My motivation to compete has largely vanished. In its wake a lack of focus leaves me training in an ad hoc fashion without real gains in fitness. I’m fine with this. In contrast Gill is highly motivated to compete, but persistent back and leg issues have left her training in an ad hoc fashion without real gains in fitness (buying a house and organising a wedding hasn’t helped us either). She is not fine with this. The half, as a relay, was a compromise that even our sporadic schedules might be sufficient for. We were wrong about that.

Being insufficiently prepared to bike or run, at least without subsequent back pain, left Gill with the swim; one of my athletes, Tim, was drafted in for the bike; and being insufficiently prepared to run left me to handle the run leg – I was just going to suffer more.

The runner at least has the potential for a lie in. That is unless they share a bed with the swimmer – I was up at 5am (a time usually reserved for my midweek coaching) a lie-in by most competitors’ standards, but still 6 hours before I would need to run. Gill made herself breakfast and pulled on her new wetsuit loaned to us by Blueseventy (she’d opted not to try the Helix in case she was tempted to buy); then we checked out and drove to the race venue. Our preparations were done and Tim had got into transition and racked before the crowds. There was nothing to do, but wait.

Much of a relay team member’s day is spent waiting. Tim and I waited to see Gill start the swim, somewhere in a mass of black wetsuits and red swim caps. Then while Tim headed to transition ready to meet her on exit I found a spot of grass opposite his bike.

Unlike a regular race report all I can tell you about the swim is what Gill told me shortly after she finished. Firstly, and most importantly, that she exited the swim with a new PB and secondly that she swallowed so much of the rank lake that she vomited during the return leg of the swim. A good result – the PB rather than the vomiting.

With Tim out on the bike there was little to do, but it’s hard to relax knowing you have to run a half marathon soon. Race nerves had kicked in. I feared for the second lap – my lack of preparation would show. Then there was always that chance I might miscalculate his return and leave him alone in transition. Naturally I headed down to our racking far to early so as to be safe.

Again I can only offer a second-hand report of the bike. The roads are rough. Tim’s Garmin mount broke in the first mile so, with bike computer in his back pocket, he rode by feel. A check halfway revealed he’d ridden far harder than intended, but feeling strong he carried on. The result was some new power PBs and an FTP test in the schedule.

Minimising the suffering was my main concern. Intentionally feel would be my only guide for the half-marathon. The first few miles felt good, the pace didn’t feel hard, but I knew that in a race I tend to push myself. I decided to stick with it and see what happened. The first lap passed pain free, but at the far end of the lake things had started to get tougher – the next lap could be a different story.

This was a mental game. I broke the race into stages: past the boat house, to the aid station, through the turn point, back to the lake, to the end of the lake and then – finally – the stretch home. Each step was harder than the last, but I stayed focussed and pushed myself on. The sooner I finished the sooner I could relax. Somehow I found energy for a few bursts of speed taking a few more places before I peeled off into the finish chute. Done.

I was pleased with the day. Gill’s PB of 37:57 for the swim, Tim’s performance (and power data) for a 3:10:34 bike and that somehow I could still knock out a 1:33:59 when I had trained so little. Most importantly though the day had been fun.

If one team was fun, then more teams would surely be more fun. A get together and some racing for some of my athletes. Perhaps put a team together for a fast time. The run might be an exception, but swim or bike offered good training sessions – most people could do them with little or no impact on their season plans. The idea was formed: multiple CoachCox teams in next years Outlaw Half.

In the meantime, to enjoy more race atmosphere, Gill and I plan to spectate at Wimbleball (weather dependent perhaps) and we’re both contemplating another 10K swim with every intention of preparing thoroughly.

Last year I briefly compared the pacing of stand-alone marathons with Ironman marathons to see how they differed. One observation at the time, which can be seen again in the chart above, is that on average there really isn’t much difference in pacing strategy regardless of finishing time: most go out fast and finish slower than they started, this is most marked among the slower runners. The classic negative split is not the average. Interestingly, assuming timing mat placement is precise, the halfway point provides a psychological boost with many running the kilometre that leads to the half marathon split faster than the previous 5K.

