Monday, 27 August 2012

Ironman UK performance analysis

As I write this I'm on a plane on my way out to the US for another lightening fast visit. Arrive Sunday night, one night in New York, work Monday, and fly back overnight landing on Tuesday morning in time to go to the office in London. Don't let them tell you international travel is glamorous.

It's an important trip, and I can see the need for me to be there in person on Monday, but I'm a bit annoyed at missing what had become until recently a bit of a bank holiday ritual. The Monday morning 4am start to get up and head over to Hampton Pool for the wonderful Thames Turbo sprint series races. One is held on each of our bank holiday weekends, so there are three in close company at the start of the year, and then one in August.

I had decided to skip race 3, instead getting some extended training in the bank, but had been looking forward to race 4. In theory, I've never been this fit - how much of that transfers from the long course racing for which I've been training to the short course sprint format is uncertain, and I intended to find out tomorrow with a balls-out (metaphorically, I assure you) effort.

In any case, instead I'm an hour out from Heathrow over the Eastern Atlantic on my way to Noo Yoik. I shouldn't complain too much - without my job with the good folks at Razorfish I wouldn't be able to fund my Ironman habit.

Recently I've been reading the rather excellent blog of Russell Cox, entitled Trains & Travels. Lately he's posted a number of articles looking at the relative performance of athletes as benchmarked against their peers and other performance levels in their race. I've always done this with my Thames Turbo results - and they've always shown the same patterns:

The swim: I am roughly at the bottom of the top third. I've been improving, slowly, but not very rapidly. I haven't competed in a Thames Turbo race since I've re-learned to swim (more on that in the future). Recently I've been 25-30% down the field overall, but, as Male 30-39 is a high performing age group, this translates to only about 35-40% in my age group.

The bike: This is where I'm strong. In Thames Turbo races I can get as high as the top 10% in the overall field. I'm also strong in my age group, hitting top 15%.

The run: Do we have to talk about the run? In my first two Thames Turbo races I was beaten by almost 83% of the entire field - I am not a runner. My best run performance was Race 4 last year where I sneaked into the top 50% - but still got beaten by 65% of the others in my age group.

Overall: My strong bike pushes me up the rankings. In recent races I have come in the top 25% of the field, and the top 35% of my age category.

Disclaimer: My Thames Turbo PB was in Race 1 of this year - I haven't conducted my usual analysis on those results as they've stopped releasing the immensely useful .xls of all the results, and have moved to an equally immensely annoying web-based results tool. Colour me unimpressed.

Russell's blog got me thinking about how this performance benchmarking compares to my inaugural Ironman. Would the same pattern be revealed? If so, what action can I take to produce a better performance next time?

It just so happens that I checked the Ironman UK results page yesterday, and they have now released a .xls of the full results. I'll say right now - all race results data should be released like this. Today's Internet is all about the cool things you can do with data - and the first step is making the data available in an easily consumable manner. Coming third only to well documented API and a tidy .csv is the glory that is the Microsoft Excel spreadsheet. Fancy web front-ends and search tools are great, but please, race organisers, make them your way of interpreting the results for those that just want to click about, don't make them the only way to get to the data.

So what does the spreadsheet tell me? It tells me the following:

My finish was a bit behind where I would expect to come in shorter races. The first race at a new distance is unlikely to give your optimal performance, and I suspect my run will have brought me down the field.
  • 501st out of 1,180 overall (42.5%) - meaning I was "chicked" by 39 women
  • 462nd out of 1,047 men (44.1%)
  • 103rd in my age category (M35-39) out of 227 (45.4%)
My swim was pretty average compared to previous performances. Interestingly I was higher up in my age group ranking than overall - this implies that M35-39 is behind the curve in Ironman swimming, as opposed to my local sprint races where it's one of the strongest groups.
  • 408th overall (34.6%)
  • 357th male (34.1%)
  • 74th M35-59 (32.6%)
My swim to bike transition (T1) was well inside the top third. Again we see the M35-39 group are a little behind the curve - come on guys, get your fingers out!
  • 343rd overall (29.1%)
  • 303rd male (28.9%)
  • 64th M35-39 (28.2%)
The bike was where, in my opinion, my race started to diverge from the plan. I went out far too hard and thought that would massively impact my ranking - I'm still in the top third, but no where near where I would hope to be.
  • 350th overall (29.7%)
  • 329th male (31.4%)
  • 70th M35-39 (30.8%)
The bike is where the spreadsheet gets really interesting. It lists the split times at 7 points on the bike course (to this day I have no idea how they did that - I don't recall going over any timing mats).

My recollection of the bike phase is that I went off like a rocket. I thought I was under control, but I was flying past people like they were standing still. I decayed hard over the second and third laps, and picked up a tiny bit (probably relief!) towards the end of the third lap.

The stats seem to validate my suspicions, only I was decaying way sooner than I realised. In fact every split was slower (relative to the field) than the one before. I managed a very minor improvement on the last split, but that was only 2.3 miles long so I don't think it counts for much! Needless to say, this is not a good example of Ironman pacing. 

However, I was significantly faster at the start than I realised, my 12.5% overall for the first split puts me in the top eighth of all competitors! My assumption that I'd drifted down to half way or even further by the end was based on my starting pace being average - I didn't realise I'd set off so relatively hard, and therefore although I slipped a long long way down, my finish was still quite good (relative to the field, if not relative to my normal short course triathlon performance).

A similar story can be seen from my average speed.

