Running Performance Predictor
In 1981 I wrote a paper called “Athletic Records and Human Endurance.” It was published in American Scientist. I had around 200 requests for reprints from people all over the world, mostly via postcards. It contained an equation which I called the “endurance equation” which could be used to roughly predict potential finish time at a distance based on past performance at another distance.
Now, after a quarter-century hiatus, interest seems to have revived. Herman Nieuwendaal has taken the thing forward into the computer age and has produced a very nice web page which may be of interest to prediction aficionados.
You can find it at:
http://www.electricblues.com/runpro.html
The original "Athletic Records and Human Endurance" may be found at:
ARHE
Running Performance Predictor
Hello Mr. Riegel,
I was wondering. Is this thread dead? Are you still available to answer a few questions?
Thanks
Charles
I was wondering. Is this thread dead? Are you still available to answer a few questions?
Thanks
Charles
Charles,
It's not dead. I'm available.
I was a runner for a couple of decades. As an engineer I always recorded my data and tried to make sense out of it.
When I graphed distance vs time on log-log paper I got a straight line. This, of course, led to the exponential equation in "Athletics Records and Human Endurance."
You can read the article in its entirety at: http://www.runscore.com/course...nt/Articles/ARHE.pdf
The link in my original posting had expired. Thanks to your inquiry it's fixed.
Best, Pete
It's not dead. I'm available.
I was a runner for a couple of decades. As an engineer I always recorded my data and tried to make sense out of it.
When I graphed distance vs time on log-log paper I got a straight line. This, of course, led to the exponential equation in "Athletics Records and Human Endurance."
You can read the article in its entirety at: http://www.runscore.com/course...nt/Articles/ARHE.pdf
The link in my original posting had expired. Thanks to your inquiry it's fixed.
Best, Pete
OMG good thanks!
I have a first very precise question.
Fatigue Decay = Log(T1/T2) / Log(D1/D2)
Did you use this function to find your best fit curve in the late '70s? Or something similar?
I know next to nothing in maths, so if possible try to keep the explanations simple enough for a 10yo kid to understand ;-)
Thanks a lot
Charles
I have a first very precise question.
Fatigue Decay = Log(T1/T2) / Log(D1/D2)
Did you use this function to find your best fit curve in the late '70s? Or something similar?
I know next to nothing in maths, so if possible try to keep the explanations simple enough for a 10yo kid to understand ;-)
Thanks a lot
Charles
Charles,
I was aware of the log(T1/T2) relationship, but it was not until I fiddled with my data and those of better runners that I saw how well it matched the running data. My interest then spread to other endurance sports and culminated in the American Scientist article. It's satisfying to see that I am still getting inquiries about that three-decades-old paper..
As for originality of concept, conclude what you wish. I make no claim to being first, but I am not aware that anybody beat me to it. I certainly was not the first to attempt to quantify athletics, but I have seen no other formulae as simple as what I proposed. The price of simplicity, of course, is the necessity to limit the use of the formula to the mile-to-marathon range, thus shutting out the sprints.
Best, Pete
I was aware of the log(T1/T2) relationship, but it was not until I fiddled with my data and those of better runners that I saw how well it matched the running data. My interest then spread to other endurance sports and culminated in the American Scientist article. It's satisfying to see that I am still getting inquiries about that three-decades-old paper..
As for originality of concept, conclude what you wish. I make no claim to being first, but I am not aware that anybody beat me to it. I certainly was not the first to attempt to quantify athletics, but I have seen no other formulae as simple as what I proposed. The price of simplicity, of course, is the necessity to limit the use of the formula to the mile-to-marathon range, thus shutting out the sprints.
Best, Pete
Hi Pete,
"you wrote:"
I was aware of the log(T1/T2) relationship
- -
The contrary would have surprised me a bit. After all, you had to come up with this 1.06 coefficient one way or the other.
"you wrote:"
It's satisfying to see that I am still getting inquiries about that three-decades-old paper..
- -
Oh sir. I think it's not the end of the road. Far from it. You were, at the time, very far up front. What happened is that Running world picked on this concept and then Dr.Daniels et al. did wonderful things with it.
Believe it or not, some cyclic endurance sports to which this concept could apply, that are nowadays overloaded with web information, probably ignore its existence.
Late '80s, like I explained you in private a famous mathematician ended up solving the time prediction mystery applied to swimming with the same function set. However, they took a completely different approach. They worked as hard as they could to keep the formula set secret in hope of commercializing a software. As a result of that, the concept isn't wide spread in the swimming world (yet).
"You wrote:"
The price of simplicity, of course, is the necessity to limit the use of the formula to the mile-to-marathon range, thus shutting out the sprints.
- -
Oupsssss. This may in fact has been Dr.DeKoninck's contribution to the evolution of this concept. You can trust my word, this formula set can predict sprint performance very accurately. It has been tested against a scientifically significant volume of data gathered from elite level swimmers from around the world.
The thing he discovered though, the caveat, is that if you wanted to predict sprint performances, you had to compute your individual fatigue decay out of a few sprint personal bests. Input the 50m (roughly 25s) then 100m (roughly 1min) and it will predict anything between 50 and 800 with a fair level of accuracy.
"You wrote:"
As for originality of concept, conclude what you wish. I make no claim to being first, but I am not aware that anybody beat me to it.
- -
I'm coming to the same conclusion. So did Alejandro Martinez, a talented engineer that paid tribute to your concept in this article here (sorry, in Spanish)
http://www.amtriathlon.com/200...ento-en-natacin.html
I also come to the same conclusion.
