6m Vaulters - a career analysis

[El Toro]

I started this thread to undertake further investigations into the career development of athlete's PV performance after discussions on the Mondo thread.

Background

'cajun5150' provided a comparative table of average annual heights by age for Bubka, Duplantis, Lavillenie & Kendricks

The young man has hops!

[ATTACH=CONFIG]248[/ATTACH]

In response to this I proposed:

Interesting. It seems to suggest that elite vaulters reach their best average year around 10 years after start of vault training and then irreversibly decline from there.

It would be nice to see a larger study but I'm not sure if there is sufficient data available.

'AS' then rightly pointed out that my underlying assumptions was incorrect

The table doesn't imply that the first year is when the vaulter's started training. It simply shows when decent data commences for the individual.

General data discussion

While there might be detailed career training records for Duplantis and some other selected athletes, it is unlikely that there will be sufficient numbers to give any meaningful analysis.

The original table was also based on average performance by chronological age. While interesting, that doesn't have much value when comparing athletes with very different paths to elite performance.

So, where to get data for a more useful analysis? I settled on the IAAF "Results" tab contained on each athlete's profile page. This contains the following data elements:

• Date (of competition in dd-mmm-yy format e.g.23-Aug-81)
• Competition (description e.g."Utrecht European Junior Championships")
• Cnt. (Country of competition)
• Cat (IAAF competition rating category A,B,C,D,DF,EF,GL,GW,OW & 0)
• Race (the nature of competition e.g. qualifying or final)
• Pl. (Place in competition)
• Result (best vault height)

I limited my selection to those athletes who have cleared 6.00m or greater (n=18) and extracted all outdoor competition data for each year of data available. I limited the data to outdoors competition only because not all athletes had indoor opportunities available and, in general, the primary peaking focus in each year is within the outdoor season.

An examination of the extracted data set shows that, for most athletes, it appears to commence in the first year for which the athlete has a recorded height of 5.00m or greater. A couple of athletes had a first year without a 5.00m jump, so that year was excluded. This provides a standardised starting point for examining the development of pole vaulters over time.

I then classified each calendar year for an athlete as a sequential peformance year (eg Bubka, 1981 = Bubka Year 1 & Lavillenie, 2006 = Lavillinie, Year 1). This allows common years of comparison across all athletes careers and, in combination with starting from a common performance standard, the issue of chronological age comparisons is avoided.

It should be noted that the IAAF does not seem to provide a statement of data collection methods, so I have been unable to confirm that the data is equally comprehensive for all athletes. However, there seem to be enough competition records (n=3,488) to provide useful information.

I will stage posts over a number of days to fit within the limitations of the message board.

[El Toro]

Detailed data discussion

Starting with the full data set described above, I excluded all non-finals performances (n=172) given they have inherent performance caps for elite performers.

I then excluded finals performances that were NH, DNS and other non-height results along with Exhibition jumps (n=307). Exhibitions were excluded on the simple basis that the height data is not included in the relevant IAAF data page.

In recognition of southern hemisphere athletes having a domestic season equivalent to a northern hemisphere indoors and some northern hemispher athletes having access to outdoor during the indoor season, performance months were limited to May to October.

The remaining data reflects as much as possible, best efforts in reasonable conditions over time. This data is the basis for all subsequent analysis.

High level analysis

The combined data set (see attached chart) for the average of all athletes shows very fast gains from Year1 to Year4 of 5m+ competition - an average of 5.08m to 5.62m. This rapid progress is followed by a more gradual rise to peak average performance by Year11.

The peak year is surrounded by a plateau containing 7 years of broadly similar peak performances (Year7 to Year 13). This plateau is followed by the inevitable annual decline interrupted by late career rallies for the 7 athletes still competing, prior to drastic decline.

It is interesting to note, that despite the apparent physical demands of the event, elite pole vaulter have considerable longevity with an average career of 15.5 years for the 14 retired athletes in the group.

I will post some more detailed breakdowns of the group data tomorrow.

[cajun5150]

Detailed data discussion

Starting with the full data set described above, I excluded all non-finals performances (n=172) given they have inherent performance caps for elite performers.

I then excluded finals performances that were NH, DNS and other non-height results along with Exhibition jumps (n=307). Exhibitions were excluded on the simple basis that the height data is not included in the relevant IAAF data page.

In recognition of southern hemisphere athletes having a domestic season equivalent to a northern hemisphere indoors and some northern hemispher athletes having access to outdoor during the indoor season, performance months were limited to May to October.

The remaining data reflects as much as possible, best efforts in reasonable conditions over time. This data is the basis for all subsequent analysis.

High level analysis

The combined data set (see attached chart) for the average of all athletes shows very fast gains from Year1 to Year4 of 5m+ competition - an average of 5.08m to 5.62m. This rapid progress is followed by a more gradual rise to peak average performance by Year11.