It looks alternately slow and fast because the 3-lap circuit had a very definite slow and fast half, but of note is that the slow halves get slower each time, and the fast halves do too. My pace was dropping continually throughout the bike phase - this is a real eye opener, I had expected a period of good stable performance followed by a rapid drop off, not a decline basically from the word go!

So how does this compare with everyone else? I figure the top 20 finishers should know a thing or two about pacing - after all, they're the only group who went under 10 hours, with 20th place just sneaking under at 9:58:53. This group is all male, the first female was Irish professional Eimear Mullen who came in at 10:08:44 - an incredible performance for her first full distance race.

So it seems the top guys also fade, just like I do - exactly like I did, in fact. Only they are going faster to start with, and they drop off at a lower rate (their reduction is therefore much less impacting). This isn't quite what I expected, I assumed that there was a magic pacing ability the top performers have and their splits would show that the three laps were taken at a similar pace each time. Actually, their pacing pattern was much like mine, only they're much fitter and can therefore go faster and resist the drop off in pace for longer.

If I conserved my energy better by starting more within my capabilities I should find that, although my initial velocity might be marginally reduced, I can limit the rate of decay of pace, and therefore my plot would look similar to the top 20, only shifted down by a few km/h. Overall this would give me a better finishing time. Something to think about for Galway where the bike course is virtually flat. If I can get the pace right from the start and hold it right through to the end, and still have good legs for the run, then I'll have got it right and be deservedly pleased.

I'm expecting my bike to run transition to be poor - basically I stopped for a picnic, application of sun cream, a chat, and generally took my time. Mentally, I needed the break, but race-wise I suspect it cost me some places.
  • 1,011th overall (85.7%)
  • 896th male (85.6%)
  • 199th M35-59 (87.7%) 
As expected - terrible! The stats show that at the slow end the M35-59 group were marginally better than average, but when you get down to the back 15% it's all much for muchness. Any longer and I'd have squeezed in a nice cup of tea (now I've thought that I'm wondering about the practicalities of leaving a small Thermos in my T2 bag, hmm...).

One of my objectives for Galway is to speed up my transitions. I needed the break after the challenging bike ride, but a better managed bike should mean a faster T2 (suggesting a well paced bike makes more difference than just that revealed by the bike split time alone). For reference, the fastest T2 was a breath-taking 1:16 by Rob Cummins, an Irish athlete who was placed 48th overall.

Onto the run, and again I'm expecting a picture of starting too fast and fading fast. If the Thames Turbo pattern holds then I'll be significantly further down the field in the run that I was the swim or bike.
  • 747th overall (63.3%)
  • 669th male (63.9%)
  • 158th M35-59 (69.6%)
So I just squeezed into the top two thirds overall, and the top two thirds of men, but my age category is clearly comprised of stronger than average runners, therefore I got beaten by very nearly 7 out of every 10 of them.

Again, the spreadsheet gives a lot of splits. Let's see if the run really went the way I remember it.

This one is a bit different - it looks like I was gaining places overall for the first 13 miles, unlike the bike where I was losing places form the start. The trend reversed fairly quickly, and I started drifting back relative to the field at a pretty high rate. 13 to 21 miles was really hard, and I got slower and slower, and then managed to halt the decline for the last couple of miles (possibly because I knew the end was close, and also it was downhill!).

Basically, my pacing sucked, and I went off at an utterly unsustainable pace.

Compared again with the top 20 finishers shows my pattern was fairly representative, but once again I started a lot more slowly. My pace from 17-20km didn't pick up again when compared with the top finishers - I expect this was one of the long gentle rises that I just couldn't get going on and ended up walking most of the way. The top 20 picked up the pace again and kept running.

Note here that although it looks like our paces were declining at the same rate because the trend line is roughly parallel, this isn't the case. My drop of about 2.5km/h represents a proportionally greater reduction given my starting pace of only 9.5km/h (26%), whereas the top 20 dropped a similar amount of about 2km/h but starting much faster at 14km/h (14% drop).

As with the bike, I expected the top 20 finishers to have a magic pacing ability that showed they hold an even pace throughout, but this isn't the case. They gradually slow down along with everyone else (along with me, at least). Where my curve differs at 17-20km I was clearly in trouble, this matches my experience - this was the darkest point in the run, although I felt mentally alert (I take that as evidence my nutrition was good) my legs just wouldn't do what they were told. Even the technique of setting small targets (the next aid station, landmark, or sometimes even lamp-post) was not working - legs just went "sorry, no way" and would play ball. Or run.

I'm taking a few things from this analysis:
  • My pattern of pacing decay is fairly representative.
  • My rate of decay is proportionately greater than those at the sharp end of the field.
  • The top 20 are faster at the start, faster in the middle, and faster at the end!
  • My swim is OK - although a third of the field beat me, the time difference is comparatively small. The 9th place finisher only beat me by 3 minutes, and despite being 22% faster the top 20 was only on average 16:30 ahead of me. 
  • My run is by far the weakest discipline, and that should be a major component of my winter focus. This is where I faded badly, and lost the most time.
  • To move up to the next level (12 hours would seem to be a reasonable target) I need to improve my run, and get better at managing my effort on the way there.
Importantly for this year:
  • Even if I control my bike better in Wales I'm not going to pluck an amazing run out of thin air.
  • The bike course is a lot harder than Bolton, and as a result I'm not sure I'll be able to set a better time, even with better pacing.
  • I now recognise that at Wales the challenge for me really is in getting to the end.
In my next races I should (hopefully) have my watch working properly so I can map all this against my own data rather than just the published splits - that will give me a lot more insight.

No comments:

Post a Comment