data requirements for different estimators?

[Brett Dikeman]

I looked through the wiki and searched the mailing list archives, but I think I'm hampered by not knowing what exactly to search for.

Is there a wiki entry or good thread discussing what sort of workout you need to provide GC with in order to get W', CP, FTP, etc values?

(probably?) related problem: I have a Trends graph with W', CP, FTP estimate lines. Two recent activities are not appearing despite having estimates in the metrics page. In Data Table view, changing between "last 7" / 14/ 21/ 28 days seems to change dates completely at random, and I don't see data from a ride 2 days ago, or today.

What am I doing wrong? (and again, any recommended links would be awesome.)

Thanks!

Brett

[Nathan Townsend]

Having done lots of CP testing, and knowing the scientific literature well on this topic, IMO this is the best practical CP + W' protocol:

Day A:  15-16 min TT -> 30-40 min recovery -> 2-3 min TT

Day B: 10-12 min TT -> 30-40 min recovery -> 6-7 min TT

All of these TTs should be paced according to the following guidelines:

1. Use a fast start

2. Do not pace conservatively ie: adopt a positive split pacing strategy

3. Try to keep the power always above your estimated CP through the middle (this mainly applies to the longer duration tests)

4. Towards the end, hitting the exact target duration is not that important.  What IS important is that you are giving a full on 100% maximal effort attempting generate as high a power output as possible, but the power should not be far above estimated CP (50 W as a rough guide).  So for example, if you reach your target duration and the power is 500 w (because you paced too conservatively and you have a big "end spurt"), but you estimate your CP is somewhere around 280-320 W, then KEEP GOING with maximal effort until the power drops to about 350 W or so.

The reason for point 4 is that a lot of people just don't expend all of their W' when doing these TTs and that can introduce error into the estimates.  The closer you get to exactly the same state of fatigue in each TT duration, the better the estimate of CP and W' should be. 

I use either a simple excel spreadsheet (see attached for linear models), or a scientific graphical software to calculate CP + W' because GC only allows two data points to be entered for the linear model.  It's better to have 4 good quality data points if you want a more confident estimate.

What I tend to avoid is the extended CP model because to my knowledge this has not been validated physiologically ie: the CP estimates are validated by conducting constant load tests slightly above and below CP, and then taking measurements of physiological variables, most importantly VO2 amongst others.  

 I don't even know exactly how CP is calculated for the extended model, but on average it seems to report a slightly lower CP and correspondingly higher W' than the 2p models.

[Brett Dikeman]

Hey Nathan - thanks for the concise advice on the testing protocols! That was a great explanation.

That said: I was hoping more for a description of what the various GC estimators need (and why I'm not seeing data in the trend graphs for a lot of activities.)

I really wish the documentation, either on the wiki or (preferably) in GC was better for the metrics, estimates, and models. For example, there should be an explanation in-program as to how and why to use the different models. The vast majority of people are not going to have much idea what model to use...

--
_______________________________________________
Golden-Cheetah-Users mailing list
golden-che...@googlegroups.com
http://groups.google.com/group/golden-cheetah-users?hl=en
---
You received this message because you are subscribed to the Google Groups "golden-cheetah-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to golden-cheetah-u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ruud Goorden]

What about that new model from mike p? I have seen he has an new extended model. I think it is based on the powerdata from the field which was asked about 1.5yrs ago. I would like to know if this can also be easily used / if the paper can be shared. If this was the one for which field data was requested I actually gave my data.

[mike veloclinic]

Hi Ruud,

Thank you for participating in the study. I just presented the validation study results at the AMSSM conference earlier this week. I am currently writing up the full paper. The final model I chose is a little different then those in GC in order to keep the model fully consistent with the original 2P CP within the CP valid domain. I'm not sure if Mark will want to implement the model directly in GC or have people just use R if they would like. One of the great parts of integrating R into GC is that freedom for people to innovate with some very powerful statistical/mathematical modeling/learning packages.

mike

[Ruud Goorden]

Hi Mike. OK Tnx. Would be OK if I can use it in R. I see Mark is doing some good work on integrating R in GC. Nice times coming for doing some own experimental stuff