Thursday, 19 February 2015

Evaluating Fitness during the Training Cyle

Motivation for assessing training

When training for big ultra races you a often faced with many months of training building up to your A races, during this period one hardly races if at all, so you don't have any race data to assessing how your fitness is progressing.

Subjectively you might feel like you are getting fitter and stronger, but you might just as equally have a what feels like a couple of bad weeks and your confidence can be knocked.  When you stand on the start line for your big races it's good to know that your training has gone well and what finishing times might be realistic to aim for.

How your fitness is progressing will also influence what training might be appropriate to do next.  This post will outline a couple of ways I'm trying to by analysing my own training logs to provide answers to just how my training is progressing.

Evaluation runs

One way of assessing fitness is to run the same course at the same intensity over a period of weeks, months and years to see how your times/paces vary.  If your training cycle contains shorter races like regular 5km, 10km, half marathons etc. then potentially you could use these.

However, such races would typically be done all-out which limits how often you can do them in a training cycle.  Short races are done at very different intensity to ultra's, above or around lactate threshold, where as ultra's are done at competitively leisurely pace.  Ultra's also stress parts of your overall fitness in a different way to short races, from my own experience my 10km PB has only improved by around 5% over the last 5 years, but my ultra's have improved by nearer 20% during this period.

With the advent of heart rate monitors we have a the ability to both record and to monitor in real-time how hard we are running.  By running on a standard route and to a target heart rate it's possible to standardise the effort level between a series of runs. 

The evaluation runs provide three bits of fitness information:
  1. The faster our average speed we can run for a given heart rate the fitter are likely to be.   
  2. How our speed changes during an evaluation can provide another marker for fitness - the fitter we are the more stable the pace will be throught the evaluation.
  3. Finally one can also look at how quickly the heart rate recovers to a lower target HR.  The quicker your HR drops the fitter you are.
Ultra runner Thomas Bubendorfer (Diary of a Rubbish Marathon Runner) regularly uses this type of evaluation run during his training. The scheme he uses is:
  1. Target HR of 161 bpm (recommend to him my Mystery Coach), I would estimate this HR is around 15pbm below his average HR during 10km's.
  2. Warm up for several miles, then wind up the speed so the HR is around 161 when the evaluation section begins
  3. Four mile segments, each mile done on a flat half mile out/back section of road
  4. At the end of each mile a split time is recorded
  5. At the end of 4 mile evaluation you stop, and stand still timing how long it takes the HR to drop to 130 bpm.
  6. Jog home.
To see examples of Thomas' click here: Evaluation Runs

Thomas is a far more accomplished ultra-runner than me (he's got a place on Austria's 24hr team) so it's useful to follow his blog and training progress.  I have longed wondered about doing a similar evaluation run, but never got round to it - until this week!

As Thomas and I have a similar HR during 10km's I adopted the same target HR, this is really just co-incidence though, runners really should pick a target HR relative to the their Lactate Threshold/10km pace.   For this particular type of evaluation I believe it's useful to be a bit below Lactate Threshold so the four miles can be completed comfortably.  Chatting with Thomas these evaluations roughly equate to his Marathon race pace.

My first 4 miles @ 161 Evaluation Run

The route I chose yesterday was at the easterly end of Loch Venachar, going from the Car Park at the end of the Loch to the Gate House just before Sailing Club.  This is about as flat a bit of road as you can get in the Trossachs and is lined with trees and great views of the Loch and surrounding mountains.  It's a lovely place to run up and down ;-)

I don't have a convertional running watch like a Garmin, instead using my Phone to record GPS and hear rate and communicate this data to my Pebble Smart Watch using AeroPilot AeroTrackerPro Android Phone/Pebble app.  I haven't worked out whether I can record splits on it yet, but since it can upload it's recordings direct to Strava I used Strava's support for Segment to provide the time/pace and HR data.

The outbound leg was into wind, and the return leg down wind which made my first attempt at staying at constant HR rather difficult.  My heart rate varied between 164 and 158 on the first outbound leg as the accounting for the gusty wind was a particular challenge.  On subsequent legs I got better and judging it. 