Now, the question I have been wondering about for years now:
Can this concept be further developed to use it not only to predict performances, but also to quantify training loads? That, Mr.Riegel, would be very handy!
The two best models at the time to quantify swim training loads are Banister's TRIMP and Skiba Swim Score. No need to comment on the former. The later is interesting since it uses the swim velocity (performance itself) as input. However, it tries to translate velocity into power prior issuing a session score.
If we could find a way of quantify a training load based on swim velocity alone, that could probably be far more accurate.
Thank you so much for your quick answers, your passion and availability. I will keep you posted about anything that I may publish in regards to your concept. In the meantime, if you're already aware of an application of this concept aimed at scoring a session, I would appreciate being made aware about it.
Thanks again!
Best,
Charles
"you wrote:"
I was aware of the log(T1/T2) relationship
- -
The contrary would have surprised me a bit. After all, you had to come up with this 1.06 coefficient one way or the other.
"you wrote:"
It's satisfying to see that I am still getting inquiries about that three-decades-old paper..
- -
Oh sir. I think it's not the end of the road. Far from it. You were, at the time, very far up front. What happened is that Running world picked on this concept and then Dr.Daniels et al. did wonderful things with it.
Believe it or not, some cyclic endurance sports to which this concept could apply, that are nowadays overloaded with web information, probably ignore its existence.
Late '80s, like I explained you in private a famous mathematician ended up solving the time prediction mystery applied to swimming with the same function set. However, they took a completely different approach. They worked as hard as they could to keep the formula set secret in hope of commercializing a software. As a result of that, the concept isn't wide spread in the swimming world (yet).
"You wrote:"
The price of simplicity, of course, is the necessity to limit the use of the formula to the mile-to-marathon range, thus shutting out the sprints.
- -
Oupsssss. This may in fact has been Dr.DeKoninck's contribution to the evolution of this concept. You can trust my word, this formula set can predict sprint performance very accurately. It has been tested against a scientifically significant volume of data gathered from elite level swimmers from around the world.
The thing he discovered though, the caveat, is that if you wanted to predict sprint performances, you had to compute your individual fatigue decay out of a few sprint personal bests. Input the 50m (roughly 25s) then 100m (roughly 1min) and it will predict anything between 50 and 800 with a fair level of accuracy.
"You wrote:"
As for originality of concept, conclude what you wish. I make no claim to being first, but I am not aware that anybody beat me to it.
- -
I'm coming to the same conclusion. So did Alejandro Martinez, a talented engineer that paid tribute to your concept in this article here (sorry, in Spanish)
http://www.amtriathlon.com/200...ento-en-natacin.html
I also come to the same conclusion.
Now, the question I have been wondering about for years now:
Can this concept be further developed to use it not only to predict performances, but also to quantify training loads? That, Mr.Riegel, would be very handy!
The two best models at the time to quantify swim training loads are Banister's TRIMP and Skiba Swim Score. No need to comment on the former. The later is interesting since it uses the swim velocity (performance itself) as input. However, it tries to translate velocity into power prior issuing a session score.
If we could find a way of quantify a training load based on swim velocity alone, that could probably be far more accurate.
Thank you so much for your quick answers, your passion and availability. I will keep you posted about anything that I may publish in regards to your concept. In the meantime, if you're already aware of an application of this concept aimed at scoring a session, I would appreciate being made aware about it.
Thanks again!
Best,
Charles
When I wrote "Athletic Records and Human Endurance" I realized that it might have some practical applications, but I was not very interested in what those might be. I did produce a small slide rule which I sold (this was a few years before the "American Scientist" article). It was described in a short "Runner's World" article. You can see it at
Race Pace Computer
Race Pace Computer
Hello Mr/Dr. Reigel,
I have been recently introduced to your name by the very Charles who posted here. We've become aquainted via the Total Immersion swim forums.
I am a tri coach and use what I believe is another derivative of your work on McMillanRunning.com. I've also been a big fan of Jack Daniels work as well, and have just recently learned of your paper.
I googling your work I found this thread, with my very own friend Charles right here! The internet is like a time machine.
Thank for your contributions!!!
I have been recently introduced to your name by the very Charles who posted here. We've become aquainted via the Total Immersion swim forums.
I am a tri coach and use what I believe is another derivative of your work on McMillanRunning.com. I've also been a big fan of Jack Daniels work as well, and have just recently learned of your paper.
I googling your work I found this thread, with my very own friend Charles right here! The internet is like a time machine.
Thank for your contributions!!!
quote:The internet is like a time machine.
Which makes lying very difficult
I hope Pete hasn't abandoned this old thread.
I am still here, but I'm no longer a participant in the time-predicting game. I'm afraid that further research will have to leave me behind.
At the time I did my work I was able to read a number of time-prediction models, but was not satisfied with them. So I took my own shot at it. It was a bit later when I realized that other sports had data that could extend the use of the basic formula.
I was surprised when American Scientist was willing to publish it, and I attribute this to a bit of luck. The editor, Michelle Press, was a runner, and this may have influenced her choice.
I like the formula as it stands, as it combines accuracy with simplicity.
So, I am watching the thread but have nothing further to contribute.
At the time I did my work I was able to read a number of time-prediction models, but was not satisfied with them. So I took my own shot at it. It was a bit later when I realized that other sports had data that could extend the use of the basic formula.
I was surprised when American Scientist was willing to publish it, and I attribute this to a bit of luck. The editor, Michelle Press, was a runner, and this may have influenced her choice.
I like the formula as it stands, as it combines accuracy with simplicity.
So, I am watching the thread but have nothing further to contribute.