I haven’t had time to examine the results further, with 35,000 finishers this is a much larger dataset than I usually work with. Too large for Google Drive, so instead of the usual spreadsheet I’ve uploaded an unwieldy (12MB) zip file containing the full data in Excel format. You can download the amateur splits from the Virgin London Marathon here (click File and select Download to save a copy). I’ll take a closer look at the results later this week.

Would we have seen the pictures of packs from Melbourne last weekend had a wave start been adopted for the shortened swim? That’s the question I’m hoping to answer for the IM Talk Podcast next week having been asked by John Newsom to look at the statistics. So what might a wave start have looked like in Melbourne?

The answer – potentially – is shown in the graph above. If we are willing to make an assumption it’s actually relatively easy to simulate the impact of wave starts using the result from last week’s race. If we assume that wave starts would not alter an athlete’s splits then I can simulate them by simply adjusting race times recorded at each mat. In this case I’m interested to see what the congestion would be like in T1 if athletes had been started in separate waves.

I tried a number of approaches to the wave start. Firstly I divided by gender, but it soon became obvious that wouldn’t work – at least 75% of the field was male, women were not the cause of congestion. More waves are needed and they need to target age groups, once I made those divisions I started to make inroads into the congestion. It was only when I applied a more intelligent split into five waves that I really made a difference; I ordered age groups by average swim speed and grouped them so that waves were around 400 athletes in size, as small as I could go without dividing the male 40-44 year old division. As the graph shows this does a better job of splitting the competitors at the start of the bike, but did require the benefit of hindsight. More waves with the fastest athletes first is the most effective way to reduce congestion at T1.

Out of interest I also tested how a full length swim might have affected congestion. Had a 3.8K swim been possible, and had the athletes maintained the same pace for the extra distance, we would not see the same problems. It seems the issue may be less pressing when a full length, challenging swim begins the race.

Of course we can never know what would have happened, a switch to wave starts would change the dynamics of the race. There might have been fewer packs, but there would still be drafting. It’s hard to measure the benefits when statistics on drafting are unreliable. Are the gains enough to warrant a change? I can’t say. There are also implications for the competition: I was careful not to divide age groups to ensure head-to-head competition for Kona slots and podium places, but there are Ironman races, like Frankfurt, that do. How much affect does that have? And what about those races that use time trial starts? Texas will be using that method, how does it influence a race? As so often happens one simple question leads to others.

To answer John’s question though, wave starts would have alleviated the congestion at the exit to T1 and this would likely have made drafters easier to catch. But it would also have largely been unnecessary had conditions allowed a full length swim.

In simple numeric terms if you have a 12m drafting zone, as they did in Melbourne, then you can fit roughly 83 perfectly-spaced athletes in a 1km stretch of road, with a field of 2000 you need around 24km to correctly fit everyone in. The process of legally overtaking may mean exact distances vary, but basically it takes a lot of road to space out that many athletes. So when you shorten the swim element of a race effectively bunching the field closer together, a crowded bike course is guaranteed.

The impact of the shortened swim is quite obvious when you look at the number of athletes leaving T1 at minute intervals. In 2012 there were fewer competitors and a full distance swim, at no point were more than 50 athletes leaving transition during the same time frame. In 2013 though, with higher numbers and the reduced swim distance, transition peaks with 100 athlete all leaving in the same minute interval. It may have been necessary to shorten the swim, but the crowded bike course and formation of packs that followed was inevitable.

By the time the athletes return to transition the numbers have changed, as the graphs above show the rate is much lower (each column now represents a 5 minute period). The bunching that we saw at T1 has diminished and groups have generally split. The pattern is a little denser in 2013 than in 2012, so it appears they haven’t broken up as much, but this may be due to the larger field. Packs formed early in the race when the swim failed to separate the competitors, 180km later groups have tended to disperse and spread; there is little evidence of strong pelotons working together for the entire course.

There were timing mats every 45km in Melbourne and I can use that data to look a little more closely at how the groups dispersed. The graphs above allow easy comparison of each split. By 45km the density seen at the exit of T1 has significantly diminished with a peak of 208 athletes in a single 5 minute window; this continues to drop, peaking at 162 at 90km, 123 at 135km and 81 at T2. All good signs that packs were generally separating. I think it’s reasonable to say that the majority of those who found themselves caught in a bunch due to the density of competitors at T1 were likely riding clear by the end of the bike. Of course there will have been some who choose to draft, but this is a much harder contingent to assess. Timings of athletes at static points on the course reveal little of the overall dynamics of the race, simply being close to another athlete at a timing mat does not mean you were drafting.

