Archive | Analysis

Rest Duration and Race Performance – a closer look at IM Hawaii

This post continues the analysis of the data from the recent Ironman Hawaii race. One of the consequences of the new KPR qualifying scheme was that every Pro who wants to race Kona has to race at least one Ironman in the year before Kona. Some athletes were even forced to race more often than they might have liked in order to get enough KPR points to be ranked in the TOP 50 (males) or TOP 30 (female). One of the concerns this raises is if this is going to have an impact on the Kona race. Or in other words, the KPR system probably did a good job of picking “the best athletes” for Kona, but did the qualified athletes race in top form or were they still tired from recent races?

What I’m analyzing is the impact of the “rest duration” to race results. “Rest duration” is the time difference in months between Kona and the race the athlete qualified for Kona (i.e. the the last Ironman race the athlete did). One caveat: The athlete may have raced other, shorter races, for example Olympic distance or even 70.3 races such as the 70.3 world championships in Las Vegas four weeks before Kona. However, these races probably didn’t add significant fatigue for the Kona racers – most of them focused on Kona and probably raced accordingly.

Data: Impact of Rest Duration on Finishing

Here is a table that looks at the number of athletes who finished or DNF’d broken down by Rest Duration:

Rest Duration
in Months
DNFs Finishers DNF Rate
11   7 0%
7   1 0%
6   3 0%
5 4 10 29%
4 4 2 67%
3 12 23 34%
2 3 7 30%
1   1 0%
All 23 54 30%

There are a few differences between DNFs and Finisher, but they are not very easy to spot:

  • All Athletes with a large Rest Duration (6 months and longer) have finished the race.
    My interpretation is that these athletes didn’t carry any fatigue from their IM any more. Some of them had injuries (Pete Jacobs comes to mind), but they were able to take care of them before starting the race. If they had some serious issues that did not resolve in time, then they didn’t even attempt to start (e.g. Terenzo Bozzone).
  • On average, finishers have one more month of rest than athletes that DNF’d.
    Largely as a consequence of the first point, Finishers have an average Rest Duration of 4.4 months whereas DNFs have only 3.4 months.
  • For shorter Rest Durations, the DNF rate is pretty stable around 30%.
    The exception is 4 months, but the number of athletes is probably too small for it to be significant. If there is any difference at all, slightly longer Rest Periods (3 and 4 months) have a higher DNF rate than really short ones (2 months and shorter). I would think that athletes with short rest periods race a bit more cautiously.

When trying to decide when to qualify (or to validate a Kona spot), this could mean to race as early as possible, but the difference is probably too small to base a decision on this data.

Data: Impact of Rest Duration on Race Performance

There is some more interesting information when including the race performance into the analysis. I did this by comparing the “expected race time” (taking the athletes rating prior to Kona and conditions on race day into account) to the actual race time. This data is shown in the following graph:


On the x axis, it shows the Rest Duration in months (shorter rest before the race to the right), on the y axis it shows the difference between the expected and actual performance of the athlete (in minutes, positive values mean that the athlete was quicker than expected). The thin blue line shows the actual data, the thicker orange line an approximation.

The data indicates that athletes with short Rest Durations (3 months and less) are not racing all that well (the data point for one month is just one athlete, so the graph may not even rise but continue to drop further). If an athlete wants to race well in Kona, their Rest Durations should be longer than three months. The data is not totally conclusive on what the “best” Rest Duration is (actual data pointing to 7 or 4 months, but the approximation indicates some value in the middle of this range). The data from this year does not show any positive or negative impact of really long rest periods (11 months).

Conclusions: What does this mean for pros and WTC?

For an athlete that wants to perform well in Kona, this means that they should seriously consider skipping the big July races (IM Austria, IM Switzerland, IM Germany, IM Lake Placid) and should avoid the August races (most notably, IM Canada). Interestingly, WTC has scheduled their new races exactly in the “not so good for Kona” timeframe: IM New York and IM Quebec have both race dates in August.

Instead of these summer races, athletes should try to qualify in earlier races, say IM New Zealand (March), IM South Africa (April), IM St. George, IM Texas or IM Lanzarote (May), maybe as late as  IM Coeur d’Arlene (June).

