Tarawera Ultra 2020 — By The Numbers
Earlier this year, my husband and I travelled from the United Kingdom to my home country, New Zealand, in order for him to compete in the 100 mile event at the 2020 Tarawera Ultra Marathon. Here he is at the finish line!
When looking at his results on SportSplits, I found that there was a huge dataset available to engage with (link here). My husband completed the 100 mile Ultra Marathon, I decided to examine all four events — 100 miles, 100km, 50 km and 21 km.
Race Demographics
The 2020 Tarawera Ultra event had a total of 2653 participants who showed up on the day, with a large proportion of participants competing in the 21km and 50km events (29 percent and 41 percent respectively). Overall, 53 percent of the participants were male, and 47 percent were female. However, these percentages did not hold steady across all the races.
As you can see above, the 100 mile and 100 km events tended to be more male-dominated, while the 50km event was about 50/50 and the 21km event actually had a significantly higher percentage of females competing. Many studies have looked at the disparity between men and women when it comes to participation in Ultra Endurance Events, and I think many of the reasons cited in these studies can contribute to explaining the gender gap. I recommend this article as a good reference point — it probably explains things more succinctly than I could!
Next, I examined the ages of participants, both as a whole, and within each race. While 16 and 17 year olds weren’t allowed to compete in the 50 km, 100 km and 100 mile events, they were able to compete in the 21 km event. Because of this, there is a separate 16–29 year old age category within the larger donut chart. Essentially, what this means is 12.81 percent of participants were 16–29, but the 16 and 17 year olds only ran in the 21km race.
I also examined age data by gender. The differences in age between gender were not huge, so I haven’t included this data in this post. But for those interested, females had a very slightly younger average age profile than males.
A particularly interesting point to note about the age of participants who took part in the Tarawera Ultra events is that the age groupings were very similar across the100km and 50km events.
Interestingly, the percentages for 30–39 year olds and 50–59 year olds were also very similar between the 21km and 100 mile races. However, the 21km race attracted a smaller percentage of 40–49 year olds than the other races.
From this data, I also found that:
- 30–39 year olds and 40–49 year olds were by far the most likely to participate in the Tarawera Ultra events, accounting for 30% and 33% respectively.
- 16–29 year olds accounted for around 12% of all the participants, and were slightlymore likely to participate in the shorter races than the longer ones.
- 60–69 year olds seemed to avoid the 100km event. However, this data point can probably just be explained by the relatively low number of participants in this age category and in the longer races.
DNF vs Finishers
Next, I decided to investigate the proportion of people who successfully finished their races. Out of a total of 2653 participants who started the race, 2528 crossed the finish line, while 125 did not. This equates to 95.29 percent and 4.71 percent respectively.
However, this data doesn’t tell the whole story. 22 percent of 100 mile participants and 10 percent of 100 km participants didn’t finish the race. For the 50km and 21km races, these percentages were 0.6 percent and 0.5 percent respectively — a tiny proportion of the people that ran these races.
DNF vs Finishers by Gender
When I looked at the gender split of finisher vs dnf data, I found some more interesting points! While the overall average DNF rate was 4.7 percent, these percentages varied quite widely based on gender — 6 percent of males didn’t finish their race, compared to just 3 percent of females.
The percentages for the 50km and 21 km events weren’t statistically significant — one or two people not finishing could have swung the ratios completely. But the data for the longer events was interesting!
- Females were more likely than males to finish in the 100 mile race. 16.7% of females didn’t finish the race, while 24% of males didn’t.
- Conversely, males were more likely to finish the 100 km race. 8.9% of males didn’t finish, while 12.6% of females didn’t.
Guntimes — A Deep Dive
Following my examination of the Finisher versus DNF data, I decided to explore finishing times. First, I explored average finishing times, and then examined those times based on age and gender demographics.
To ensure clarity when reading the graphs below, I decided to convert the times into hours. This means that a time such as 1.5 hours doesn’t mean 1 hour 50 minutes, but 1 hour 30 minutes. If you were a participant in one of the Tarawera races and want to compare your guntime to the ones discussed below, you can use the following formula:
- ((hours*3600)+(minutes*60)+seconds)/3600
Guntimes by Gender
Across the board, males had an average finishing time that was faster than females. But, the time difference between male and female participants was significantly lower for the 100 mile event than for the others:
- For the 21km race, males were 15 percent faster,
- for the 50km, they were 11 percent faster,
- for the 100km, they were 12 percent faster; and
- for the 100 miles, they were just 3 percent faster.
The differences in time based on gender grew when comparing the top 5 males and top 5 females of each race. The top 5 males in the 21km race ran 22.6 percent faster, in the 50km race ran 19 percent faster, in the 100km race ran 18 percent faster, and the 100 mile race also ran 22.6 percent faster.