I am experimenting, examining groupings of athletes over consecutive timing points and identifying those who maintain close proximity across the course. Too raw to give numbers from and it has its flaws: it can’t distinguish a legal pace line from a team time trial, nor could it spot an opportunistic drafter who jumps from athlete to athlete. Post hoc drafting identification will never be possible, but perhaps there is a way to compare groupings on different courses. With more work I may be able to suggest which courses see the highest rates of pack formation, I suspect it will correlate with number of athletes. In the meantime, for the curious, the chart below gives the athlete density at T1 for all the 2012 Ironman races.

Drafting will always be a problem, there are always those willing to cheat, but the density of athletes on a course amplifies this by drawing in those who might otherwise ride honestly. If you left T1 surrounded by dozens of other competitors, would you feel it was your obligation to drop back or theirs?

In raw form we see a race not too dissimilar in bike or run, perhaps marginally slower, but barely, and in the overall ever so slightly faster thanks to the much shorter swim. Conditions may have been tougher, but not so much as to make a significant dent in race results. However to better compare the two years I can apply the adjustments I made to the professional times yesterday, correcting swim times under the assumption that each athlete would maintain the same pace. The results below show a similar, but marginally slower race than in 2012.

Another way to consider the race is to consider the median split times across the different age categories and professional ranks. It eliminates some of the impact of the fastest and slowest and gives some indication as to the relative difficulty of each stage of the race.

For the most part, in both bike and run, 2013 is slower, but not always and not always by that much. Notably the slower average bike time of the male pros is countered by a faster average run time and the average bike split of the female pros was faster in 2013. As a general guide though we are seeing slightly slower splits, but as is often the case the impact of poor conditions is actually smaller than we imagine.

The front of pack can give a very different picture to the overall figures, particularly in a race that draws a strong age group field with a lot of Kona slots. Examining the top 20 in each age group may give us a slightly more stable set of results to compare across the years.

Each graph above also includes the “corrected” finishing times where I have proportionally adjusted the swim time. Obviously before adjustments the top 20 times are quite consistently faster than in 2012. Adjusting the swim times to estimate what might have been produces a more mixed picture – some performances still beat the 2012 equivalent while the majority fall slightly short. The differences are surprisingly small though, hardly significant over the duration of an Ironman.

I have heard stories of packs and drafting on the bike, but then I’ve heard these stories about every race and it’s not something I can discern from the timing data. Certainly with the shortened swim the number of athletes exiting transition at the same time would increase noticeably and this would likely create more bunching on the ride; groups would potentially reduce the impact of windy conditions on bike splits and subsequent run splits. I can’t say if it was a significant factor any more than I can for any other race and although it is an unfortunate reality of our sport it’s not an area I’m going to speculate on (for now at least).

You race what you get on the day and this time round that involved high winds and a short swim. Obviously a shorter race means quicker times, but adjusting for the difference in length shows performances on a par, or perhaps slightly better given the conditions, than 2012. Ironman Melbourne appears to be a fast race even in difficult circumstances.

There are many ways I could have dealt with the abbreviated swim. I could, as I have with shortened New Zealand races, simply ignore it and choose never to compare Ironman Melbourne with other events, but that’s not very interesting. I could, as I have with the current assisted New York swim times, only compare bike and run times between races to eliminate the differences, but this underplays the impact of a short, but tough swim. Alternatively I could make an assumption, as I have in the graph above, I could assume that every athlete would maintained the same pace over a full 3.8 kilometres as they did over 1.5K (which it appears the final swim distance was) and I could adjust their times accordingly.

The results of that very simplistic adjustment is the sometimes slower dashed blue line of the “Corrected” 2013 times in the graph above, it’s suggestive that conditions were tougher this year and while the shortened times are faster, the equivalent times would be fractionally behind 2012’s. Of course it’s speculative and based on a major assumption. If you change the swim you change the race, the dynamics of the day would shift, different packs would form, different pacing strategies would be adopted, but it may be the only way I have to consider the swim times. I’ll add the same adjustments to my age grouper charts tomorrow for comparison.