This analysis also  raises questions about WTC’s tries of establishing regional “Championship races” without hurting Kona. Other than the new race in Australia (IM Melbourne in March), these races are in July (IM Germany) and August (IM New York). It almost seems that pro athletes have to choose between Kona and one of the regional championships as it seems very tricky to race well in the regional championship and in Kona. This is certainly not in the interest of WTC and the people organizing the regional championship races. However, WTC may not have too much of a choice in this matter: An ideal date for a great Kona race (say May) is too early for the bulk of the athletes starting in these races that are not focused on qualifying for Kona.

One thing that seems obvious to me: WTC should move the cutoff dates for Kona qualifying by at least one month. Currently, these cutoffs are at the end of July (40 males, 25 females) and end of August (10 additional males, 5 additional females). In order not to devalue the July races too much, they should decide all spots at the end of July (or change the relations between the first second batches to 25-15 for May qualifiers and the same number for July qualifiers). Races after these dates would then count for the following year’s Kona race (so points from IM Canada 2012 should count for Kona 2013). This avoids some hectic racing in August: People that had to race in August in order to qualify, didn’t do too well in Kona – they were just too tired to race really well (think Mary-Beth Ellis).

I’m not sure if WTC is considering changes to the KPR. As far as I know, the system remains pretty much unchanged for 2012, but maybe we’ll see some adjustments for 2013. Other than some general grumbling about the KPR, I haven’t seen many specific suggestions how the KPR could be made better. Hopefully, WTC is going to solicit feedback on improving the KPR.

How many people have completed 12 IMs and will now be able to get a slot for Ironman Hawaii? (Part 2)

My last post triggered some discussion on the TriTalk forum’s thread on the WTC Announcement.

A few people came up with other ideas for “guesstimating” the number of eligible athletes. One example was user Stengun who wrote the following:

I recently attended the Ironman Lanzarote "Special Achievement" Ceremony. For those of you who don’t know: They give a special award/medal for anyone that’s completed 5 Lanzas or more. I did go this year, and felt quite excited by it. However I left feeling a bit average and not very special by end. This was because of the numbers of people receiving this award. I assumed it would be me and hand full of others. But it was not. There must have been 40+ people all receiving this, and not just people with 5 Lanza finishes, there where plenty with 10+ and the one guy had 18! Remember this is just Lanza finishes. So almost everyone would most likely have other finishes at other event (as do I). I understand some events like Lanza have a very loyal following and that could skew the numbers a bit. But I’d estimate, there’s an equally loyal following at the other big and long standing events.

With similar qualifications as in my first post, this is certainly something I can look into! Here then is a list of the long standing Ironmen and the number of athletes who have raced in all six years that I have age group data for:

    • IM Canada: 50 athletes
    • IM Coeur d’Alene: 17 athletes
    • IM Lake Placid: 28 athletes
    • IM Wisconsin: 26 athletes
    • IM Lanzarote: 28 athletes (This number seems to be consistent with Stengun’s observation.)
    • IM France: 8 athletes
    • IM UK: 23 athletes
    • IM Switzerland: 4 athletes
    • IM Austria: 17 athletes
    • IM Western Australia: 23 athletes

This is a total of 224 athletes from 10 races. Trying to remove Kona qualifiers (again, using the crude 10-Hour-barrier as outlined in the first post) reduces this number to 207.

Then there are at least 6 more “long-standing” IMs (Arizona, Florida, New Zealand, South Africa, Australia, Germany, maybe Louisville) – which would give a total number of 331 athletes that have consistently race their “home race” in the last six years without having raced Kona.

It is a bit hard to estimate how many of these have completed 12 or more races overall. I’m guessing that not all of these would have raced the required number of “other” races (either at home or in another race), but that this number can be balanced by those athletes who have missed a year here or there but have done other races a few more times.

So this would give a slightly higher number than when estimating by “at least 8 races”. (That number was 273 which is in the same ballpark.)  If I had to give a number, I’d put it at around 300. But based on all this analysis, I’m very certain that the number can’t be close to “a thousand” that people have been throwing around.

How many people have completed 12 IMs and will now be able to get a slot for Ironman Hawaii?

