Analyzing IM South Africa 2011

Chrissie Wellington had another amazing race in South Africa – this is what most triathlon new sites are reporting. Here’s a closer look at the statistical outcome of this race.

Race Adjustment

IM South Africa is usually a medium quick race (course handicap around 10 minutes), but in the last few years has been a bit faster – not sure what has caused this, but there might have been some course changes before the 2008 race. Compared to these years, the 2011 race was typical with an adjustment of 17:27. (To recap, the adjustment is the average difference between the participant’s ratings and their actual race time.) Conditions have been reported to be quite nice, but maybe a bit hot during the run. This may have contributed to quite a few big names dropping out, for example the pre-race favorite Marino Vanhoenacker, Luke Bell, Rutger Beke, Jason Shortis, Michael Göhner, Bert Jammaer and Natascha Badmann. Tom Lowe didn’t even start the race, but I haven’t heard anything of why that was the case.

Female Results

As noted above, the big result this weekend was Chrissie Wellington’s quick time, so I’ll start with the females:

Rank Name Nation Actual Time Expected Time
1 Chrissie Wellington GBR 08:33:56 08:40:32
2 Rachel Joyce GBR 09:08:23 09:23:52
3 Diana Riesler GER 09:20:37 n/a
4 Silvia Felt GER 09:24:31 n/a
5 Amanda Stevens USA 09:25:50 n/a
6 Belinda Granger AUS 09:28:59 09:28:29
7 Simone Benz SWI 09:31:21 10:55:35
8 Martina Dogana ITA 09:47:47 09:40:28
9 Yvette Grice GBR 09:59:28 10:14:56
10 Eva Dollinger AUT 10:00:24 09:34:34
11 Claire Horner SAF 10:11:55 10:47:25
12 Helen Buley SAF 10:26:29 10:17:14
13 Caroline Koll SAF 10:37:39 10:17:59

Expected time is based on athlete’s rating (prior to race) and race conditions.

Chrissie posted a quicker time than her 8:36 at IM Arizona time from last November. (I find it a bit silly to call it an “IM world record” knowing that she posted an 8:19 in Roth 2010.)  Looking at the statistics, the TOP4 best results ever are from Chrissie – and all on different courses!

Rank Name Adjusted Result Actual Result Race
1 Wellington, Chrissie 08:40:26 08:19:13 Challenge Roth on 2010-07-18
2 Wellington, Chrissie 08:48:59 08:54:02 IM Hawaii on 2009-10-10
3 Wellington, Chrissie 08:50:32 08:33:56 IM South Africa on 2011-04-10
4 Wellington, Chrissie 09:00:42 08:36:13 IM Arizona on 2010-11-21

Simply amazing – for me that’s about as impressive as the fact that Chrissie posted the overall best run split. The sky seems to be the limit for her.

Behind Chrissie, Rachel Joyce had another great race, but her excellent 9:03 – twenty minutes quicker than expected – pales in comparison with Chrissie’s performance. With Natascha not finishing, I predicted Belinda Granger to round out the TOP3, but even with her solid 9:28 (almost exactly as predicted) it was a trio of three unrated pros coming in before her and Diana Riesler getting the third spot.

Male Results

On the male side, there was quite a bit back and forth at the top of the race, and finally home favorite Raynard Tissink was able to defend his title from last year. All the athletes that were much better rated than him dropped out (most notably Marino Vanhoenacker who was right in the mix), but it was a hard-fought win and he had to post a new course record of 8:05, more than 20 minutes quicker than expected. A really great performance from such an established athlete!

