Golden Cheetah: 2-parameter vs. 3-parameter model

Over the last three months I have performed 6 sec, 1 min, 6 min, 12 min, and 30 min tests indoors and recorded them in Golden Cheetah.

The two parameter model estimates what I think is a reasonable Cp and a W’ of 22 kJ while the 3 parameter model estimates a lower Cp but a W’ of 60 kJ (which from what I read is unusually high).

If anyone can shed light on the pros and cons of the two different models it would be much appreciated. Maybe it’s just fluke with Golden Cheetah?

The 3-minute all-out test is all you need. No guessing about the validity of models

here is a thought experiment you can try out in GC. In one of these trials/test you must have come close the the point of failure, ie not being able to continue.
For that workout then enter into the setting the CP and W’ and see how the W’ graph looks for that workout. If it is close to right then when you reach or get really close to the failure point then W’ should approach zero. If it goes below too soon then it is too low or the CP is too low, or if it does not drop far enough then CP is too high and or the W’ is too high. W’ does not start to decline until you exceed CP/FTP/TP what ever you want to call it. So if your CP is too high then it does not start to use up the W’ reserve when it should. Also if the W’ is too high your have more reserve than reality and when you reach failure it says you still have more… At least that is my understanding but I look forward to other points of view on this topic. I am curious which 3 parameters it uses for the 3 parameter model?

1 Like

scooter - taking your suggestion, W’ doesn’t get close to zero before test completion if set at 60 kJ and goes negative just before completion when set at 22 kJ. That would seem to indicate the two parameter model is more accurate (for me). From the (few) technical papers I’ve seen I can’t say I understand how the three parameter model works.

kjeldbontenbal - I may try a three minute test at some point, but still want to understand how the two models give such radically different estimates with the same data points.

maybe you have seen this but if not here is Mark describing how the critical modelling works in GC:

there may be some nuggets in there.

Thanks, scooter. I had read GC forums and faqs but not seen this. Interesting that in discussion of the 3 parameter model he emphasizes 10 - 20 minute efforts but not longer ones. In fact he demonstrates a tail-off after 30 minutes assuming people won’t do full out efforts for that long. I don’t believe that is the case for the 2 parameter model.

A video can be worth a million words.

Are you referencing the protocol where you empty yourself as fast as possible and the power you can hold at the end of three minutes is CP?

Yes, that is the one

1 Like