TTR – Adding female Pros and IM Florida

I’ve put some more work into TTR – Thorsten’s Triathlon Rating. With IM Florida I added a third course (and six more races from 2005 to 2010) and I also managed to include female pros.

The amount of data is getting quite impressive by now:

  • 3 courses (IM Hawaii, IM Arizona and IM Florida)
  • 19 races (6 times each for Hawaii and Florida, 7 races for Arizona)
  • 632 athletes (412 male and 220 female)
  • 1,513 individual results

One thing that I have to think of now is to make sure that I’m working on high quality data – I was able to catch a few obvious problems, but I have to do some more systematic work on that.

An interesting little tidbit: The athlete with the most results in my database is Petr Vabrousek with 14 results! He’s also way ahead of the next athletes, who “only” have 9 races (Christophe Bastie, Michael Lovato, Heather Gollnick and Tamara Kozulina).

Adding IM Florida and new male TOP10

Adding IM Florida did not require any changes in the methodology. So the changes in the TOP10 are a reflection of the Florida results:

  • As was to be expected, IM Florida is another fast course, in fact it is even faster than Arizona (course rating: 25:53, compared to 20:38 for Arizona and 3:57 for Hawaii)
  • Dirk Bockel moved a few spots down with his 8:21 in Florida 2010 (which was adjusted to an 8:44).
  • Raynard Tissink fell out of the top 10 – his two great Kona results were joined by a not-too-stellar 8:40 in Florida 2009. He was replaced by Rasmus Henning.

Here is the new (male) TOP10 list:

Rank Name Rating # Races
1 McCormack, Chris 08:20:57 5
2 Alexander, Craig 08:20:58 4
3 Raelert, Andreas 08:24:05 3
4 Llanos, Eneko 08:29:15 5
5 Al-Sultan, Faris 08:29:40 6
6 Bockel, Dirk 08:34:04 3
7 Bracht, Timo 08:34:58 6
8 Vanhoenacker, Marino 08:35:47 5
9 Henning, Rasmus 08:35:58 3
10 Potts, Andy 08:36:37 3

Macca is still in the #1 spot, but it’s getting very close between him and Crowie.

Adding female Pros: Women’s TOP10

The main challenge this time was adding the females into the mix. I wanted the male results to be pretty stable and not change much when the women were added into the mix. It was clear that there would be some small changes in the course ratings and adjustments for the race, but I wanted these to be pretty small. In order to achieve this, I had to play a bit around with the calculations until I settled on a slightly changed algorithm. (The change is a bit technical: The adjustments are calculated as a percentage of race time instead of absolute change in seconds. In order to show a course rating as a time, that percentage has to be changed back to a time by applying it to a 9 hour race.)

With this new results and updated algorithm, here are the women’s TOP10:

Rank Name Rating Last Race # Races Total Rank
1 Wellington, Chrissie 09:01:40 IM Arizona on 2010-11-21 4 54
2 Carfrae, Mirinda 09:07:51 IM Hawaii on 2010-10-09 2 67
3 Van Vlerken, Yvonne 09:28:09 IM Hawaii on 2010-10-09 2 114
4 Jones, Michellie 09:29:42 IM Arizona on 2008-04-13 4 119
5 Steffen, Caroline 09:32:19 IM Hawaii on 2010-10-09 2 133
6 Berasategui, Virginia 09:33:21 IM Hawaii on 2010-10-09 4 136
7 McGlone, Samantha 09:34:09 IM Hawaii on 2010-10-09 4 141
8 Thuerig, Karin 09:35:33 IM Hawaii on 2010-10-09 2 146
9 Preston, Rebecca 09:35:55 IM Hawaii on 2007-10-13 2 148
10 Lawn, Joanna 09:38:58 IM Hawaii on 2010-10-09 6 153

The first two spots are not a surprise – Chrissie and Rinnie are currently the top athletes. What is interesting is the large gap to the next spots – more than 20 minutes! For me, Michellie in #4 was a bit of a surprise – her last race was in 2008 and to me I did not really think of her as an active athlete that might show up in the ratings.

Another interesting thing is the Total Rank that shows how the females would rank in all athletes (i.e. male and female). Chrissie, for example would be ranked #54 – which means that there are 53 males that have a better rating than she has. Even if some of her great results from Roth are not in the database yet, it’s hard to see her crack the Top10 in Hawaii or the male ratings.

Adding a second course: IM Arizona

In the last week, I’ve worked to include a second course into my triathlon ratings. As I want to include races as the year progresses, I wanted to work with races that have been between IM Hawaii in October and today. As far as I know, these races were IM Arizona, IM Florida, IM Cozumel and IM Western Australia. Again, Craig Harris helped by providing race results for Arizona and Florida. For no particular reason, I chose to start with IM Arizona.

