In webinars from Trainingpeaks about WKO4 and 5 they explain that de classic levels in % of FTP works fine till FTP. So i use for Z1, Z2, Z3, Z4a(SS) and Z4(FTP) the classic zones. For Z5 VO2max i use the PDC between 8 and 3min (for me is that 106-137%). Z6 (Anaerobic Cap) is used the PDC between 3min and 10s. So thats 137%+ and for Z7(Pmax) is use the PDC of 10s till max.
There are always a lot of crossovers, for example a 2min all out will use a lot of anaerobic cap and vo2max. I believe that use PDC for finding your target for intervals, works better for me than zones. I use the zones only to check my Time in Zone for a workout and weekly targets. Thats also the reason that i dont divide the anaerobic capacity zone to FRC and Pmax/FRC
This is really interesting @david , thanks for sharing the highlights of your search for ways to estimate VO2max.
I’ve had a go at determining these for me and comparing to figures from my Garmin. I’ve shared my spreadsheet if anyone is interested (google sheets).
The comparison to grand tour champions is really straightforward and compares well for me (59 ml/kg/min vs 60 ml/kg/min from Garmin, but note if I use FTP instead of CP it gives 61 ml/kg/min).
For the ramp / step test, I “modelled” performing the tests based on my CP and W’; resulting in PPO of 423W for the ramp and 367W for the step test respectively. (The associated times on my PDC are 7 & 13 mins respectively for info.) This “modelled” test method utilising W’ & CP gives comparable results, but the step test is closer to Garmin’s estimate giving 59 ml/kg/min (-1 difference) whereas the ramp test gives 62 ml/kg/min (+2 difference).
I hope that is of some help and interest - I look forward to seeing how this develops within intervals.icu!
Nice discussion…I think there will be problems relating FTP to VO2max precisely because definitions of both are in dispute. However if you want a really quick (back of the envelope!) way of relating FTP to VO2max I wrote a little calculator for CyclingApps.net which is online here (no permission required) and also works as an app on your phone: http://bit.ly/cawvo2max . I actually aimed it at runners but cycling watts is also displayed on tab1. It has a go at demographic corrections too.
Thats a clever idea! I didn’t think of that. It nicely solves the problem of how to use the PDC for those formulas.
One Intervals.icu specific issue is that with the default FTP estimation algorithm (FFT curves + Mortons 3P) everyone on the same FTP gets the same W’. So if you want a “custom” W’ number you need to have 5s, 3-5 min and 10 to 30 min max efforts done recently and use Mortons 3P (“Options” on the /power page).
I think I understand what you are saying. General formulas like these assume everyone is the same. I don’t see any way around that, if you really want to know your own personal VO2max you need to get tested in a lab. However the the margin of error is pretty good so the results are unlikely to be way off (not like the 220-age max HR formula for example). Same thing would apply using whatever HR stuff Garmin have going for this.
It might be interesting to compare the continuous ramp VO2max estimate (short duration) vs the CP one (long duration) but using only 12m+ efforts to estimate CP. That might help answer the original question. However if you haven’t been doing VO2max work then you aren’t going to have a good “ramp test” estimate to compare with.
I am going to add a new chart on the /power page showing the W log(t) relationship for each selected power curve (42d, season etc.) with little table with MAP, T-map and Vo2max (using continuous ramp formula) for each.
I am making progress with this but as usual a lot of work will be required to handle amateur power duration curves like mine. I usually don’t have power in races (PowerTap hub in training wheel and I have carbon wheels for racing) so my power curve is “missing” durations that I never max in training.
The observation I would make is this: in the paper, measurements are reported for discrete times – eg, 4h, 3h, 2h, 1h and so on down – whereas you have presented a [nearly] continuous PD curve. Your PD curve has a lot more points at the right hand end than the paper’s PD example. So the weight of the observations at the right hand end is greater in your curve than in the paper’s. This means that your regression line is weighted more to the observations on the right than in the paper; this has the effect of making the regression line fit the right hand observations better and the left hand [short duration] observations worse than if you had fewer observations on the right. In other words, your regression line is flatter than it would be if you used the paper’s discrete times. With steeper regression lines, your middle graph would perhaps give a MAP duration of 3.5 minutes and the last graph a duration of 5 minutes. I did a rough check of this [visually estimating your data] and found the middle MASP duration to be about 3.5 minutes.
Also note that the paper starts from 2 min, not 3.
This method seems therefore to depend heavily on the choice of durations at which power is measured.
Incidentally, how do we get hold of our PD data to check this out?
Just according to the paper. I still need to do some more work e.g. no estimate if your power curve is coming up from under the PO 10m-4h line, if the PO model is really bad etc… But if you have a good curve it will work. I also need to add help etc. and other things.
I couldn’t thing of a way to pick points from the power curve instead of using the whole thing. Mine is often bad because of training artefacts i.e. I do 8m efforts a lot, don’t do longer than 14m much (longest climb near my house), hardly ever do long hard rides (cause I don’t like sweet spot!). I don’t have power in races which would fix a lot of this!
Tx. If you change to “MAP” in “Options” on the /power page it shows the PO regression line and the 95% CI line above that with a circle where your MAP is. So you can get an idea if the data is worth something or not (i.e. does your data look like the plots in the paper).
Your power curve needs to come down from above and it the dotted line to count. I think you need to do some max shorter efforts (5m or less) to get a number.
