Reconciling vastly different PD models

Hi everyone,

I was hoping someone might be able to provide feedback/guidance on how I might reconcile three very different power-duration curves generated from my data by Golden Cheetah, WKO5, and Xert.

Maximal or close to maximal efforts over the past 90 days (based on what Golden Cheetah is selecting for its CP calculation). All but the last were completed very recently (last week) as I was purposely trying to feed the model.

  • 1 sec - 942 watts
  • 6 sec - 920 watts
  • 1 min 30 sec - 493 watts
  • 5 min 24 sec - 326 watts
  • 14 min 30 sec - 269 watts (not ideally paced)
  • 31 min - 246 watts (three weeks old)

Models

Golden Cheetah
CP = 235; W’ = 28.9 KJ

Xert
TP = 254; HIE = 23KJ

WKO5
TP = 272; FRC = 13.2 KJ; TTE = 33:20

Xert’s 30 min power estimate (266) is not much different from WKO’s model of threshold (272, which is says I can only hold for 33:20), so those models might be closer than they first appear. However, the modeling of the glycolytic contribution is very different.

One final piece of data: when I did an INSCYD test done with @steveneal two years ago, my Vlamax was modeled at .57 mmol/l/s. This, combined with “feel” on longer efforts makes me think threshold is closest to the Xert number, though I don’t know that I can hold that power for 60 minutes.

Any ideas other than HTFU and doing a 60 min TT?

Thanks!

1 Like

For fun you could see how long you can hold:

235 watts

254 watts

272 watts

If I was your coach and you asked me that question :slight_smile: I would ask you to answer it for me!

Looking forward to your findings.

Steve

2 Likes

Hi @steveneal, thanks for responding! This seems like the obvious solution now that you have suggested it. I will start tomorrow or Wednesday, depending on how my legs feel. I think I’ll take two easy days between efforts and start with 235. I’m wondering if WKO is inflated because I don’t have any longer hard efforts in the last 90 days. I’ll keep you posted.

1 Like

@fazel1010

Awesome and good on you!

There is more to the model than the last 90 days…especially if you are trying to plan the next 90!

So see where you BODY and MIND is at for these durations.

Be well fuelled … and I agree with a few days between each

Talk soon.

Steve

1 Like

What did you find with your duration tests?

I never actually did longer duration tests. CP seems to be a better model for me and the type of training/testing that I do. Xert might be the happy medium, but I believe it’s pretty closely tied to CP as well.

WKO uses MMP and therefore averages efforts. This means maximal efforts - if few in number - live above the PD curve (my situation).

CP tends to handle this much better (from Golden Cheetah).

Obviously, the models are very different and basically paint me as two very different athletes. WKO says threshold is around 4 w/kg and FRC is relatively low. Golden Cheetah says threshold is significantly lower (and this is closer to being right; best 30 min power is around 250; best 60 min is around 240) but that I have a very high anaerobic capacity.

I have data from three of my buddies in WKO as well - and the model works best for a guy who Zwift races a lot. See how most of the area under the curve is filled in? (Yes, he is strong AF…)

That is to say - if you are going to model power using a couple maximal efforts - it appears CP is much better. If you are racing, I don’t know that it matters much as you’re going to have a load of maximal efforts from one-to 20 or 30 minutes so it’s likely the models will be comparable.

If you have WKO the PD curve page has normalized residuals and test targets to help fill in the curve, so that’s a way to keep the model updated.

At least this is where I’ve ended up.

1 Like

Here’s a huge discussion / argument over the models.

https://forum.slowtwitch.com/forum/Slowtwitch_Forums_C1/Triathlon_Forum_F1/WKO4's_Power-Duration_Curve_Model_Fundamentally_Predisposed_to_Underestimate_Power_Output%3F_P6476332/

Andrew Coggan is the guy who generated the model for WKO. Liversedge (Mark) heads up Golden Cheetah’s development. They both participate.

CP modeling seems to like those short efforts. I look at WKO5 and Intervals.icu. Intervals.icu uses CP and whenever I do a max 3-5 minute effort Intervals will tell me I have a new FTP.

The other day I did a 3 minute max effort to fill out the WKO model. It was a new PR for me. Intervals.icu said ‘well done, your ftp is now 256’. WKO5 bumped my mFTP from 226 to 229.

When I do a longer FTP test (I like the Kolie Moore baseline test (about 45min all in)), WKO5 will give me an mFTP that is spot on. It’s a really good test for me mentally as it’s a ramping test. You start out in sweet spot, roll through threshold, and finally get to the ‘no way can I hold this pace for an hour’ wattage. I think this test helps you get a great feel for your threshold.

Yeah - I prefer Moore’s longer test as well. As I said, I’ve done several 60’ efforts in the last two years and I feel like finding the line is much easier if that makes any sense. This is what Liversedge says in the thread I referenced.

The GC model is based upon the hydraulic model of Morton and Magaria. In small words it is based upon peer reviewed science and is rather more complex than other models in GC…selecting peak values from 5s, 20s-90s, 2-5 mins, 10-50mins and 1hr+

In intervals ICU you have some settings that you can mess with to change the version of CP model used, and at what duration to start looking for maximal efforts. I think the default is a single maximal effort 3’ or longer, which is likely why your effort resulted in such a skewed estimate of FTP.

Long time WKO user (~15+ years), used GC on & off for about 10 years. Played around with Xert for a couple of years. Just started playing around with Intervals.icu

The PD curve in WKO uses OLS (Ordinary Least Squares), in part, to model the athletes PDC. The notion being that the PDC should also represent repeatability. GC fits to the peaks using, as noted above, one of a couple CP models. I don’t know about Xert.

I don’t have a dog in this fight. As noted in the link to an old Slow Twitch thread, there’s plenty of debate about which is “better”. I like WKO as I’m more familiar with it and it makes more sense to me. The PDC in WKO is also used for much more than targetting interval intensity/duration.

It’s great that we all have this selection of tools for training. Just pick one that works for you and go with it.

Dave K

1 Like