Rank Name Nation Actual Time Expected Time
1 Raynard Tissink SAF 08:05:36 08:28:38
2 Andreas Bocherer GER 08:08:36 09:12:29
3 James Cunnama ZAF 08:13:18 08:47:47
4 Daniel Fontana ITA 08:18:51 08:33:00
5 Cyril Viennot FRA 08:30:42 n/a
6 Eduardo Sturla ARG 08:32:52 08:27:20
7 Uwe Widmann GER 08:33:39 08:30:42
8 Petr Vabrousek CZE 08:40:36 08:46:06
9 Trevor Delsaut FRA 08:41:57 n/a
10 Balazs Csoke HUN 08:43:06 n/a
11 Dominik Berger AUT 08:52:24 n/a
12 Kent Horner SAF 08:58:39 09:18:39
13 Toby Radcliffe GBR 09:01:17 09:17:13
14 Mike Schifferle SWI 09:04:02 09:07:13
15 Sergio Marques PRT 09:09:30 08:50:00
16 Craig Twigg GBR 09:10:01 n/a
17 Gilad Rotem ISR 09:12:06 n/a
18 Erik-Simon Strijk NDL 09:16:32 n/a
19 Bryn Freeman SAF 09:23:29 09:27:21
20 Erich Felbabel FRA 09:23:36 n/a
21 Marcel Bischof GER 09:32:47 n/a
22 Heinrich Sickl AUT 10:08:14 09:28:07
23 Freddy Lampret SAF 10:46:11 09:38:47

Expected time is based on athlete’s rating (prior to race) and race conditions. 

After Raynard, my young countryman Andy Böcherer had a great result and came in second – more than one hour quicker than expected! (Then again, I just had one race in my database for him, probably one of his slower results.) Another South African, James Cunnama, managed to snag the last spot on the podium, he was also more than half an hour quicker than expected. A lot of people had told me that Cunnama was underrated before the race, so I’ll have a look at my methodology before my next round of predictions for IM Australia …

3 Responses to Analyzing IM South Africa 2011

  1. Chris April 12, 2011 at 3:07 am #

    Interesting attempt at modeling. Seems like you’re pretty good at predicting racers with HUGE numbers of races under their belt (maybe Tissink excepted), but it just seems like the sample size of IronMan races for each individual, plus the dynamic factors and unique situational reality of each race (who’s in the field, what the weather is like, etc.) would make any more accurate modeling difficult.

    Good attempt though, and I wish you the best in tweaking your formula, and will pay close attention to future results!

  2. Rob April 12, 2011 at 12:51 pm #

    Not too bad, but I see a few oversights. Sorry if I sound a bit detailed, but just my statistical mind.

    One oversight would be the lack of a historic weighting adjustment. There should be an adjustment for current form. Looking at someone like Raynard Tissink, he came 5th in the World Champs 6 months ago, yet he gets a poor time prediction affected by 2 lesser performances at Roth and Florida a few years back. And Natascha Badmann hasn’t raced in 3 years, but you predicted her to come 2nd just behind Chrissie?

    Furthermore, what about people who drop out or don’t finish. Their time is not included…? Maybe a heavy penalty for a DNF – and make this adjusted by how far they were and position in the race? This could be added to the time that they were currently situated? ie: 3rd place on the run with 5km and you drop out, doesn’t hurt your time as much as 3rd place at halfway on the bike….

    And presentation… how about a column or two showing what the difference is/was adjustment factor etc… red for under/blue for over it will show at a glance who over or underachieved (relative to your forecast).

    Overall, quite a nice tool…. a long way to go before it gets useful, but hopefully it will get there. It will be fun to further develop it to include the various legs of a race, and then for age groupers to put there times in… would make a very useful training/pacing aid!!

    • Thorsten April 19, 2011 at 9:17 pm #

      Rob, thanks for the suggestions .. I’ll try to include them as soon as I can.

      Let me try to address some of your points:

      • historic weighting adjustment – it’s in there with a factor of 0.85 .. maybe I have to use a smaller constant (and I’ll play around with different factors), but I wanted to be results quite “stable” and not fluctuate too much by one good (as with Tissink) or bad performance
      • “aging” of ratings when not racing – I’ve got some ideas in mind, and will include them in my next major revision of the rating algorithm
      • DNFs – I’d like to include these, but I’m not sure in which way. I’m sure that no one DNFs in order not to have abad result included in my ratings, so there is no “skewing” by not including DNFs.

      Thanks again, feedback like yours helps me make improvements to how the ratings are calculated.

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.

Close