Changes to the lottery system

The guys at IMTalk scored a real scoop this week. The day after the Kona race they interviewed Andrew Messick, the new CEO of WTC, the company who owns the Ironman brand and runs most of the Ironman races. In the interview (listen to the full interview here, the announcement regarding the lottery starts at 22:50) he announced that

“We’re changing how our lottery works. For athletes who have done 12 or more Ironman over their career – and who are still racing – we are going to guarantee that they have a chance to race Kona.

Andrew also announced that when you’ve entered the lottery and get not picked, you will get an extra chance when you enter the next lottery. In my opinion, these are very smart moves on their part, again encouraging people to continue to participate in Ironman races (or the lottery) and choosing one of their races instead of a Challenge or Rev3 race. (On a more personal note, this may give me a chance to get to Kona, but I would still have to finish 10 more IMs.)

Analyzing the data

This change in the lottery system prompted Thomas Peoples to send me an email with the following issue:

I’m interested to see how many people have completed more than 12 IM races for this new lottery system.

I don’t really have all the required data to do a proper analysis:

  • As I focused on PROs, I don’t have the agegroup results for all races (for about probably about 1/3 of races I only have PRO results).
  • My results only go back to 2005.
  • It is tricky to properly match results from different races, for example there are differences in spelling or the handling of special characters (technically “synonyms”). (Are Andi Boecherer and Andreas BÖCHERER the same athletes?) Especially on the female side, athletes change their names after marrying (two notable examples are Bella Comerford/Bayliss and Marilyn McDonald/MacDonald).
  • One other issue is “homonyms” – different athletes with the same name. For example, there are probably more than one “Peter Brown” or “David Smith”.

But even with these caveats, I did some analysis. I have 55 athletes that with my limited data have 12 or more races. When I look for athletes that have 8 or more races (as I only have age group data for 2/3 of the races), I get 339 athletes.

This number has to be reduced by subtracting the athletes that have already participated in Kona. As I currently don’t have age group results for Kona, I have resorted to looking at the finishing time for the athletes and remove the “fast” athletes. If I define “fast” as having at least one result of under 10 hours (probably allowing them to be in the mix for a Kona slot) and remove all of these athletes, I’m still coming up with 273 athletes.

With all the uncertainties mentioned, my guess is that the number of athletes eligible for an automatic slot is about 250.

How is WTC going to handle this?

It will be interesting to see how WTC is going to handle these issues. At least based on their public facing data, they do not have a “unified customer view” across all their different races. (Although it would be quite valuable information for them – how many “repeat customers” do they have?) Therefore, I assume that when you try to claim your “12 IM slot” you will have to submit a list of your IMs which they would manually cross-check against results in paper, HTML, PDF or maybe even in a database. They would also need to have a way to make sure that these athletes have not been on the IM Hawaii start list (or finished?). One can only hope that WTC has the necessary data available and that they diligently check the claims before assigning a “12 IM slot”. I would be very interested in a “look behind the scenes” on how they plan to handle this issue.

Kona TOP3s who didn’t qualify in an Ironman (outside of Hawaii)

With the KPR, every pro that wants to race Kona has to complete at least one Ironman in the “qualifying year” outside of Kona. Now that I have a good number of race results in my database, I can have a look at recent TOP3s in Kona that would not have been able to race under the new qualifying rules. (Even so, my data may not be 100% up to date or complete – please let me know any errors in my post.)

Here’s an overview:

  • winners: Michellie Jones (2006), Chris McCormack (2007), Craig Alexander (2008 and 2009), Mirinda Carfrae (2010)
  • 2nd places: Sam McGlone (2007), Yvonne van Vlerken (2008), Mirinda Carfrae (2009)
  • 3rd places: Kate Major (2007), Virginia Berasategui (2009), Julie Dibens (2010)

This is one winner from each year and 11 out 30 possible (5 years * 3 spots * m/f) places! To me, this was a bit of a surprise, I thought that it was “just” Crowie who only raced Kona.

Where did they get their slots?

  • The TOP10 in one year used to get an automatic spot for the following year.
  • For some time, the 70.3 champions got an automatic qualifier for Kona.
  • Even though the number was pretty small, some 70.3 races also had some pro slots.