I was pleased with my pace during the each of the legs, feeling pretty comfortable kicking out each mile at around 7min/mile pace.  At the end of the four miles I stopped and stood watching my Pebble report my heart rate looking for the point when it went below 130bpm.  For the first 10 seconds my HR hardly budged sitting hight 150's, this felt like an eternity but in reality was probably only about 10 seconds, then my HR dropped steadily down to 110bpm and then jogged off.

The final figures report by Strava (click here for the Strava page) for my 1mile Segments were:

    1st mile :  6:56 min/mile, average HR 161
    2nd mile : 6:56 min/mle, average HR 160
    3rd mile : 6:56 min/mle, average HR 160
    4th mile : 7:00 min/mle, average HR 161

Thomas uses a 3.5 sec/mile per beat correction when the averages are not 161.   would add this in, but... I can't help feel that the figures are just too consistent to be exact.  I strongly suspect that the GPS recording affected the time resolution of when I completed each section so wouldn't be surprised if it was being rounded by 4 sec/mile.

Having a manual split recording would be thing to do so I can press a button when I cross the turn about point and have a more precise time value.

Looking at HR data it looks like it took around 25 seconds for my HR to drop from 161 to 130.  Again using Strava for post analysis isn't idea so here a manual button press at when the HR crossed 130 would be best.

The figures should be roughly in the right ball park though.  And they are really encouraging.  Thomas typically sees pace figures around 6:35 to 6:55 mark so I'm actually quite surprised at how close my little legs got me.  Thomas's marathon PB is 2:55 so I'd guess I should be now capable of a 3:05 marathon, but not yet a 3hr marathon.

The thought of running for 3hrs at the same pace as my evaluation is rather daunting though.  I'm confident that I could knock out a half marathon at that pace right now, but a full marathon would be really tough.

Another really encouraging sign is how well I maintained my pace through the four miles.  Sure the method of using Strava and GPS data is crude but I don't expect it that it would have hidden a big slow down.

Finally the rate at why my HR dropped is pretty astounding.  Thomas is normally chuffed is his recovery time to 130 is around 30 seconds.  Again my method of analysis isn't ideal, and could easily be 5 or 10 seconds out, but even with adding 10 seconds it's very respectable.

All these figures suggest that might aerobic fitness right now is excellent.  My training is going well :-)

Easy paced evaluation runs

Doing a 4 mile evaluation on flat route at around marathon pace is certainly useful, particularly if your are training for a marathon as it'll tax your body at the roughly the same intensity as you run the marathon at.  However, we run ultra's at much lower intensity.

One approach you could take is to do the above 4 mile evalation at a lower target HR.  My average HR during 53 mile Highland Fling is roughly 15 bpm's below that of the marathons I've race at, so one could simply run the evaluation at 15bpm lower.  I do think that at such an easy pace the heart rate drift you'd get would be some much lower than at marathon pace, and the so little oxygen debt to clear that the pace variations over the mile segments would be smaller than the timing rounding errors of evaluation and the rate of HR dropping would also be less meaningful measure of fitness.  This would leave the average pace on a flat 4 mile road section at a given heart rate as the only useful measure.  However, I don't run any road ultra's, let alone anything that is flat.

To try and make easy paced evaluations a bit more representative to what type of terrain I encounter in most of my ultra's I have adopted route known locally as the Four Bridges, from my house it's 6.55 miles with ~420ft of ascent/descent with a mix of forest tracks, paths and around 2miles of road/pavement.  This route has been one that I typically run once most weeks during training and since I also record my average HR for each training run I've amassed quite a few runs.  I have also developed a way of normalizing all my other training runs to map them to what an equivalent pace would be on this Four Bridges loop.

Over the years my pace for a given HR has improved, and for this course I've long had the ambition of being able to complete the course at a HR of 140 at 7mph, over the last few months I have been inching closer.  Finally this Monday I did the an easy pace evaluation run (strava page for run) around this route averaging 7.08mph, which equates to roughly to 8:10 min/mile pace on the flat.
Sure this is not a fast pace, but 5 years ago I could hardly run at this low a HR and could only manage around 11min/mile pace. So this is another very encouraging sign that my fitness is continuing to progress.

