Trains, Travels

Perhaps it’s the weather. As I basked in the sun on the homeward leg of my early morning run I had a thought, “maybe I can race well later this year.” The arrival of an unusually warm spring has broken the negative cycle – my mood is lifted and my desire to train, specifically to run, has returned with full force. And so positive thoughts arrive, the idea that perhaps I can be both fit and capable of racing well enters my head, only compromised by the knowledge that it will take some time and some planning. The path ahead is littered with events, too many to hope to be on top form across them all. I need to take the smart approach, utilising races as part of my development, rather than an end in themselves, stepping stones to greater things.

I suffer from an allergic reaction to strict training schedules, and while there is a strong case – one that made me stop and think – for the benefits having my own coach would bring me, it’s not something I want to pursue right now. But I need something more than loose guidelines to make the longer term transition to greater fitness, something that appreciates the impact of my heavy race season, using those events to build me, not break me. Rather than limiting my use of the Performance Management Chart to retrospective training critiques, I can use it constructively to manage this coming year; utilising my experience of the system to set training targets and a rate of growth for the weeks and months ahead. Perhaps it seems at odds with that desire to train for pleasure, but there are many ways to achieve fitness growth.

I’ve dabbled with the approach before – planning tapers or testing out ideas for a build; manipulating the underlying values of the model to see the effect different work loads are likely to have. It’s a matter of trial and error, adjusting those numbers to produce the pattern I want. At this point I am guided by experience, those times I’ve analysed historical PMC data actually come in useful as I avoid my previous mistakes and attempt to replicate the successes. I tweak until I produce a pattern I like.

By now a familiar chart, but this time the product of good intentions. Reviewing the last four years of run training showed 2009 to be my most successful season, a race heavy year much like the one ahead. I achieved it through a consistent, moderate training plan, 30 runs of at least 30 minutes in 30 days forming a large part of it; I could take that consistency, without excessive overload, and apply it again. Firstly building up through April – perfect for thirty days of running – then maintaining the run fitness from race to race, my work then being to minimise fitness loss. There isn’t time to chase high peaks of fitness, 2010 has taught me the danger of building too rapidly; the volume of running this pattern suggests is a sufficient challenge after a season out of running.

It sounds like a plan, albeit an abstract one defined by the shape of a curve rather than the details of sessions; and it’s that abstraction that gives me the freedom I desire. The curve is the product of a training load deliverable through whatever (sensible) combination of intensity, duration and frequency I desire and I’ve already indicated my preference for a repeat of the high frequency, mixed duration and intensity approach to run training that worked well in 2009. As long as I accumulate a sufficient training load each week my fitness will progress in line with the chart, within reason further details don’t matter; my training is flexible to match my moods.

Finally I’ve found some much needed direction and – perhaps – a way to guide myself towards this goal without feeling overly restricted by a plan. I may not have written a single session down, but I know the kind of work I need to do; starting in April with thirty days of running at least thirty minutes.

• Tweet
• 4 Comments

For a long time I was a running luddite. My training tools were perceived exertion and at most a stop watch, sometimes I simply checked the clock before I left and on my return; I only touched technology to find out how far I’d run – resorting to Google Maps to estimate distance. That changed with a Christmas gift of a Garmin FR60, enabling me to record training statistics in a similar fashion to cycling. But I was resistant and would often neglect the heart rate strap, it wasn’t like I looked at this information while running. The collection of data was sporadic and incomplete; even when I upgraded to a Garmin 610 on my last trip to the US, I still tended to leave that heart rate strap behind.

Predictably, once I’d finished playing with my four years of cycling data, I turned back to running. Every run has been recorded, but those luddite ways limit the depth of information, where cycling is data rich, running is impoverished. Duration and a rough estimate of speed stretch back four years, cadence is present for most of the last two, heart rate only occasionally crops up. I was keen to apply the same analysis I had just performed on cycling, but this was tempered by the awareness I probably had less than a year of rich data to use. My expectations were low.

The last year holds the highest quality of data, so for the purposes of the charts I focussed on this. Even then the sparsity of heart rate data in comparison to speed is notable, in this regard I seem set in my ways. Cadence is consistent throughout the period, hovering on average just below 90; speed and heart rate have apparently fallen over the year, something to be concerned about? At least until I remember that each point on the chart is an average, the downwards trend in run speed is a product of what I’m labelling the ‘Girlfriend Effect’ – regularly training with my partner at a slower pace has pulled the trend line down. I’m not concerned, a second cluster of solo runs shows me holding a comparable pace to last year; I’m happy to be building from that point. With a reasonable correlation between speed and heart rate, this has fallen on average too.

It’s unfair to place all the blame on my partner, such a small data set using per session averages is bound to be susceptible to outliers. The problem is highlighted by a group of runs averaging below 9kph, these seemingly pedestrian efforts will have actually contained some of my fastest running of the season. A single average speed makes a poor representation of interval sessions involving strength or technique work during the recovery; 5 minutes of hard running followed by 3 minutes of static activity will register as slow. And of course I can’t discard the fact that just as for cycling, speed is a poor metric of performance, highly dependent on external factors.

So when I plotted the scatter graphs to look at correlations I also incorporated Normalised Graded Pace, a concept similar to normalised power, attempting to adjust pace to account for terrain and better represent the effort involved.

