How are the rankings calculated?

• ​Our starting point is the best British athlete in 2023 for a given gender and multisport discipline. For the open gender category in triathlon that was Alex Yee, so let's use him as the example.

• 1000 points is defined as the score achieved by a performance at the 2023 GB Number 1's average level during the 2023 season.

• We then ingest the results from a wide variety of different races.

• For each race, our algorithms (a series of fancy maths equations) estimate the time it would have taken Alex Yee to complete the race. We basically make series of very detailed calculations to determine what would likely have happened if Alex had turned up and then had an average performance (by his 2023 standards!).

• We then compare your actual performance in the race to the time estimated for a 2023 Alex Yee. We award your performance a score (almost certainly under 1,000, unless it was so good you would have beaten Alex!)

  • Worked example - you completed the race in question in 1 hour and 10 minutes (4,200 seconds). Let’s imagine we calculated that Alex Yee would have completed the race in question in a time of 50 minutes (3,000 seconds). This means that your score for that race would be 714.28 ((3,000 ÷ 4,200) x 1,000).

• Once you have completed one scoring race in the last 12 months then you will appear in the rankings. If you do more than one scoring race then we take the average of your best two races to determine your ranking.

​

If you are a female triathlete then our point of comparison is Beth Potter rather than Alex Yee. James Teagle (Open) and Emma Pallant-Browne (Female) have been used for duathlon.

​

So I can compare how my performance change over time by looking at my scores?

• Yes, absolutely! The beauty of this approach (which is explained in more depth below) is that it makes it possible to accurately compare performances on different days, in different weather, on different courses, against different opponents etc. The score you are given for each race indirectly takes into account all of those factors. If your scores are getting higher, then you are getting better!

• The only caveat here is that the 2023 GB #1 in Triathlon and Duathlon are not the same. Duathlon scores are relative to the 2023 GB Duathlon #1 who is different to the 2023 GB Triathlon #1. This means that scores in duathlon can only be validly compared with other duathlon scores, not with triathlon scores (and vice versa).

​

What if the GB Number 1 gets better (or worse) over time?

• The fact that the ability of the GB Number 1 may change over time does not matter. Scores in MyTriRanking will forever be relative to the ability of the 2023 GB Number 1(s) during the 2023 season.

• There is nothing statistically special about the 2023 GB Number 1. Their ability at that time was simply used as an arbitrary standard that everyone else can continue be measured relative to. We could have chosen any arbitrary standard, but making scores relative to the 2023 GB Number 1 seemed the neatest solution.

 

How do you know how fast the GB number one would have gone in the race I just did?

• The My Tri Ranking algorithm makes use of athletes we call “benchmarkers”. Benchmarkers are athletes whose ability relative to the GB Number 1 has already been defined.

• Let’s imagine Jonathan Brownlee raced Alex Yee (2023 GB #1) multiple times in 2023, and on average he was 3% slower. If you take a race that Jonathan Brownlee did, but Alex Yee didn’t do, one might logically deduce that if Alex Yee had competed, his time might have been 3% faster.

• But a calculation based on just one benchmarker (Jonathan Brownlee) wouldn’t be very robust. So the algorithm makes use of many benchmarkers. Every benchmarker produces an independent estimate of how fast Alex Yee would have completed the race. If you have a large enough number of independent estimates of a variable, it becomes possible to use statistics to analyse them and derive a more precise estimate of how fast Alex Yee would have completed the race (assuming he performed at the average of his 2023 level).

• When a new result is processed, anyone who already has a ranking in MyTriRanking becomes part of the pool of available benchmarkers. The algorithm decides which are best to use, and simulates the race imagining the GB Number 1 competed.

• Occasionally a race cannot be scored, and you might see a result without a score on your profile. This happens if the algorithm decides that the quality of the available benchmarkers is not sufficient to produce a reliable estimate of the GB Number 1’s time. This is most likely to happen when the field is small, or when few of the athletes participating have participated in other races already scored by My Tri Ranking.

 

What races are considered in the rankings?

• We have tried to include a broad set of sprint and standard distance races from across the triathlon / duathlon pyramid.

• Races covered:

  • Elite International Events

  • Age Group European / World Championship Events

  • GB Age Group Team Qualification Races

  • British Elite Super Series Races

  • Welsh Super Series Races

  • British Universities Championship Races

  • Age Group National Championship Races

  • British Championship Races

  • Mass Participation Events (Blenheim, London Triathlon, JLL Property etc)

  • Local grassroots events (preference given to results published in 5 year age-groups).

 

There is no perfect way to rank multisport. Sports like swimming are easy to rank as time is everything. Triathlon is not one of those sports. Courses, distances and conditions vary wildly, time in the absence of context is uninformative. Ranking triathlon is inherently complex and has required innovative mathematical approaches. This is a first serious attempt to address these challenges in the UK with the help of a computer model.