Using logs to compare Historical Fitness

My next race is the Highland Fling which will be held at the end of April. I have run this 53 mile race twice before, first in 2012 when I completed it in 10hrs and 46 minutes, and then in 2014 when I completed it in 9hrs and 43 minutes.  I have all my training runs from these two years and this year all recorded in the same spreadsheet, and by applying a normalization for elevation per/mile and HR drift all my training runs can be plotted on graph of average HR for the run against equivalent speed that I would have achieved on the Four Bridges route.

The follow graph plots my training runs in Feburary 2012 in yellow, February 2014 in red, and my runs so Far this February (2015) in blue.
Training runs in February 2012, 2014 and 2015, normalized to equivalent Four Bridge Speed

What is immediately apparent is that between 2012 to 2014 to 2015 my pace for a given HR has improved.  The biggest improvement is between 2014 and 2015 - around 0.5mph, with just a modest improvement between 2012 and 2014 - around 0.25mph. 

This would indicate that my training over the last year has been twice as effective as all the training between 2012 and 2014.  Rather than seeing my year on year improvements tail off my rate of improvement has gone around four fold better than it was.

This graph doesn't tell the whole story though.  Between 2012 and 2014 I improved my Highland Fling race speed by roughly 0.5 mph, despite my training paces for a given HR improving by 0.25mph.   My best guess to the reason for this discrepancy is that the training data primarily maps general aerobic fitness, and isn't able to account for the effects of Metabolic and Structural Resilience.

In the Spring of 2013 I adopted a diet inspired by reading the Perfect Heath Diet
 book. This diet recommends a ratio of  50% Fat, 30% Carbs and 20% Protien in your diet, which is quite a shift from the high carb, modest protien and fat that I had previously.  As well as being generally being healthier on this new diet and better able to tolerate higher training volumes I'm convinced that the changing to using Fat as my primarily fuel has helped improve my Metabolic Reslience - now rather than running out glycogen stores half way through an ultra I find I can maintain my energy levels right through to the finish.

Since 2014 I have broadly maintained the diet so I wouldn't expect any dramatic differences in my Metabolic Resilience, but my training has clearly led to pretty remarkable improvement in my general Aerobic Fitness markers. What appears to be the key reason for this big improvement is training Consistently (blogpost: Consistency is the key to improving fitness)- and for me it now means training everyday (blogpost : 100 days later - joys of streaking.)  The remarkable thing for me is that improvements have not been about long runs, not hill sprints, not intervals, not tempo runs, not core work, not strength training it's *simply* getting outdoors for a training run everyday.

So looking forward can I improve my Highland Fling speed by another 0.5mph?  Go for a sub 9hr Highland Fling???

The big question for me now is whether I can build the Structural Resilience to run 53 miles over lots of rough terrain to average sub 10 min/mile pace.

I believe the best way for me to build this Structural Resilience will be via overall training volume and an emphasis on lots of feet of hill descents each week.  Going down lots of hills requires going up them, so this will no doubt help with Aerobic Fitness too.

I am not planning to do lots of really long runs before the Fling, as basically I don't need them as I get all the benefits of Aerobic Fitness, Metabolic Resilience and Structural Resilience via other less stressful training stimulus .  I may do a modest ultra run/race a couple of weeks before the Fling, but will be happy just string together lots of 8 to 15 miler's week in week out.  If you run everyday then in effect every day is back to back which takes the pressure off having to do big single runs, or even big back to backs.


  1. This is all very interesting and something I have been focused on now I am back in full training. Only problem I sometimes get is the Garmin HR strap going nuts at the beginning of a run meaning I have to ignore the HR reading for up to a mile. When doing a race it usually sorts itseld out in the warmup and with this workout requiring a steady warmup then that won't cause much of an issue.

    Would you say the best method is trying to work out my max heart rate and work back from there to determine an evaluation HR?
    Looking at my last full effort parkrun where I have worn a HRM my average was 170 with a max near the end of 187.
    I haven't worn my HRM enough to determine any better maximum or average rate.