The peak year is surrounded by a plateau containing 7 years of broadly similar peak performances (Year7 to Year 13). This plateau is followed by the inevitable annual decline interrupted by late career rallies for the 7 athletes still competing, prior to drastic decline.

It is interesting to note, that despite the apparent physical demands of the event, elite pole vaulter have considerable longevity with an average career of 15.5 years for the 14 retired athletes in the group.

I will post some more detailed breakdowns of the group data tomorrow.

How much variation is there around the mean for each year?

[26mi235]

Year 4 though 18 (17?) at within about 10 cm of the peak so there is a very long peak period with a plateau within it. I suspect that there is a 'survivor' bias on the right tail as those no longer able to make 560 departing and their capability with training might be 5.30-5.50, lowering that tail. Of course, in the last several years athletes might be training only intermittently so they might have been able to go higher. Just thoughts about the data.

[El Toro]

How much variation is there around the mean for each year?

Thanks for the question 'cajun5150'.

There is quite a bit of variation between the best and worst performances in any year. This is not unexpected because we are looking at all athletes, so it only takes one good or bad competition by one athlete to set the limits.

I have presented this data in the attached chart. Despite this being a crude presentation of the data, it does show some interesting things.

As expected, the range is much greater in the early years, most likely reflecting the establishing of consistent technique.

Once established technically, there is much higher minimum performance that aligns with the peak average years, whereas the peak performance barely changes. This tends to indicated that increases in average performance derive from eliminating crap performances rather than necessarily increasing peak performance. This is where Duplantis currently shines with his high level minimum consistency despite a PB below this elite group.

Interestingly, even though average performance in the declining years return to around the same level as the early years, the minimum performance doesn't decline to the same extent. This probably reflects an ability to maintain a minimum technical skill even while peak performance declines.

I will look at this issue in more detail in the next post.

[El Toro]

In the previous post, I looked at the range of performances around the mean. As discussed previously, the outliers are for single performances and, while interesting in themselves, do not tell us anything about the bulk of performances.

If we want to understand how much variaton there is for all the performances, a better approach is to use the standard deviation. This measures how concentrated the data is around the mean; the more concentrated, the smaller the standard deviation.

I have attached the same chart as the last post but replaced the best and worst performances with +/- one standard deviation from the mean.

This smooths out the chart substantially but still confirms the three stages of rapid growth, long peak and steady decline. Interestingly, in Year9 and Year 10, the maximum declines while the minimum holds or increases prior to the peak average in Year 11.

In the next post, I will chart the standard deviation by itself.

[El Toro]

As promised, I have attached a chart with the standard deviation by itself.

Because the standard deviation represents a variation in performance, the units for standard deviation are the same as that for the base performance - metres. So 0.27 for Year 1 represents 27 centimetres.

You can see how the standard deviation declines until Year 10, meaning that the athletes are achieving more consistent performance as technique and other factors mature.

You may note that consistency appears to get better in the latter years after a mid-career plateau but caution needs to be used here because the numbers of performers and performances decline. Given that '26mi235' raised the issue of the tail, I leave it to him to comment on the data in the table below.

Year	Performances	Performers
Year1	19	11
Year2	36	12
Year3	63	11
Year4	93	13
Year5	106	13
Year6	119	13
Year7	167	14
Year8	146	14
Year9	201	14
Year10	146	13
Year11	170	14
Year12	180	14
Year13	147	14
Year14	121	12
Year15	95	9
Year16	85	9
Year17	76	7
Year18	55	4
Year19	28	2
Year20	15	2
Year21	8	1
Year22	6	1

Let me know if you are finding this information useful and any feedback on visual presentations are treatment of the data is welcome. I will start start to look at groupings of athletes tomorrow.

[gh]

how I remember why I quit taking math as soon as possible

[El Toro]

how I remember why I quit taking math as soon as possible

That's why I put in the pretty pictures!

[26mi235]

I am interested but otherwise occupied.

[El Toro]

So far We have looked at all 6m vaulters as a group, so it is now time to take a look under the hood at subsets of the group and eventually individual athletes.

Firstly, let us consider whether the location of an elite vaulter affects their career performance profile. Traditionally athletes from Europe and North America dominated the pole vault but in the last 20 years, athletes south of the equator have started to have an impact.

There are inherent difficulties for southern athletes in combining domestic competition with travel to key competitions in the northern hemisphere. Will this make a difference to the shape of their overall careers?

The athletes based in the Southern hemisphere are; Braz, Brits, Burgess, Hooker & Markov (n=5, 28%).

Northern hemisphere athletes are; Bubka, Gataullin, Hartwig, Kendricks, Lavillenie, Lobinger, Mack, Otto, Stevenson, Tarasov, Trandenkov, Walker & Lukyanenko (n=13, 72%).

The attached chart shows the average performance by hemisphere as well as for the entire group - the latter is the same data as the original chart.