The 2012 Finisher distribution below highlights the speed of the course and athletes. Strong left skews and narrow distributions, patterns comparable with the fastest European events.

A fast race simply makes the competition at the front tighter. There are 100 slots available meaning double figures for some age groups, but this won’t make them any easier to obtain. Kona slot allocation is determined from the starters on race day, but having refined my calculations I can make a reasonable attempt to estimate the distribution; as the table below shows there will be a lot of 40-44 year old men going to the roll down. I’ve also included my own allocation algorithm that distributes slots a little more closely to the actual athlete distribution.

The real question, of course, is how fast do you need to be to earn one of those slots? I only have one year of data to work with, but we can use the finishing times from last year’s race to gain some insight into the requirements of Kona qualification. By plotting the times of the top 20 in each age group along with the number of slots available, as in the two graphs below, it’s possible to see what would have been required last year to earn those slots.

One year presents a limited picture: I don’t know if those times will prove typical of the event or not, there weren’t even any athletes in the female 18-24 category to make predictions from. But we can see that in the major male age groups qualifying times will be tight, with seconds separating some finishers. If you’re between the ages of 25 and 50 and a man I would be aiming to finish in around 9:20 if I wanted reasonable odds of a Kona slot. As usual in the female field, with fewer slots, it’s harder to make predictions, age group winners are often a step above those that follow.

Ultimately it’ll come down to your training, on race day you do the best you can. This is the clearest indication I can give, if these times seem reasonable, then the odds are on your side. And if you don’t quite make the cut remember the roll down, with lots of slots on offer the potential to roll is high, 37th place in the male 55-59 age group went home with a slot in New Zealand last month.

After the race I’ll fetch the results and see how the second Ironman Melbourne compares to the first. I am expecting it to be quick.

My starting point is the finisher distribution chart above that gives a small indication of race day conditions and quality of the field. In comparison with some of these charts from 2012, Los Cabos does not lie at the fast end of the Ironman spectrum, professional times and the broad spread of overall splits with their mild right skew support this. The distributions themselves seem typical of American Ironman races, just shifted a little further right than we might expect; each discipline a little slower suggesting conditions were a contributory factor.

Looking at the median split times of individual age groups brings the focus more towards the bike. The bike is consistently slower than the Ironman averages of 2012, while the swim is marginally faster and the run varies in its deviation. I’m hesitant to draw strong conclusions from such a broad picture, but whatever the causes Los Cabos seems to be a slower than average race.

The final area I’ll look at today is the front of the pack, the top 20 in each age group, and to make the data more useful I’ve compared this years finishing times for age group placings with the averages for all Ironman races in 2012. Again the pattern, at least among the men, places Los Cabos somewhere between the average and the slowest times for a given placing. As is often the case in the smaller female field the picture is less clear, trending closer to the averages of 2012.

Entering a new race comes with an element of the unknown and so does the analysis of new data. It’s hard to place a race in context when there is no history to examine. Finish times at Los Cabos seem slightly slower than most American races, but not hugely so. With a typical allocation of Kona slots the competition for a place at the World Championship was likely as stiff as ever, front of pack athletes varying less from course to course.

I started playing with ideas earlier in the week, New Zealand was my test bad having recently imported all its Kona qualification data into the Ironman database. I trialled different solutions in Excel and after a morning with the numbers had a method that fairly distributed all slots with a minimum of one, but required tweaking for each race and, more importantly, didn’t match New Zealand’s allocations. It was a solution, just not the solution, and I’d reached the limits of my Excel skills.

The Los Cabos guidelines gives an example that if 8% of the Age Group starters are Female 40-44, then 8% of the slots are allocated to the Female 40-44 category, but if I compared the distribution of slots in New Zealand 2013 to the distribution of athletes in Age Groups, as in the graph above, the match was poor. I wondered if the actual allocation process was, at least in part, manual with race organisers adjusting the allocations by hand to make their numbers work. That, or I was missing something obvious.