    If I was to do this soon, you would recommend warmup over a few miles winding up the pace till you hit the evaluation HR and then hold it there for 4 miles. So lap button press at the start and let it auto mile split for the 4 miles.
    Then at the end of the 4 miles let it auto split and stop to a standstill? Wait till it drops below 130bpm then lap button press again.

    What is the 130 based on? Do you adjust it depending on your max HR.

    You did the evaluation just as part of a normal weekly session (I think) would this be something you would normally do or replace that as one of your speed sessions for the week. Just thinking that if you are looking for an accurate representation on race fitness you won't want to do so on very tired legs, certainly not tapering for it but in a more rested state than say a tempo run the previous day.

    Starting to feel the benefit of getting consistent training in again. A good feeling from yesterday’s 12 x 1:00 session was I covered 4.81 mile, whereas when I done the exact same route starting and finishing the session from the same point and exact same warm up and cool down at the same time last month i only covered 4.44 miles for 15 x 1:00. Encouraging to see improvements.

    1. Max HR is probably something you'll get close to at the end of 5k. There are various online guides for how to determine max HR, general principle seems to work sub max level for a few minutes so the HR is raised to near the max but without exhausting you, then up go all out. A progressive steeper hill is probably a good way to do it.

      How useful Max HR is debatable. Sure it's nice to know, but if one is doing evaluations to training/racing to HR zones then you don't go anywhere near your Max HR. The highest I've seen when going up Ben Guilipen in the last year is 190, but previous years I've seen 193 a couple of times.

      I feel a more useful HR for training purposes is your HR at Lactate Threshold. Again there are various online guides to how to determine it, from going to a fitness lab through to doing a 30 minute run at race pace and then measure the average HR for the final 20 minutes. I personally have used HR during 10k's and tempo runs as a way of gauging roughly where my LT is. Last time I did lots of tempo runs and 10k's my LT HR was around 175bpm. I suspect LT HR is now a bit lower though as my HR at all paces has dropped significantly over the last year (around 10bpm for the same pace), but I don't feel my LT pace has gone up proportionally. Perhaps if I focused on 5km and 10km races I could push my LT pace up more.

      With getting accurate HR reading getting good contact between the chest strap and the skin. Pre wetting or licking the strap contacts prior to your run is one way of helping get a decent contact before sweat does the job for you later in the run. Another route is to use conductive gel. Since using a Wahoo Tickr strap I've found that when I wear certain tops, especially rain jacket I get lots of mis-reading in the first mile - I presume because of static build up. I bought some conductive gel intended for a tens machine and it works well most of the time.

      When doing evaluations the important thing is to do a controlled run where you can consistency reproduce the same intensity and terrain. Using the same route is one of the best ways of doing this. A track evaluation would be great as well. The HR to use will depend upon what part of your fitness you are trying to track, a little below LT/marathon pace seems like a good general purpose evaluation as it fast but not so fast that it's struggle to complete.

      The HR target to aim for at the end of the evaluation is a little arbitrary, again the important thing is the same target each time you do an evaluation, not the specific figure. Using 161bpm target for the 4 miles, then 130bpm target for post run recovery test seems a decent enough combination for giving you enough time to measure it - so 30 bpm less than you evaluation HR would be OK, but rounding it to the nearest 10 would probably make the measurement a little easier on the run fogged brain.

      So if your LT HR is 160, then 150 for the evaluation run and 120 would probably be perfectly reasonable figures to go for.

    2. Second part...

      W.r.t fitting evaluations in, yep I just substituted the evaluation for an everyday training run. Where I did it meant a 3.5 mile w/u to the Loch and 3.5 c/d back home so the total mileage was 11 miles. This mileage is up at the upper end of what I can manage to run on daily basis (I've built up to averaging 9 miles right now) so the addition of 4 miles @ marathon pace made it a bit of workout so I had to be careful about running easy the day before and after. As speed session it's pretty low key though, so a bit of different stimulus than a tempo run or interval session.

      I have found one doesn't needs to do lots of high intensity work to build good aerobic fitness so only doing 4 miles @ marathon pace this week will be fine. Next week I can go back and do a tempo or an interval session.