Guntimes by Age
Next up, I looked at average guntimes by age. Here, it’s important to bear in mind that very few 70–79 year olds and 80+ year olds competed in these races, so data findings about these age groups are not statistically significant. For example both the 100k and the 100 miles had a single participant over 70!
What’s probably unsurprising is that, almost across the board, as the age group got older, the average guntime got higher. However, in the 100km race, 30–39 year olds were faster than the younger 20–29 year old age group, and in the 100 mile race, 30–39 and 40–49 year olds were equally as fast as one another.
40–49 year olds also tended to run at around the overall average speed of the races, except in the 100 mile race, where they were a little faster than the average.
Guntimes — Average of Top 3 Participants For Each Race
Next up, I decided to get an idea of how the pros performed when compared to an average competitor. To do that, I looked at the top 3 participants from each race, and compared their average guntime with the overall average guntimes for each race. Here, I found another interesting tidbit!
For each race, the average of the top three participants guntimes were between 71 and 106 percent faster than the average of all participants in that race. This meant the top three participants:
- for the 21km event were 106 percent faster;
- for the 50km event were 101 percent faster;
- for the 100km event were 90 percent faster;
- for the 100 miles event were 71 percent faster.
Percentiles
Finally, for this section, I also examined percentiles, and looked at maximum, minimum and 25th, 50th and 75th percentiles. Although I didn’t get too many insights from these graphs, they may be fun for participants to look at to see where they were in the percentiles, so I have included them below:
Splits
To finish this project off, I dived into split times. Given the number of splits, I’m going to take a bit of a different approach here, and discuss each race separately.
100 Mile Race
To start, let’s have a look at the 100 mile race! The split averages are up top, the female participants’ split averages are on the left in pink, and the male participants’ split averages are on the right in blue (sorry about the gender stereotyping!).
Interestingly, males and females took the same amount of time to complete split 1 — Green lake, split 4 — Titoki and split 5 -Tarawera Outlet. Further, female participants were actually faster than male participants when travelling between Green Lake and the Buried Village.
This data shows that it was only towards the end of the race where the average male became significantly faster than the average female, with the average male being 40% faster than the average female in the final split.
I then decided to look at the top participants from this race, and decided to first compare the average stats for the top 5 female participants with the stats for the top 5 male participants. The differences between male and female participants were much higher here than when looking at average participants.
The average guntime for the top 5 male participants was 4.07 hours faster than the average for the top 5 female participants, which translates to males running 22% faster. However, I’m aware that this was significantly skewed by Vladimir Shatrov’s super fast time of 15:53:30, with the following competitor almost an hour behind him.
My last step in looking at the split data for the 100 mile race was to compare the top participants with the average times. To do this, I examined the difference between the average split times for the top 3 participants in the 100 mile race and the average split times for all participants in each race.
Noticeable here is that it’s around Rerewhakaaitu where the top 3 participants really began to shine. In the Green Lake split, the top 3 were 47% faster than the average participants, and for the Buried Village split, they were 42% faster. However, these percentages then grew. The average time for the top 3 participants were:
- 62.5% faster for the Rerewhakaaitu split,
- 72% faster for the Titoki split,
- 160% faster for the Tarawera Outlet (!!!),
- 81% faster for Okataina,
- 94% faster for Lake Tikitapu,
- 75% faster for the Redwoods; and
- 220% faster for the last split — taking less than a third of the time of the average runner!
100 km Race
Following all those exciting revelations, I moved on to the 100km race to see what insights I could find there.
Again, I started off by examining the averages of female and male participants in the race. Unlike the 100 mile race, males didn’t become significantly faster than females near the end of the race. For the 100km race, male participants were 10–16% faster than female participants.
The difference between the genders decreased as the race went on, with males being 16% faster than females in the first split, but only 11% faster in the last one. This suggests females may have paced themselves a little better than their male counterparts.
I wanted to compare the top 5 male and top 5 female participants for this race. Unfortunately, when I looked at the data a little more closely, I discovered that two of the top 5 male participants don’t have data for the Titoki split of the race. This meant that these two participants’ Tarawera Outlet splits included the time for both Titoki and the Tarawera outlet.
Because of this, the Titoki and the Tarawera Outlet splits need to be looked at together in order to compare times. For these two splits, the top 5 females were 15% faster than the top 5 males. However, in the other splits, males were faster.
- Split 3 — Okataina: The top 5 males were 23.5 percent faster than the top 5 females
- Split 4 — Lake Tikitapu: The top 5 males were 23 percent faster than the top 5 females
- Split 5 — Redwoods: The top 5 males were 33 percent faster than the top 5 females
- Split 6 — Finish: The top 5 males were 25 percent faster than the top 5 females
To finish off my examination of the 100km race splits, I compared the average of the top 3 participants’ splits with the average splits for the whole race. This data had the same issue as the one above, with one of the top three participants not having data for the first split, resulting in a lower average time for the Titoki split, and a higher average time for the Tarawera split.