Because the lines overlap in some places, I've shaded the edge of data boxes for Northern athletes to avoid confusion. Southern athletes have just the plain numbers and the average for the group does not have any data boxes.

As we can see in the chart, there is a clear difference between Northern and Southern athletes with Northern athletes having a higher average performance and a much smoother performance curve above the average for all athletes.

The only exception to higher average performance by Northern athletes is a much higher average in Year4 for Southern athletes but this could just be an artefact of the smaller sample. Southern athletes also have a shorter peak plateau of 4 years vs the 8 years of Northern athletes.

In addition, there is a clear difference in longevity with no southern athletes making it past Year16 compared to 7 athletes from the northern hemisphere (39% total,54% Northern).

Of course, the still active South athlete, Braz is only at Year9, so may still get there. Other active athletes, all from the North are Kendricks at Year8, Lavillenie at Year 13 while Lukyanenko is already at Year16.

In the next post, I will look at continental differences.

[El Toro]

This post continues the examination of geographical location on performance profiles. Of course, geography is really a proxy for training systems and the quality of support available for this most technical of events.

Asia is the only continental association without a 6m vaulter. However, given the recent demonstration in high quality sprinting and horizontal jumps in China, there appears to be no reason why a 6m vaulter could not be produced in the next decade.

There are three continents with 3 or more athletes - Australia, Europe and North America. Africa and South America only have one athlete each, so I have combined the data for Braz and Brits for convenience.

In a further examination of Europe, I have broken down the continent into West (France, Germany) and East (Ukraine, Russia and URS) again to reflect the different training systems.

The breakdowns are as follows:

Chart 1
Europe (n=8) - Bubka, Gataullin, Lavillenie, Lobinger, Otto, Tarasov, Trandenkov & Lukyanenko.
North America (n=5) - Hartwig, Kendricks, Mack, Stevenson & Walker.

There is a clear difference betwen North America and Europe. Although, Europe gets to an average performance greater than 5.60m a full year faster than North America, it's progress slows down before hitting a peak of 5.77m in Year 13 with a full 11 years averaging 5.65m or more.

In sharp contrast North American athletes continue a more rapid improvement to hit maximum performance in Year8 with only 6 years above 5.65m. Does this represent the differences between collegiate and club-based systems?

Chart 2
Europe_East (n=5) - Bubka, Gataullin, Tarasov, Trandenkov & Lukyanenko.
Europe_West (n=3) - Lavillenie, Lobinger & Otto.

In the same way that developmental processes for athletes are different between Europe and North America, there were and perhaps still are, significant differences between the Soviet (and subsequent constituent countries) and the Western nations.

Chart 2 shows that both parts of Europe have a broadly similar shape to their careers. The primary difference is that Eastern Europe appears to be better in every year and has 14 continuous years (Year4-17) with an average of 5.65m or more compared to half that for the West (Year7-13).

Despite the differences in performance levels, both systems hit their peaks at the same time - East in Year11-13 and West in Years12-13.

Lukyanenko is the only "modern" Eastern athlete with all others having at least half their careers in the USSR. A further breakdown could be interesting.

Chart 3
Australia (n=3) - Burgess, Hooker & Markov.
Africa/South America (n=2) - Braz & Brits.

As we saw in the earlier post, athletes from the Southern hemisphere have much less consistent career progression than Northern athletes, so it is not surprising to see that smaller subsets have even more volatility.

While Brits and Braz show the consistent linear increase in earlier years evident in the Northern hemisphere, Australian vaulters have a much less consistent progression than any grouping examined so far.

Australians also have the shortest careers of any grouping, noting that Braz is only in Year9 but is still active.

I am not sure why Australian development is like it is without digging a bit deeper. That will be a post for another day.

[lonewolf]

Commendable stuff.. El Toro is hereby awarded a Phd in PV analysis.

[El Toro]

Commendable stuff.. El Toro is hereby awarded a Phd in PV analysis.

Thanks. I hope you find at least some of it interesting or at least that it's useful in addressing insomnia.

However, I think I'm a bit short of the necessary 100,000 word thesis for a PhD (and let's not talk about the limitations of my analytical skills).

Of course, before I started this thread, I assumed that somebody would already have done something similar and I would just post a link and move on. Sadly, not the case.

There are lots of papers on physical attributes and their link to performance as well as biomechanical analysis of technique but nothing I could find on career progression.

[El Toro]

In advance of the Euro Champs this week, I will post a short series of posts of Duplantis vs "..." and then take a break from this thread for the duration of the Euros.

First up, Mondo vs the USA 6m club (see attached).

It's pretty obvious that he has clearly surpassed the average career progression of the other vaulters. More impressive is that in each year from Year2, he increases the distance between his average and the average of the rest - Yr1 -3cm, Yr2 +12cm, Yr3 +13 & Yr4 +29!