Spine Race - Some Stats (2012-2024) - Updated following 2024 race.

This is an update to a post I originally published in following the 2023 event.  I have now updated it to reflect the 2024 race.

Over the years there have been a number of stats published about the Spine races, usually focusing on finish percentages etc.  I decided to take a look and see whether I could do something different with the data.  I've focused purely on the Winter Spine Race and none of the other races in the series.

Some headline stats:
  • 28 countries are represented in the finish results (increase of 2 as a result of 2024 event).
  • There are 610 recorded finishes by 478 people
  • Only 16 people have gone sub 100 hours (exc. 2015) - an increase of 4 during as a result of 2024 event.
I've attached the spreadsheet I put together.  To explain it in a bit more detail take Jasmin Paris (row 6):
  • She was the 230th recorded finisher
  • The 185th individual to have ever finished the Spine
  • The 5th fastest recorded time (includes people who have multiple finish times)
  • The 5th fastest individual performance (i.e. it removes people who have finished more than once), but includes 2015 (which is discarded for record purposes)
  • The 3rd fastest ever individual performance (when discarding 2015).
Any surname highlighted in Green means it is a multiple finisher, and is their fastest performance.  Any name in Yellow is a multiple finisher, but not their fastest performance.  Anything highlighted in Orange means there is a faster time recorded, but this is their fastest time when excluding 2015.

Access to spreadsheet here:

Some more general stats:

Q1 - means time of the 25th percent finisher, Q2 means 50th percentile (I.e. average), and Q3 means time of the 75th percent finisher.

The analysis shows that the event by and large is getting quicker and quicker compared to the long term average.  This is probably down to two factors :a) the event attracting a greater calibre of competitor, and b) there is a greater understanding of what it takes to finish the event compared to earlier years.

In terms of 2024, at the front end it was the fastest ever year which is not surprising considering the calibre of the entry list.  However, although a course record was set this year, the front 25% of finishers averaged only a little under an hour faster than 2022.  There is a question though about the 2022 results and whether they should be treated in a similar vein to 2015 as a large part of the course was cut due to storm damage in Northumberland.

Next question is the difference in Male and Female results since 2013:

So female performance is marginally slower, but only by very small margins at the front end and on average.  It is only towards the back end of the race in there a more pronounced difference.  What is perhaps the most interesting output is that the times have tightened up considerably compared to the same analysis that I undertook last year.  Looking at 2024 in isolation the average female finish time was 130:00, whilst the male equivalent is 130:49.  There has long been a theory that the longer the ultramarathon, there becomes a tipping point where female physiology is more advantageous.  There is definitely a hint of this here, but we need to be cautious as there are relatively few female finishers (16 out of 91) so the averages can be skewed by having fewer numbers.  There is also a question about whether the distribution of ability is similar to males to make a meaningful comparison valid - perhaps some females may be put off from entering for various reasons.  Regardless, it is clear that overall female performance is at or near to the same level as males.

Next up is the relative distribution of performances by year.  The dotted line below is the average distribution of times (excluding 2015).  I have excluded 2012-2014 due to there being so few finishes in the infancy of the event (albeit these years are included in the overall).  What is clear though is that on the whole performances have generally improved considerably since the early years which makes sense as more people enter, attracting of elite athletes, better understanding of strategy etc.

2022 seems to have been the best year overall for performance (grey line), but I suspect this is due to a significant chunk of the course being cut, so there is actually an argument to exclude the times from the record books.  Otherwise 2024 has a strong claim to be the fastest on record.

Finally, a bit of fun analysis new for 2024.  I took a look at whether there is any difference in performance between GB / IRL performance against the Rest of the World:

Contrary to what I was expecting (GBR & IRL performances to be better due to knowledge of conditions), the exact opposite is true regardless of whether you are a front end or back end finisher.  This may in part be explained by differing completion rates.  Whilst I haven't looked at all years, I did calculate the finish rate for 2024.  GB / IRL finish rate was 58%, whilst for the Rest of the World it was 45%.  The conclusion therefore is that GB / IRL competitors are more likely to finish,  but RoW athletes are more likely to finish faster.

What I think this means is that GB / IRL are more likely to finish by having a better understanding of conditions in the UK, but those from the Rest of the World may be of a slightly better calibre, perhaps attracting more serious ultra runners who have undertaken a variety of events prior to entering the Spine Race?  In contrast the UK contingent have a wider range of experience, and for many of whom the primary goal is simply to finish?


I have stayed away from looking at finish rates each year as there has been plenty of analysis in this regard undertaken by others.  What I haven't seen is an analysis on where retirements happen?  All things being equal, you would expect more retirements in the second or third quarter of the race as fatigue hits, accumulated injuries, getting timed out etc. all becomes a risk.  I would then perhaps expect fewer retirements in the last parts of the race as the finish line is almost in sight.

The graph below suggest a different picture...

The above does not mean all the retirements necessarily happened at Hebden Bridge, but that either people were successful in reaching the checkpoint and retired there or continued but were unsuccessful in reaching the following timing point.

What is clear though is that there is a significantly higher attrition rate in the earlier phases of the race - if you are going to retire there is a 41% chance that either people retire at or before CP1 or were unsuccessful in reaching Malham.  If you include checkpoint 2 at Hawes 66% of all retirements happen by then.  Or looking at it another way, 66% of retirements happen in the first 108 miles, but only 34% of retirements happen in the following 156 miles.  Statistically, once past Hawes the overwhelming odds is that you reach the finish.  Again, to reaffirm the graph does not mean that 40% of people will retire between the Start and CP1, but it does mean that if a retirement does happen, then there is a 40% chance it will happen at that location.

In terms of 2024, the analysis shows that it was more or less an average year in terms of the locations of where the retirements occurred.  There was a slightly higher percentage of drop outs at Torside, but offset by slightly lower numbers than usual at Hebden Bridge.  After that there are differences here and there but nothing that I would deem significant.  It is interesting to note the weather that hit a number of competitors between CP3 and CP4 didn't have any material effect on the location of drop outs.  Perhaps putting a hold on the race whilst the worst of the weather passed negated any potential impact it would have had.

Whilst I tried to sanitise the data and remove obvious errors in the published results, there are bound to be one or two issues.  For instance one common name I've kept as separate people in different years results, as I can't say for certain they are one and the same person.  Also there are quite a few people who clearly finished at the same time, but have been given slightly different times by a few seconds.  Anything within a minute I tended to group together, but there are bound to be errors.  However, it is going to be there or thereabouts.


  1. Very interesting, I love the stats! Do you know what happened in 2015 to make it such a fast year (course diversion or course cut?)

    1. I believe the race was paused for 24 hours due to a storm and bits of the course were diverted.

    2. An interesting read. One major fact you have not mentioned, possibly because you are unaware of it, there was no gpx route file available to runners until 2014. With a gpx file downloaded onto a GPS runners are able to travel faster making less route errors. 2012 and 2013 relied heavily on map and compass skills. In recent years runners can travel a lot quicker as GPS units improve with route notifications etc.

    3. Thanks for your comment. It certainly will be a factor. My gut feel is that the early races dont have the accumulated benefit of learning from earlier editions and as race craft knowledge improves so does race times. Its also noticeable that recent editions have benefited from 'bigger' names being attracted now that the race is well established. There are then other factors that vary one year to the next such as weather and route variations. Certainly having a GPX file loaded onto a watch removes a lot of jeopardy- as a traditionalist I think its a shame as map reading should be part of the skill set needed to finish.


Post a Comment

Popular posts from this blog

Ranger Ultras - High Peak 100k

Ranger Ultras Pennine Bridleway 270km