What is critical power?

A brief historical sketch of the critical power concept, with a few things you might not know about its development. I also cover the basics of fitting data and some of the pitfalls in doing so. Some of the resources used include:
Wilkie (1985): journals.biologists.com/jeb/a...
Garcia-Manso (2006): doi.org/10.1016/j.jtbi.2012.0...
Rohmert (1960): link.springer.com/article/10....
Monod and Scherrer (1965): www.tandfonline.com/doi/abs/1...
Poole et al. (1988): www.researchgate.net/profile/...
Martin et al. (2015): onlinelibrary.wiley.com/doi/a...

[@languid100]

Great presentation, thanks. Look forward to next week.

[@sameroulis]

What an awesome presentation! Much appreciated!! Thank you

[@Surfsailwaves]

Thankyou very much for this masterclass: fascinating, clear and entertaining.

[@nitro1sport]

Great work, thank you mark. Can one improve his critical power and if so, how can it be done ?

[@all-outphysiology2177]

Yes. Aerobic training increases it. So will breathing oxygen-enriched air. The latter is usually impractical outside a lab.

[@nitro1sport]

@@all-outphysiology2177 Thank you. To the best of my knowledge, most (if not all) training studies does not use CP as an intensity indicator for their intervention protocol (instead of % of HRmax, % of VO2max, LT1 and LT2 and so on). Is it so and if it is, what on your opinion, is the reason for that ?

[@all-outphysiology2177]

@@nitro1sport simply the length of time it takes to establish it if you don't use the 3 min test!

[@akreutzer82]

Thank you for this and all of your other great videos! I have a quick question. If you fit a CP model using the 1/time-power relationship, what would you consider a minimal R2 to deem the model a good fit? Thanks!

[@all-outphysiology2177]

The fit is more than R^2 of course, but you should typically see values of >0.98. Standard errors of

[@akreutzer82]

@@all-outphysiology2177 Thank you!

[@hallo108]

Great presentation! In terms of using CP for evaluation of training improvement in highly trained athletes, would you say 3 test effort would be sufficient? For example 3, 6 and 14 min.

[@all-outphysiology2177]

Thanks. That would work, certainly.

[@gilleek2]

I must have this @rseways. I would have expected that the more trained you are, the more data points you'd want in order to be sure of whether you've made improvements or not as beyond a certain point your improvements are gonna be smaller and smaller?? Would testing for CP and W' be useful for amateur ironman athletes or is it most useful for shorter more stochastic events?

[@all-outphysiology2177]

@@gilleek2 yes, and also no! The big issue is that each predicting trial is exhaustive in nature, and do doing 5+ trials in a rested state is not really feasible for athletes. In practice, athletes and coaches use performance in training and competition to construct a power-duration curve. This is useful in practice, but may not really represent the limits. With respect to ironman triathlon, yes the CP and W' are probably of less relevance than lactate threshold given the duration of the event. But it will still be useful in setting training intensities and interval sessions.

[@gilleek2]

@@all-outphysiology2177 Okay. Could you do tests on different days perhaps after a rest week to make it most manageable or does that reduce the predictive nature of the protocol?

[@all-outphysiology2177]

@@gilleek2 You could, but I don't know of many athletes who would be willing to sacrifice a week of training for fresh exhaustive tests that last less than 20 minutes each! This is why they usually stick with lactate threshold and Vo2max testing rather than critical power.

[@patrickkelly885]

Do you find many outliers who cant hold CP for 30 mins plus ? and is it standard practice to test time to exhaustion at CP to test the CP for an indvidual is correct in a study?

[@all-outphysiology2177]

Generally no we/I don't test endurance at CP as it's not really how CP is defined. At least, not in the last 15-20 years. Seeing steady state responses below CP (that is, CP minus at least one 95% confidence interval) and non steady state responses above it indicates you've found the right power.