      I personally wouldn't do an evaluation run carrying too much fatigue. The key with an evaluation is that is consistent and reproducible so that you can compare progress month on month. Carrying fatigue into an evaluation will likely make the results a far bigger measure of level of fatigue than aerobic fitness.

      When I do my training log spreadsheet I have a column for "Effective Calories per mile" (I say Effective and I normalize the data so account for hills and HR drift) this gives me a measure data to day measure of my efficiency. In general as I get fitter my Effective Calories per mile goes down. Having this measure gives me an idea of general progress as well as showing when I'm struggling to recover for faster or longer runs, or simply that I'm coming down with a cold, or had a bad nights sleep.

      This type of day to day evaluation isn't a controlled test like the 4 mile evaluation so doesn't replace it, but having these different types of ways of consistently measuring fitness is helpful it get a full picture.

      With your own training, it sounds like the interval session does suggest great progress. It's not a controlled experiment like the evaluation so not reproducible in the same way but it sure is useful to start recording patterns in training like this for signs of progress or possible over-training.

      All this data is only useful if it enables you to make decisions about what to do next. If you continue to make progress then your training is working, if you stall or find yourself falling back then it's time to spot the problem and fix it before injuries or over-training nullify all your good work.

  2. What do they say about the most sincere form of flattery? :)
    When I ran the 2:55, my evaluation was closer to 6:20-6:25 pace (on the Garmin, that is).

    You should find a more accurate way to measure your data, though. 4 seconds rounding in pace is okay I guess but I would be surprised if your 25 seconds recovery time is correct. I have never seen such a quick recovery, neither from me nor from any of the other runners that MC used to coach.

    1. The bit of imitation that I'd love to flatter you with would be matching your 2:55 marathon :-)

      Currently I can't imagine pulling off a 3hr marathon, it might be possible one day, but I guess it's something I'd need to focus on for a while, rather than these lovely trail ultra's that we have in Scotland :-)

      I agree about finding a more reliable an accurate means for timing. I've gone back over the Strava graph of pace and HR for the end of the run, the two data points that are either side of the running section and below 130, are 6:50 pace @ 59:33 with a HR of 160, and then at 1:00:00 with a HR of 125. My HR at 59:50 was still 159, so it stayed high for 17 seconds, then plumpted rapidly to 125 over the next 10 seconds. 6 seconds later it was 115, so at that point I just jogged back home.

      Looking at the GPX files that I can export from Strava the samples come in at 2 and 3 second intervals. I presume this is what my phone app is recording rather than Stava re-interpreting.

      It could be that my Wahoo Tickr HR monitor is not suitable for detecting rapidly changing HR, but usually when I run it doesn't take too long to respond. I don't get the sense that it's as good as my old Polar watch/HR strap though, but this never had an facilitate for recording or download traces so there is no way to compare.

      Ideally I'd have a function in the phone software that could compute the time times for me, accounting for the fact that the samples are far enough apart that they cross over the 130 range very rapidly and can estimate the cross over point. A manual time marker for the end of each segment would be useful too. I will investigate to see in my phone app has this facility, if not then might need to be speak the phone app author with a "great feature addition".

  3. An interesting blog. Yet further evidence that your fitness has improved greatly with the current diet and training regimen.

    I suspect that resilience of your leg muscles will be a key factor in the Fling. Doing some down-hill running sounds sensible. However it is difficult to know how to test for improvement in resilience during relatively short training runs. I consider that leg muscle resilience is also key factor in the marathon and I have been experimenting with counter-move jump before and after a training run. However I do not have any way of measuring jump height accurately enough. The preferred equipment is a force plate. Also I am a bit concerned that a max effort jump is itself somewhat stressful

    1. Doing plenty of downhills seems to work really well for toughening up my quads so would recommend this. My calves still seem to be a weak point though, perhaps jumps/drills would be helpful here.

      W.r.t measuring hight, could you measure the time time in the air and then calculate the height from this, the maths should be pretty simple. A slow motion camera would be one way, another might be to use a device that you carry that is able to measure time in air and time on stance - some phone apps can do this, as can some recent sports watches. A carried device would be subject to issues of the device not being tightly coupled to the persons mass so would not be a perfect gauge but you might well find them reliable enough to gauge trends and make comparisons.