Course Ratings – Comparing Courses

The methodology I outlined previously worked well for including the second course. It was very interesting to see the adjustments for IM Arizona:

Year Adjustment # of Athletes Rating
2005 n/a 0 n/a
2006 567 15 567
2007 1052 15 810
2008 1149 14 923
2008 1289 24 1014
2009 1575 21 1126
2010 1335 23 1161

(Note: There were two races of IM Arizona in 2008 as they switched from April to November.)

The adjustment for IM Hawaii hovered around 0, the final course rating is 5. Comparing this to the course rating for
IM Arizona of 1161, it means that IM Arizona is 1156 seconds faster than IM Hawaii (almost 20 minutes – or 19m:16s to be exact).

To put this in another way: Chrissie Wellington’s blazingly fast 8:36:13 in Arizona 2010 is the statistical equivalent of an 8:54 in Kona – a bit faster than Mirinda Carfrae’s winning time of 8:58:36, but still short of her own course record in Kona. Of course, there are a lot of ifs and buts with these comparisons, but I really like to play around with the numbers without taking them too serious.

New Top10 Rated Athletes

Here are the new Top10 Athletes (based on IM Hawaii and IM Arizona):

Rank Name Rating
1 McCormack, Chris 08:17:07
2 Alexander, Craig 08:17:58
3 Raelert, Andreas 08:21:26
4 Bockel, Dirk 08:23:59
5 Tissink, Raynard 08:25:32
6 Llanos, Eneko 08:25:55
7 Al-Sultan, Faris 08:26:08
8 Vanhoenacker, Marino 08:30:22
9 Bracht, Timo 08:31:56
10 Potts, Andy 08:33:01

There are two main changes compared to the last rankings. Timo Bracht has entered the Top10 – this will be analyzed in the next section. The most obvious change, however, is that we have a new number one: Chris McCormack. Andreas Raelert has moved down from 1st to 3rd. This is a “result” of his performance in IM Arizona in 2008. Even though he won his first IM in 8:14, as this was on a fast course the adjusted result is “only” an 8:35. This “bad” result brings down his rating quite a bit. It’ll be interesting to see how his rating will change when I include IM Germany where he had two faster results.

Adjusting the  “Aging Factor”

In order to favor newer results over older results, I include an aging factor. After playing a bit more with different values, I’ve adjusted the factor to 0.85. This results in a few little shifts in the Top10, but also to a new number 9 with Timo Bracht, instead of Cameron Brown (who is now ranked #12).

Here’s a closer look at the results of these two athletes:

Name 2005 (49) 2006 (66) 2007 (261) 2008
2010 (290) AZ 2010 (1335)
Timo 9:02:33 8:30:24 8:37:52   8:28:52 8:21:00 8:07:16
Cam 8:19:36 8:25:22   8:26:17 8:34:10 8:53:41  

(The number under the races is the adjustment for this race, a number >0 meaning the race was faster.)

When I look at these results, then Timo was the faster athlete in 2006, 09 and 10. Cam’s result in 2008 is much better than Timo’s in 2007, and Cam had the better result in 2005. All in all, it seems that the “old” result of 2005 shouldn’t carry so much weight any more and Timo should be ranked in front of Cameron. This was achieved by a relatively small adjustment in the age factor.

It’ll be interesting to see what happens when both athletes home races (IM New Zealand and IM Germany) are added to the ratings – both have had their most impressive results on their home turf.

Thorsten’s Triathlon Rating “TTR” – First Results

I’ve worked on a first version of a triathlon rating system, focused on long-course athletes. For now, the races I have analyzed are limited to the males pros in IM Hawaii races from 2005 to 2010. (A big thank you to Craig Harris, who was making these available as an Excel file to me. Craig has an interesting little stats site.) I plan to extend that to female pros and add more races in the future, but even with this limited data there are some pretty interesting results.

TTR Top10 Rated Athletes

In TTR, a rating is expressed as a time. This time is based on the previous results if the athlete and represents the expected finishing time in an “average” race in “average” conditions.

Here are the Top10 Athletes:

Rank Name Rating
1 Raelert, Andreas 08:16:45
2 McCormack, Chris 08:18:05
3 Alexander, Craig 08:18:33
4 Bockel, Dirk 08:24:56
5 Tissink, Raynard 08:25:25
6 Al-Sultan, Faris 08:26:10
7 Llanos, Eneko 08:26:29
8 Vanhoenacker, Marino 08:30:52
9 Brown, Cameron 08:32:50
10 Potts, Andy 08:33:15

In order to be included in this list the athletes have to have more than one race and the last race cannot be more than four years old. (In fact, all these athletes have raced in IM Hawaii 2010.)

