Monday, 1 February 2016

Marathon prediction and junk mile calculator

Race predictor Version 0.2
Age: Male Female
Previous race distance:
Time achieved (h:m:s):
Pace (mm:ss):
Age-grade:

Pace: per km per mile
Distances: Standard Custom

Average weekly distance:
Average weekly pace:

19 comments:

  1. In your article in Running Monthly, which is how I found your blog, you say that this relationship between training and marathon time is causative as well as correlative. Unless I missed something, the examples you gave are of people choosing to run further but at a slower pace and managing to improve race times.

    Have you any experience of the reverse approach, keeping mileage the same but purposively raising average pace? The most efficient way to do that would seem to be to increase the pace of "easy" runs, on the assumption that they constitute the majority of mileage. But that would fly in the face of received wisdom that it is time on feet, not miles per hour, that matters on easy runs

    ReplyDelete
    Replies
    1. I believe you are missing the point of what Christof has been trying to graph with this extensive set of blog posts- namely that a scientific study with a good size data set and accurate GPS measurements of training data says something more complex than "received wisdom". The Tanda study comes up with an exponential equation that implies that it's not "time on feet" ALONE that matters- it's also how fast you are running while on those feet.

      Take a hypothetical runner Bob, who might run his training at 9:00 pace average and 30 miles a week. The equation predicts that if Bob adds in a 10 mile long run per week (or 2 shorter runs), he'll take 6 minutes off his marathon time. Respectable. For Bob to do that by "speeding up", ALL his 30 miles have to be run at 8:34 pace now.

      The reason it may seem to you (and the wise crowd) that time on your feet is the only thing that matters is probably because most recreational runners can't just "decide" to run their 20-40 mile weeks at a 30 seconds/mile faster pace and keep it up for an 8 week training cycle. Remember that some of Bob's miles are likely already tempo runs- he can't make them faster or he will not finish his workout. So that turns his easy runs into more like "1:30" faster, which makes them into tempo runs too! That is MUCH harder than it sounds. Every single workout will hurt. Given the two available options (add in a long run, or do everything so it hurts) that will give Bob the SAME improvement, which do you think will work better for Bob?

      Delete
    2. Sorry, I did not mean to imply that I accept the received wisdom.
      I agree with your analysis of the problems Bob would face, but I can think of a couple of other scenarios that may apply.

      Take Alice, whose ageing joints can't cope with an increase in mileage. Suppose she undertook strength and conditioning training and some cross training in addition to her existing running: perhaps that would enable her to maintain a higher pace without increasing mileage and without breaking her?

      Or her younger friend Peter, who has been doing pure LSD for a few years and decides to put in some speed for a change. Instead of the traditional reps, intervals and tempo, he could decide to try upping his cruising speed.

      I appreciate the risks of injury and burnout, but my question to Christof still stands: is there any experience of people using this equation from the point of view of increasing speed rather than distance? Or does it express a relationship that is causative only in the distance dimension?

      Delete
    3. My gut feeling is that Peter, who just does long slow runs, may well benefit from upping his cruising speed. I don't have direct evidence for that, but we know from experience that within limits it must be true. If Peter's long slow runs are near walking pace, then speeding up will add a training stimulus. We know that as his speed increases further - towards interval paces - the same distance brings greater benefits. There is lots of literature supporting that. High speed running over shorter distances drives similar muscle adaptations to long slow running. But, I would be worried about Peter...is he going to survive it? I guess he is young!

      Delete
  2. You are quite right in pointing out that there is - possibly - another space that runners might try to train in to achieve faster marathon times, but the space is rather small. By running faster (over the same distance) one might expect a faster marathon performance, but I have no data supporting that at very short distances. If you look at the Tanda dataset, his 3 hour marathon runners are already averaging 90% age-graded performance every day. Given they must be doing some easy running, that means a lot of high speed interval work. My view is that there are not many people who could survive running even faster on the same mileage. I know of a few runners who train fast at relatively low mileage, but they are putting in average weeks similar to the fast Tanda runners. Certainly HITT and other such all-out efforts (including races in the training cycle) do drive fitness gains, but the maths suggests the gains will be limited. It is simply very difficult to run fast enough to make-up for the lack of distance (at least at low distances). I have looked at what I would need to do to maintain a decent marathon performance with minimal distance - and it means racing 10km at full speed every day. I think that would break me within a few days.

    ReplyDelete
  3. Regarding average weekly pace/distance - over what period? And including or excluding the taper and race week itself? I definitely fall into the low mileage high pace, rarely running slower than 8 min miles and covering a fair amount of ground at close to or faster than mp which got me a sub-3 off 35-40mpw (depending on how you measure it). I'd still struggle to fit my performance into what's predicted here though. Admittedly I don't have duration recorded for all my runs, but looking at my running log I find it hard to believe I was doing an average weekly pace off sub-6.50. I could believe sub-7.30...

    ReplyDelete
    Replies
    1. It is the eight week average finishing one week before the marathon. So, 40MPW at 3hour marathon pace translates to a 3 hour marathon prediction according to Tanda's equation. If you were averaging 40MPW at 7min30 per mile that translates into a 3:10 prediction. I have not come across any runners yet who have pulled off a sub 3 from that kind of training. Either you are doing more running or exercise than that, or the training you do is somehow more effective, or you are blessed with more miracle stuff than most. I have heard from one long term performance runner who had been heading towards 2:30 marathon but got injured and now years later pulled off a 3:07 when Tanda predicted a 3:15. There is a scatter, but usually the formula gets pretty close - closer than most people expect from such a coarse equation.