Allowing for the fact the figures above include a couple of age group roll downs (male and female 70-74 slots rolled to other age groups) the allocation tends to underrepresent the largest age groups and overrepresent the smallest. I can see some sense in that – it helps prevent the major male age groups taking all the slots. I was still interested to see if I could come up with a better distribution, either closer to the actual numbers or more accurately proportional to the age groups.

I tried several methods – you can skip this paragraph if you’re not interested in the more technical details – all based upon the principle of calculating the number of age groupers per slot and then assigning slots to age groups based on this number. For example, at Ironman New Zealand 2013 there were 1,284 age groupers on the start line and 50 slots giving 25.7 age groupers per slot, so in the male 40-44 age group with 202 members there’s potentially room for 7.86 slots. By assigning slots in rounds, giving one to each age group with space from largest to smallest, I ensured each age group took a fair share. Refinement brought the numbers closer to the actual proportions, but the approach I finally settled on assigned each age group a number of slots proportional to its size and then iteratively adjusted the distribution until the correct total number of slots were allocated. The results are shown below.

My allocation doesn’t match the final slot distribution, but it more closely matches the competitor distribution. I’m no nearer to replicating the methods used by Ironman, but if they are at the organiser’s discretion I may never be. What I have is a means to estimate future distributions, such as Ironman Melbourne below, on the basis of athlete numbers.

We’ll see how my predictions stand in two week’s time. With more slots to distribute they’re likely to be more accurate, but we can anticipate the largest groups having fewer slots and some of the smallest groups having more. Whatever method Ironman uses their actual distributions don’t seem to match the precise wording of their allocation policy, but I still can’t help thinking I’ve missed something obvious.

Apart from reminding us that luck plays a role in Kona qualification, Ironman New Zealand offers a rare chance to examine the roll down process. Despite the large number of age groupers who would be interested, for most events it is remarkably difficult, often impossible, to find a list of Kona qualifiers. Ironman New Zealand, however, offers records back to 2003 which can be combined with the results data to build a more accurate picture of the qualification process. Naturally I couldn’t resist.

My starting point was to adapt the chart of average finishing times for age group placings to show each individual race and indicate whether the finisher accepted a Kona slot or not. In the densely packed male 30-34 year old results shown above it can be a little difficult to read, but it does show that it’s not uncommon for slots to roll from the front of the race. It is however unusual for them to roll far; outside of the top 10 Kona is an increasingly rare thing. It should be remembered that the number of slots available in New Zealand has changed over the years – older races had more.

Slot availability also changes in proportion to the size of the age group, as one of the larger divisions male 30-34 will typically have a lot of slots. For contrast here is the same graph for the female 25-29 year old age category.

Roll down happens, but with at most two slots available it rarely gets far. As with all the women’s age groups you need to place first or second to be in with a chance of going to Kona.

Much as I like the graphs above they are relatively time-consuming to produce and too detailed to appreciate the broad picture. Instead I need to look at what happens on average (or in the next case on median) to see how many slots there are for each age group and how far they roll. The chart below attempts to summarise this information.

It gives a fair indication of where there are the most slots and the most roll downs, unsurprisingly the two go hand-in-hand: the more slots available the more chance there is they will roll. If there are a lot of slots in your age group it looks to be worth going to the roll down, just in case.

The problem with the above is two fold: firstly it doesn’t account for the variation in slot numbers available and secondly it doesn’t account for the variation in age group size (and hence competition for slots). To factor for this I calculated the percentage of the slots each age group received and then, assuming (it’s a big assumption) that the roll down is proportional to the number of slots, I adjusted those numbers accordingly. Finally I converted these figures into percentages of the age groups to give a relative indication of where you need to finish to gain a slot or be in with a chance of a roll down.

Complicated, but the chart above shows that for the main age groups you need to finish within a similar percentage of your age group field to be in with a chance of a guaranteed slot and that, generally, the roll down allows a small margin on top of this. Long roll downs, as happened this year, are too rare to significantly increase the odds outside of the guaranteed slots. Looking at past finish times for top age group placings isn’t a bad guide to qualification, you will need to be in the top 10%.

Of course this is evidence from one race and there are many factors that influence slot acceptance: race date, athlete location or finances are obvious examples. Expanding it beyond this scope requires more data. At least in New Zealand we can see most people offered a Kona slot take it.