Does this large number of athletes who didn’t race an IM during the year (or none at all) mean that WTC should change the qualifying rules? For now, I don’t really think so:

  • With the exception of Macca, all 2010 TOP10 athletes qualified under the new rules. This shows that the new rules have been accepted, even if a bit grudgingly.
  • Giving any number of TOPx athletes from Kona slots for the following year would be a bit unfair compared to those that have to race to qualify: They have a full year to focus on the next Kona race. While this may be a disadvantage, it usually allows them to get to Kona a bit fresher than those that have to race in the summer to get a slot.
  • Especially on the women’s side, the 70.3 champions fared quite well in Kona (even if it meant that it was their first IM). However, the way the points system works now, all the 70.3 champion has to do is to race an Ironman – this should give them enough points to qualify. Also, I think it is fairly reasonable to expect pros not to race Kona as their first Ironman.

So even considering the large number of athletes with good Kona results  that wouldn’t have been able to race under the new KPR rules, I don’t see any need for major changes based on the analysis of the data. From a WTC viewpoint, you could even say that the KPR succeeded in drawing athletes to race more often. However, as indicated earlier, I suggest to wait for the Kona race before a final verdict on the KPR.

A question from Chrissie Wellington: Do women have to race more often for a Kona slot?

Chrissie Wellington has posted a long and as usual thoughtful blog post on “Thoughts, comments and suggestions on the KPR”. One of her points is as follows:

[It] seems that, because the women are competing for 30 total slots, they are having to race more, than the men who are competing for 50. Those at the cusp of the men’s cut off have done far less IM racing than their women counter parts.  In short, some data suggests that the women are racing more to accrue sufficient points to get to Kona.

I thought that with my database of results, I should be in a pretty good position to have a closer look at this question.

All qualifiers

First, lets have a look at the number of races that the Kona qualifiers have raced:


This graph shows the relative frequency of the number of races that female (red) and male (blue) athletes have raced. The shape of the graphs are very similar, and the average number is also pretty close (2.66 for the male, 2.82 for the female. To me, there is no significant difference between the genders.

There are no changes when we limit the number of males to 35 (as with women), either the top 35 or the low 35 qualifiers.

Athletes close to the cutoff

But Chrissie didn’t seem to look at all athletes. So I looked some more and limited the athletes to the 10 above and below the cutoff. With only 20 items in each dataset, the graph looks a bit wonky, but the base shape is still pretty similar:


Here, the male average is 2.25, and the female at 2.85 races – the difference between the genders is a bit larger than for all qualifiers.

So, did females race more often?

Even though Chrissie is a lot closer to the female athletes than I will ever be, her intuition of females racing more often is not supported by my data analysis:

  • Female athletes have a similar average number of races, regardless of weather they qualifies or were close to the cutoff.
  • The larger difference in the averages of athletes close to the cutoff seems to be caused more by the male athletes racing less than the females being “pushed” into racing more often. (Maybe some of those male athletes weren’t really interested in qualifying for Kona and didn’t chase points.)
  • Also, the data gets a bit skewed by prolific female racers close to the cutoff (Miranda Alldritt with 6 races, Hillary Biscay with 5 races) and Mary Beth Ellis having to win three IMs within eight weeks in order qualify.

Then again, I may be missing something, so please leave a comment or send me an email.

Another idea: Change the points scheme for the females

There is however, one thing worth discussing a bit further: Because the size of the female fields are usually a lot smaller (average pro fields are 10 for the females and 18 for males), it is much easier for the females to score quite a lot of points even when they are far behind from the top racers. (I.e. it is much easier for a female to finish in the top 10 than for a male.) The way I see this, it results in the women racking up more points than their male counterparts (case in point: Male cutoff at around 3.000 points, female cutoff at 4.600; male counterpart to the last female qualifier “only” needs 4.200 points).

Chrissie discusses some changes to the points scheme in her blog post as well (although I’m not sure her thoughts were limited to the females). Maybe WTC should have a closer look into this?

Powered by WordPress. Designed by Woo Themes

By continuing to use the site, you agree to the use of cookies. more information

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.