      Delete
    2. Christof, I fear I am going to weary you with this question (and may have asked it before) but if Tanda is based on an eight week average finishing one week before a M, does that mean that there is a week;s overlap between the final week of the eight week Tanda block and the first week of the taper? THanks Julian

      Delete
  4. Yes, you need to factor in a slightly lighter final week in the eight week period (2 weeks out from the marathon) - probably only 15% lower mileage than in the previous week. The last week (leading-up to the marathon) - outside of the formula - still includes about 50% of the distance with the race making it up to a similar distance to the previous week.

    ReplyDelete
  5. Great blog, thank you. If you live in a hilly area and/or run off road a lot I presume you would adjust for this in the calculation?

    ReplyDelete
    Replies
    1. Yes, both hills and off-road running apply additional loads. One way of correcting for it might be to look at flat pace for the equivalent heart rate. So, if an off-road run of 1 hour at 5 mins per km (12 km) required an average heart rate of 130 bpm you would need to find a flat terrain run at around the same heart rate - it might be 4:45 per km. So, that would make your off-road run equivalent to 12.6 km at 4:45 per km (5/4.75*12 km). Not perfect - but, probably better than no correction at all....

      Delete
    2. Thank you for the reply. What are your thoughts on the accuracy of the GAP adjustment function on Strava? My thoughts are that it is too conservative based on my own experience using heart rate adjustment as described above.

      Delete
  6. Another fascinating read - I'm glad I stumbled across your blog! Unless I've misunderstood something, what you have outlined above doesn't really apply to how I (and I'm sure many others) train, though. I've been following a polarized training approach for the last year or so (and something not too dissimilar to that for a couple of years before as well), which means I do the majority of my running at something like 8:00 to 8:30 minutes per mile, but once or twice a week I'll do speedwork sessions where I'm running at something between 5:15 and 6:00 pace, depending on the nature of the reps. Even though the weekly average pace I arrive it via that (let's say 7:30) is exactly the same as if I went out and did my 40 miles per week all at 7:30 paced runs, I would suggest that the physiological effect would be very different. Does it make to talk about average weekly distance and pace for someone who's following such a polarized approach?

    ReplyDelete
    Replies
    1. The equation Tanda fitted to his dataset of marathon runners probably followed a very similar approach to yours - i.e. they had some speed sessions. Our findings are that the equation seems to work reasonably well independently of how you arrive at the totals. That is 40 miles per week at 7:30 will get you to about the same place. It is true that some polarization does improve performance, but not by as much as you might imagine. If you look at my training on Strava you will see that I do no speedwork at all and yet Tanda predicts my marathon performances very well. It is possible that with speedwork I might do better - but, in the past I have got injured with that approach. Slow running is safe and it drives-up fitness in a nicely predictable fashion.

      Delete
  7. Hi, Does this work for half marathons? I have a half marathon coming up in 12 weeks (from Monday). I have been doing some higher mileage (80-90 miles for a couple of weeks) but wondering if I should continue doing that or to start following a training plan...with the training plan I would definitely have to do less mileage to recover from the faster stuff. Would you train just with high mileage/slow pace for a half or what your approach change? Thanks for any insight.

    ReplyDelete
    Replies
    1. Yes, many of the same limits (cardiovascular) that apply to the marathon also apply to the half marathon - we have had enough runners start the high mileage training and then do a test half marathon in which they have been shocked by just how fit they have got. If you are happy running high mileage then you will (most likely - there are no absolute guarantees) do better sticking with it than most plans based around fast running. Of course, if you can do enough fast running without getting injured then it is possible to get fast without the high mileage - but in my experience only a small number of people can do that. Unfortunately they also tend to be the ones that are 'bomb-proof' and get to be really fast. Trying to emulate such runners usually ends badly. Stick with a gradual build in the Tanda prediction and try not to stagnate at a particular pace/distance. Each week try and run like a slightly faster runner - but, take care and listen to your body. A down week every so often is a good thing. If you are already at about 90 miles per week with 12 weeks to go then you are in an excellent position.

      Delete
  8. Thank you for your response - very helpful. I much prefer the high mileage - it is easier to deal with...I seem to become inconsistent with faster stuff (I have days after faster stuff sometimes that my heart rate is so elevated/I sleep for 12 hours and have to take a few days).

    I might experiment with doing that then as its much easier for my body to handle.

    Thank you again.

    ReplyDelete
  9. Hi, I've just started reading your blogs which I enjoy and find very interesting. I've had a look over my training for the recent London Marathon and it appears I fall well outside the time predicted by the Tanda equation. In the 8 week period I ran 90km/week at 13.6km/h (4:24/km). Tanda predicts a time of 2:55, however I ran 2:34. This was a 6 minute PB from Manchester the previous year(off 99km/week at 13.6km/h). There is nothing about my training that adds additional loads. There were some slow stop/start miles done running next to my son on his bike but removing time for these made minimal difference. In London I ran a 30s negative split, while in Manchester I struggled the last 7km and ran a 3 minute positive split. My training has typically involved lots of running around LT and marathon pace but very few fast interval sessions. I'm 39 years old and have been running since I was at school but have only really trained properly for marathons in the last 2 years. Just wondering if you had any thoughts or insight? Cheers.

    ReplyDelete
    Replies
    1. Dear Glen, I have had a quick look at your Strava log and there is nothing special that stands out in your training. I can see you did a few faster longish runners, but they are not especially unusual.I think you can consider yourself blessed with the right adaptations! Whether that is just your long term status as an active athlete or whether it is luck with your morphology, physiology or genetics I don't know. But, I do look forward to seeing how you do at London next year. I think the Tanda relationship still applies but with an offset. We have seen similar, but smaller offsets in elite runners

      Delete