100m WR predictions

Members Area In the News
1

How Fast Can A Human Run The 100 Meter Sprint?

By News Staff | August 6th 2009
.
Usain Bolt, sprinter from Jamaica, currently holds the world record in the 100 meter sprint with a time of 9.69 seconds. Whenever new records are set, people ask ‘what is the limit on human performance?’

So how fast can a human run?

Two econometricians from Tilburg University in the Netherlands, Professor of Statistics John Einmahl and former student Sander Smeets, say they have calculated the ultimate records for the 100-meter sprint.

The good news; there is still room for improvement in both the men’s and women’s times in the near future.

They used extreme value theory to calculate by how much the current records for the 100 meter sprint could be improved.

Extreme-value theory is a sub-sector of statistics, which tries to answer questions about extreme events (which by definition are uncommon), using information about less extreme events. The theory is normally applied within the financial and insurance world to estimate the risk of extreme damage resulting from storms, earthquakes or the bursting of a dyke, for example, in order to calculate premiums.

With a little modification, they say it can apply to sports as well.

Einmahl and Smeets analyzed the records of 762 male and 479 female athletes. Each athlete was listed once, and the times were recorded between January 1991 and June 2008.

Times run before 1991 were discounted on account of the inadequate doping controls before this date. The men’s times varied between 9.72 and 10.30 seconds, and the women’s from 10.65 to 11.38.

(LOL, they presume things have improved? kk)

According to Smeets and Einmahl, [b]the fastest time that the men are capable of sprinting is 9.51 seconds, which knocks 0.18 seconds off Usain Bolt’s current world record. For female 100m sprinters, another 0.16 seconds can be knocked off the 10.49 run by Florence Griffith-Joyner, which would mean a time of 10.33 for women.

In a more cautious estimate (with a 95% confidence interval), the predicted times are 9.21 for the men and 9.88 for the women.[/b]

Sander Smeets studied Finance and Actuarial Sciences at Tilburg University and now works as a junior actuary at AZL, in Heerlen. John Einmahl is Professor of Statistics at Tilburg University.

Paper: ‘Ultimate 100m world records through extreme-value theory’, CentER Discussion Paper nr. 57

2

I love stuff like this.
econometricians?

Using a complicated formula involving what I ate for breakfast and a Miles Davis solo, I’ve determined that the 100Meter world record in 2050 will be whatever the 100Meter world record is in 2050!

3

I take back what I said.

4

Usain Bolt; 9.69 seconds.

Eased up; cost about 0.08 secs, = 9.61.
Slightly improved reaction 0.02, = 9.59

Train for nothing but the 100m instead 100 & 200m. 0.04 secs, = 9.55

A few years older, a little stronger. 0.05, = 9.50 secs.

Wind in Beijing was 0.0 secs. (+ 1.5 m wind, = 0.09 secs?) = 9.41 secs.

High altitude instead (say 0.08 ) = 9.33 secs.

2 round grand prix event, instead of four round tornament over 2 days (0.02/3 for Usain, not a lot of differance) = 9.31 secs.

Improved track (beijing was a quick track), improved spikes (they’re allready good) 0.01
= 9.30 secs

A better training method. 0.10 secs.
= 9.20 secs

An even more gifted athlete than Usain Bolt. (If that is possible). 0.05 secs.
= 9.15 seconds.

I’ve bored myself so I’ll stop there. So then:
9.15 seconds is the llimit, cause that’s where Goose232 ran out of scientific patiance. Lol

5

You’re right. You should be embarrassed. You did not bother to read through the assumptions or claims made by the econometricians, instead you took what was a headline article and took it at face value.

They did not estimate what the human limits were in the 100m - they estimated what the record could be in the very near future. This is explicitly stated in the paper and implied in the article. So basically all the assumptions you have listed sarcastically are accounted for. And as a maths major rainy, you should know better than to take stats literally - it is not physics, it does not purport to be deterministic. If your degree is the paper it’s printed on why don’t you critique their methods and give us something interesting to discuss rather than making “horribly stupid” generalisations which add absolutely no value to the board.

6

You make a good point. I will read through the paper (http://center.uvt.nl/pub/dp2009.html), but what about:

Two econometricians from Tilburg University in the Netherlands, Professor of Statistics John Einmahl and former student Sander Smeets, say they have calculated the ultimate records for the 100-meter sprint.

7

Dazed, you’re right. Thanks for the check. My apologies to the board and the paper authors. I have no issues with what they’ve done. Rather than attack them, I should have discussed the poor reporting of what they said.

My issues with the paper are fairly minor. Without spending considerable time on it, I’m unable to understand most of the math leading up to the results.

They ignore all times before 1991, due to ‘inadequate doping’ controls. Due to forum rules, I’m not going to discus further why I think this is a mistake. It removes quite a few strong times.

They use times from 1991 to 2008, but then use this data to predict the current fastest possible 100m time. The majority of the times they’ve used are from people no longer running.
Their data analysis does not include anything pertaining to when someone’s PB was set, so the result should not be interpreted to be for the current top runners, just any particular crop of runners in general.

They don’t include Bolt’s 9.69. While his 9.72 is pretty close, and I don’t think it would change their result much if any, it seems silly to leave out the current WR, and I’d want to know what their model would predict with it included.

8

It’s a news paper article reporting on a working paper, each with completely different objectives in mind. Surely you’ve been around long enough to have witnessed how scientific findings are sensationalized to capture the public imagination. which although not necessarily a bad thing for the sport, doesn’t make for good science of analysis of findings.

It’s like the conclusions that are reached on athletes training from tidbits reported in news articles - they aren’t necessarily untrue, but are devoid of the context that the statements were made under, or sometimes just misinterpreted by the author.

9

I discovered while reading the paper that the authors used the term ‘ultimate record’ to mean the very fastest time that would seem statistically possible right now, not the ultimate fastest in the future.

10

I agree, it does seem a bit strange that they are not using the most recent times. Mind you, they are likely to have been fairly restricted by the data set that they used. I am only guessing, as I am not familiar with extreme value analysis and first thought they may have just been using a 95% prediction interval.

I also thought that the analysis would be biased by time. Its a shame that they didn’t adjust for this as it could also correct for possible evolution in technology.

11

Interestingly, their 95% confidence limit of 9.21 is 95.05% of the current WR of 9.69.

12

Its ~95% of 9.69, but unless you know how the times are distributed a confidence interval, which is more or less an area, can’t be computed.