Having identified some of the difficulties with averages in my run data I didn’t anticipate many useful correlations. What is there, is what you might expect – a relationship between heart rate and speed or pace, run faster and your heart rate will rise. There is also a relationship between cadence and pace with faster running tending to be related to slightly higher cadence. It’s small and at the other end of the scale those interval sessions come into play, when I’m stopped between reps average cadence is rapidly pulled down. That said it feels easier to hold a high cadence when running fast, those ‘Girlfriend Runs’ have challenged me to maintain stride rate at a much lower effort.

This was more an exercise in completion, matching the work I’d done with the cycling data and in the process highlighting many of the limitations in both analyses. It did confirm my sense that cadence is solidly ingrained and select workouts aside my pace remains similar to last year; given the level and quality of winter training I’m happy with that, for now. And it comes with an unintended consequence – I’ll be wearing my heart rate strap from now on, just for completeness. A repetition of this analysis another year on should be much more detailed and precise.

I’ve spent enough time digging through history for now, I’m putting the data to rest. There’s more I’d like to do, I’ve pondered the idea of building a database of raw session logs and looked into developing new charts using R. Projects for the future, it’s time I focussed on data production.

• Tweet
• Comment

It started with a distraction. As I plotted the four year performance management charts side-by-side I was – as usual – side tracked. A blog post and discussion of using the new Garmin 910XT to measure swim metrics was far removed from the topic, data collection being the only common theme, but my experiences with the Finis Swimsense drew me in. Conversation on how you could utilise stroke rate information turned to the broader topic of cadence applied to swim, bike and run; from utilising a stroke rate ramp test to applying the principle to the other disciplines. Would a structured ramp test of cadence be worthwhile on the bike? I suspected not and with 765 separate ride files already in use building my charts, I had the data to look deeper.

I have never specifically worked on cycling cadence. I’ve simply concerned myself with developing the power I can sustain for a given distance and let cadence follow, whatever felt comfortable. In my case that’s a relatively low cadence, I don’t spin in training, preferring to push a big gear; it works for me – bike power has increased over time and my run performance relates more to fitness than the cadence I hold for 112 miles. Reading the comments of other coaches and a handful of papers suggesting cadence should vary according to the type of cycling has helped solidify my view that for long distance time trialling what matters most is sustainable power, cadence can be a distraction.

Still focussing on four year trends, my first consideration is the pattern in cadence over that time using the average cadence from each ride. As usual there are limitations, ride averages incorporate freewheeling and terrain will have an impact – when I ride in the Pyrenees I’d expect a lower cadence up the steepest mountains and freewheeling during the descents. With no easy way to summarise the information more accurately, I have to accept these limitations and hope the dataset is sufficiently large to prove useful.

Cadence takes a curiously sinuous path from a high start in 2008, dipping to a low during 2009, before rising into 2011. In part, as is often the case, the product of my training – until June 2008 I frequently used a turbo trainer, during 2009 I mainly put in long steady miles outdoors and I started 2011 with a winter riding only in the little ring. Throughout this period the average heart rate for any given ride has remained largely unchanged and overall the normalised power has risen, taking a similar path to cadence. The routine turbo sessions of early 2008 would have higher normalised powers, in their absence I was simply riding harder during 2011. What interested me was how power and cadence seemed to track each other.

Slicing the data in another direction – a quick scatter graph – seemed to support that increases in average cadence were matched by increase in power. Anecdotally this matches my perception that I tend to spin faster when putting in an effort; a Maximal Aerobic Power test (not in the data set) had shown an extreme of this – I averaged 100rpm throughout. So in self-selecting cadence it seems I have a slight tendency to pedal faster when I want to go harder, like I am tuning the RPE to match the output, but it isn’t a strong instinct, I appear most comfortable around the 80rpm mark.

I thought I’d go one step further with the scatter graphs…

The data is there for the comparison, so why not cross reference cadence, heart rate, power and speed to see what I find? Unsurprisingly speed tells me next to nothing, as ride metrics go it’s a poor measure of training, too dependent on external factors to say anything much about fitness. Power and heart rate trend well together, the best relationship on the chart, with the tightest distribution; not surprisingly, averaged over many rides we’d expect that higher power means a higher heart rate. Again cadence trends a little with power, and perhaps heart rate, but as before that correlation is comparatively weak.

An interesting detour. Plotting the data does nothing to change my view, the patterns I see are largely what I expected, though the way my cadence varies over four years is surprising. There is certainly nothing here that will change how I train – I’ll continue to regard cycling cadence as largely a red herring, power matters more to Ironman performance. And I remain convinced that a specific test session of cycle cadence will be largely uninformative.

Finishing with a second anecdote, riding with power for a number of years I’ve played with gearing and seen many occasions where a shift up or down by a single cog could produce a jump in power with little change in perceived exertion. It doesn’t always last and during many intervals I’ve found myself shifting between two gears, adjusting cadence back and forth, as each in turn feels more comfortable. All the time holding my power around my target, but cadence varying according to what feels right. I’m comfortable and I’m riding well, I’m not convinced time spent training towards riding that power at a specific cadence would be a good investment.

• Tweet
• 1 Comment