda equation. This is where considering heart rate becomes important.
Figure 2 shows heart rate data from the last 8 weeks. It does not include my 3rd run of the day since it is usually pretty horrific since I am glycogen-depleted probably even beyond the level I would expect during a marathon. On the y-axis is my average heart rate on each run plotted against speed (on the x-axis) in units of meters per min, which may seem odd. First, I use speed rather than pace since heart rate is typically linear with speed over a reasonable range which allows some simple predictions to be made. I also use units of minutes since heart rate is measured in beats per min. The straight line is a simple linear regression fit to the heart rate versus speed data. This fit predicts a heart rate of 28 beats per minute for me when I am running at 0 meters per min and a rise in heart rate of ~0.47 beats for each meter per min increase in speed. I have plotted on the graph (in magenta) two symbols which happen to overlay one another. The first is a square, the x value of which is the Tanda predicted speed (using the 8 week dataset) with a y-value that is my 'expected' marathon heart rate (148). The second point is a triangle which is plotted at 2 hour 45 min marathon speed and again at my expected marathon heart rate.
 | |
|
Three faster runs are visible on Figure 2. The top two are my 5 km pacing on the 10th September and the 20 km effort on the 20th September. Neither of these runs fall on the line. There are several possible reasons for this the most obvious of which is that I have got fitter since then. The other fast point - which sits directly on the line is my parkrun today. This causes me some concern since the line is the average of 8 weeks and my performance today suggests that all I can do is the average of those data points - I was hoping that things would be better than that and I might be able to pull off a performance closer to my recent runs. To illustrate I have replotted the graph using just the last two weeks worth of heart rate data (see Figure 3).
 |
| Figure 3. Heart rate data for the last 2 weeks plotted against the average speed of the run. The triangle is plotted at 2 hour 45 min marathon speed and a heart rate of 148. The square is the 2 week Tanda prediction (2 hour 37 mins). |
Obviously the 2 week Tanda prediction is not realistic since the heart rate scaling would suggest I would need to hold an average of 152 beats per min. But, now the average of my last 2 weeks heart rate data suggests the 2 hour 45 should be achievable under a heart rate of 148. It also suggests that my parkrun today was sub-optimal (by more than is shown here since that data point is biasing the line upwards). I am concerned that the sub-optimal nature of my 5 km performance is that my heart rate does not scale in a linear fashion at high speed and that as a result such extrapolations are dangerous....without more data it is hard to know whether tracking close behind Fergie on a winding route wearing a coat at the end of a tough week and following a 10 km warm-up was the problem - or whether I am bio-mechanically limited in some way....
On the question of whether the full 8-week period for the Tanda prediction is optimal for a 'non steady-state' training regime (i.e. one where there is a rapid ramp-up of fitness) I do have a bit of relevant data. The 7-week Tanda prediction was reasonably accurate for my performance at VMLM and for what might have happened at Amsterdam had my legs, intestines and the weather been more 'helpful'. The current 7-week Tanda for Frankfurt is sitting nicely at 2:43:37.
So, I have got a stack of three marathons - Frankfurt, The Raceway and Pisa - before the end of the year, if necessary, to try and see whether 2:45 is possible. I will gain 'battle-testing' from each failure and learn more about what this data means. I know that 2:45 is going to be tough to achieve and probably rate my current chances of success at around 10% on my first outing. If the weather is perfect, I get a good draught from a group of steady runners who are tracking the line, my guts play ball, I don't lose the feeling in my legs and feet for 10 km and I get my carb-loading and my plasma-volume in the right place who knows perhaps it might just work. But, there is no limit to just how far things can go wrong in a marathon.......Now, I hope my Achilles settles down this week, my heels stop hurting, the pain in my hamstrings disappears and my groin strain doesn't flare-up. I am now off to find some even lighter socks, underpants and shorts and shave-off any hair that isn't performance enhancing! I am not obsessed, just motivated - Kevin will be trying to beat me and I can't let a youngster win without a fight.
No comments:
Post a Comment