A lot of big names in this list, and certainly no surprises in the first three names. In order to validate the list, let’s have a closer look at the results of Andreas, Macca and Crowie:

Name 2005 2006 2007 2008 2009 2010
Andreas 8:24:32 8:12:17
Macca 8:23:52 8:13:07 8:15:34 8:25:20 8:10:37
Crowie 8:19:04 8:17:45 8:20:21 8:16:53

Between Andreas and Macca, the two races that they went head-to-head in Hawaii are pretty even, probably with a little advantage to Macca. However, Macca has had some slower races in the past, and that leads to a slightly better rating for Andreas. (That advantage should get a bit larger when this summer’s IM Germany is included in the ratings.)  Looking at Macca and Crowie, Macca was able to win two of the three races that both finished. Again, it makes sense that Macca comes a little bit ahead.

Comparing the Years

Here are the numbers comparing the results of the different years:

Year Adjustment
2005 n/a
2006 128
2007 309
2008 -68
2009 -500
2010 298

What the “adjustment” means is that for example 2009 was 500 seconds (or about 8 1/2 minutes) slower (compared to the base number of the 2005 race). If you look at the results of the top 3 rated athletes in the previous section, you can notice that the 2009 times were quite a bit slower than the 2010 races. In fact, on average the 2010 times should be 500+298 seconds = 798 seconds or 13 minutes 18 seconds quicker than in 2009. Macca was even a little bit better (almost 15 minutes, so he consequently placed higher up and won the race), while Crowie’s PR time (but only 3 1/2 minutes faster)  was not good enough for a top 3 finish in 2010.

Another observation is that both of Macca’s wins (in 2007 and 2010) came in “fast” races – probably consistent with his observation that he struggles in hotter conditions.

Averaging these adjustments, you get a “course rating” for IM Hawaii of –33 seconds. This course rating will get more important in the future when I’m going to analyze results on different courses that will help compare the different courses.

TTR Top10 Performances

Another interesting look is at the top 10 adjusted performances in Kona. The adjusted result accounts for “harder” and “easier” races:

Rank Name Adjusted Result Actual Result Race
1 Alexander, Craig 08:12:01 08:20:21 IM Hawaii on 2009-10-10
2 Stadler, Normann 08:14:04 08:11:56 IM Hawaii on 2006-10-21
3 Al-Sultan, Faris 08:14:17 08:14:17 IM Hawaii on 2005-10-15
4 Lieto, Chris 08:14:36 08:22:56 IM Hawaii on 2009-10-10
5 McCormack, Chris 08:15:15 08:13:07 IM Hawaii on 2006-10-21
6 McCormack, Chris 08:15:35 08:10:37 IM Hawaii on 2010-10-09
7 Raelert, Andreas 08:16:12 08:24:32 IM Hawaii on 2009-10-10
8 Alexander, Craig 08:16:37 08:17:45 IM Hawaii on 2008-10-11
9 McCormack, Chris 08:17:00 08:25:20 IM Hawaii on 2009-10-10
10 Raelert, Andreas 08:17:15 08:12:17 IM Hawaii on 2010-10-09

(The absence of any 2007 results looks a bit suspicious to me, I’ll have a closer look at this and may adjust the scoring system accordingly – after all this is pretty much “work in progress” and if you spot any more inconsistencies, please let me know in the comments!)

Interestingly, this time Crowie has the best result (from his win in the “slow” 2009– 8:20:21 minus 500 seconds). What is evident is the “Aussie dominance” of Crowie and Macca – they both have three results in the Top 10 which is not surprising when you look at the success these two athletes have had in Hawaii.

Looking closer at the Macca results, his best performance was not in the two years he has won the race, but in 2006 when he lost to Normann Stadler – a year when he was really happy with his performance and some of his comments rubbed some people the wrong way.


I want to continue to work on TTR and extend it a bit. Here are some pointers on what I have in mind:

  • As noted above, I’ll investigate some of the questions I have with regards to some inconsistencies and oddities and probably adjust the algorithm a bit. I also have to work a bit more on my data base to make sure I can correctly group the results for an athlete.
  • I want to extend the results to include female pros. At this point I’m not sure if I want to just “lump” them into the male results (probably okay for a Chrissie Wellington, but maybe a bit harsh for second or third tier athletes).
  • I want to extend the analysis to more races, both on the IM calendar and for other big IM races (notably the Challenge Series), but that depends on getting more results. I’ll also have to think about getting the results into my database without too much manual overhead.
  • For now I’ve looked at the final times, it would be great to extend this to the individual disciplines and draw some more conclusions from that.
  • It should be possible to do the rating for agegroupers and shorter distances (IM 70.3 and Olympic distance). There would be too many athletes and races to cover – probably a bit too much work for a side project.

Let me know what else you would like to see covered.

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.