Because of this, I added the two together to compare differences in time. Doing this, I found that for these splits, the top 3 participants were 50 percent faster than the average speed of participants in this race. They were also:
- 88 percent faster in Split 3 — Okataina
- 120 percent faster in Split 4 — Lake Tikitapu
- 130 percent faster in Split 5 — Redwoods; and
- 125 percent faster in Split 6 — Finish
The top 3 participants for the 100km were much faster than their competitors. However, this pace was steady throughout the race, whereas for the 100 mile race, the top 3 participants appeared to have paced themselves and then used their last bit of energy to really power from the Redwoods to the Finish!
50 km Race
Next, I moved on to the 50 km race. This race had two segments called Green Lake, so I have added extra labelling to be clear that they are separate, distinct segments.
This race was interesting, as it was the first one I had seen where males and females took the same amount of time to run the last split from the Redwoods to the finish! Here’s the percent differences between the genders:
- Split 1 — Green Lake 1: Males were 16% faster
- Split 2 — Green Lake 2: Males were 9% faster
- Split 3 — Lake Tikitapu: Males were 18% faster
- Split 4 — Redwoods: Males were 5% faster
- Split 5 — Finish: Males and females took the same amount of time!
This may, again, suggest that females paced themselves a little better than males, as the difference in time it takes for males to run the splits when compared to females decreases significantly as the race progresses.
Next up was the comparison between the top 5 male and top 5 female participants in the 50 km race. In terms of gun time, the top 5 males were 19% faster than the top 5 females. The differences by split were as follows:
- Split 1 — Green Lake 1: Males were 17% faster
- Split 2 — Green Lake 2: Males were 18% faster
- Split 3 — Lake Tikitapu: Males were 19% faster
- Split 4 — Redwoods: Males were 15% faster
- Split 5 — Finish: Males were 18% faster
Here, you can see that the top 5 male and top 5 female participants tended to pace themselves in an equal manner, as the percentage differences in split times don’t vary much across splits.
Following this, I moved on to the average split times for the top 3 participants in the 50km race versus the average split times across the 50km race. In regards to guntime, the top three participants were 101% faster than the average runner in this race. The differences for splits were as follows:
- Split 1 — Green Lake 1: The top 3 participants were 85% faster
- Split 2 — Green Lake 2: The top 3 participants were 112% faster
- Split 3 — Lake Tikitapu: The top 3 participants were 140% faster
- Split 4 — Redwoods: The top 3 participants were 109% faster
- Split 5 — Finish: The top 3 participants were 90% faster
21 km Race
To finish this all up, let’s have a look at the 21km race!
In the 21km race, the average split time for males was consistently shorter than the average split time for females. Here’s the percent differences between the genders:
- Split 1 — Lake Tikitapu: The average male was 16% faster than the average female
- Split 2 — Redwoods: The average male was 17% faster than the average female
- Split 3 — Finish: The average male was 14% faster than the average female.
The margins between genders grew further when I compared the top 5 male participants with the top 5 female participants.
- Split 1 — Lake Tikitapu: The top 5 males were 20% faster than the top 5 females
- Split 2 — Redwoods: The top 5 males were 23% faster than the top 5 females
- Split 3 — Finish: The top 5 males were 20% faster than the top 5 females
Finally, I examined the average split times for the top 3 participants in the 21 km race with the average times for all participants in this race. In terms of guntime, these participants were, on average, 106% faster than the average guntime.
In regards to each split, the top 3 participants were :
- 94% faster in Split 1 — Lake Tikitapu
- 108% faster in Split 2 — Redwoods
- 110% faster in Split 3 — Finish
This was consistent with other races, where the top 3 atheletes tend to be a bit under 100% faster in the first split, before increasing to more than 100% in later splits. This is presumably because the average participant is excited and may start out fast, not pacing themselves as much as the pros!
Limitations
I would have liked to look more into race geography and how speeds between pros vs average competitors compared when it came to more dramatic inclines and declines. However, the dataset was a little limited.
For example, for the 100 mile race, the official race map showed 15 checkpoints. However, SportSplits data only provided information from 9 checkpoints. Because of this, the extra knowledge gained from this information would be fairly limited, as the splits are spread out across long distances.
Final Words
So that’s it! If you have any questions about the data, I’d be happy to respond. Also, if you want me to do more of a deep dive into a specific race or a question you have about the race, let me know — I’d be more than happy to add to this to help people explore the data even further!
Credits
Sportsplits for their data — link here
Tarawera Ultra Marathon — link here
