Sorry, to clarify, I’m not suggesting my result indicated a pace I could maintain for 60 minutes or longer (it indicated 29 minutes to be exact). I’m saying that if you treat the model purely mathematically, it does suggest you can run at the CV pace obtained for 60 minutes or longer, even though in practice it is observed and recognized that you can’t.
My question, acknowledging that, is why not just define it as 30 minute race pace? And I don’t mean you specifically but the idea in general. Save the 4-5 time trials and hyperbolic function for sprint and middle distance athletes and just use a single race effort to approximate CV. Also, from that race effort, you can obtain classic race distance predictions via standard methods. The hyperbolic function can’t provide that (using the 2-20 minute race effort protocol) because it asymptotes at approximately 30 minute race pace, in practice.
On that last topic, I read a paper that Andy Jones published where he used 1500m - 15km times of 12 elite athletes to approximate CV and compare it to marathon race pace. That’s obviously slightly different than the 2-20 minute race time protocol but the results he obtained were still interesting. Of course, that interested me (as you say, I’m data man), so I pulled the data for the top 200 times run in the 1500m, 3000m, 5000m, 10000m, half, and marathon. Using just the 1500m - 10000m I fit the model to those 200 performances in each category, so it is an approximate average. The CV for that dataset is 360.26 meters per minute, which if you multiple by the average time for the 10km of the group works out to 9,748.64 meters, so slightly slower than 10km pace for a group that averaged 27:04 for 10km (369.55 m/min).
To illustrate my point about the mathematical implications of the model though, for that group, the hyperbolic function is
Time = 274.99 / (Speed - 360.26)
Say you run 3 meters per minute slower than CV, so 363.26 meters per minute. The result would suggest you could run for 91.66 minutes at that pace. Using the half marathon from that dataset shows an average time of 59:16, which is 356.01 meters per minute. So obviously slower than the speed predicted, meaning it wouldn’t be possible to maintain that speed for 60 minutes, in practice, let alone 91.66 minutes.
This is my point about the undesirable asymptotic behavior of the model. I hope that clarifies it. I was tempted to use more graphs and analysis but refrained… for now haha. Also, I hope this shows that I’m not being disingenuous in saying I’m interested in this concept. There are just some aspects of it that seem peculiar when going from theory to practice.