my Stryd RSS implemention seems to give small values?

[TT]

Hello,

I made some custom metrics based on the following posts but they seem to give values LOWER than if the power data was from biking.  From the physical formulas, data tables, and the Stryd blog (links in posts below), I was under the impression that RSS should be higher per Watt than TSS (Bikestress)?  Am I making some programming error?   I took the formulas from these posts, but used units of Stress instead of TSS.

https://groups.google.com/g/golden-cheetah-users/c/MgtlbMqAUqY

https://groups.google.com/g/golden-cheetah-users/c/rHWJTqjVD5g/m/OyG6mIx2AwAJ

They both roughly agree with each other, but not with my custom metric which just uses the Bikestress formula?  i.e. taking Stryd power and pretending it's bike data.  This custom TSSBikeRun metric agrees with the built-in Bikestress metric for Bike activities, but for Run activities the RSS metrics end up as 50-90% of this TSSBikeRun metric.  My runs are all one hour or longer.

Thanks for reading.

TT

[Ale Martinez]

Try this formula for RSS:

{

    relevant { isRun && Data contains "P";  }

    value { 0.03 * Duration * BikeIntensity^3.5;  }

}

BTW, using v3.6 you can download sample Stryd custom metrics from CloudDB.

[TT]

Hi Ale,  thanks for the response. 

Your formula gives the same low result as the second one here in my list.  Here are screenshots of the RSS formulas I tried, in addition to yours, just now.  At the bottom, is a Coggan PMC.  TSSBikeRun and Bikestress agree exactly.  I can click on the Bikestress symbol in the Legend to remove it and check the TSSBikeRun sticks behind.  RSS is always lower than TSSBikeRun.  I will try 3.6 after the final release and check out the downloadable version.

[Ale Martinez]

El domingo, 2 de enero de 2022 a la(s) 21:00:58 UTC-3, TT escribió:

Hi Ale,  thanks for the response. 

Your formula gives the same low result as the second one here in my list.

Yes, it is the same formula, in a more compact way.

 

  Here are screenshots of the RSS formulas I tried, in addition to yours, just now.  At the bottom, is a Coggan PMC.  TSSBikeRun and Bikestress agree exactly.

That's expected.

 

  RSS is always lower than TSSBikeRun.

If you want to use RSS the reference to use is the Stryd implementation, not BikeStress/TSS which is a metric designed for cycling.

[TT]

On Monday, January 3, 2022 at 1:30:31 AM UTC+1 Ale Martinez wrote:

That's expected.

Yes I used this as a check to ensure that my custom metric TSSBikeRun was working properly.

 

 

  RSS is always lower than TSSBikeRun.

If you want to use RSS the reference to use is the Stryd implementation, not BikeStress/TSS which is a metric designed for cycling.

Yes agreed.  The problem is that these implementations give a lower Stress than TSSBikeRun.  Even though running at threshold power for an hour is supposed to give an RSS > 100. 

Just mathematically, the RSS formulas look like they should give a higher result.  They do not, however.  Example:  in your formula, at BikeIntensity=1 the result for 1 hour (Duration=3600) is 108.

 

[TT]

Did another test.  In the User Defined Metric panel, using the Test button:

value { 100*(Duration/3600)* BikeIntensity^2; }   --> 75

value { 0.03 * Duration * BikeIntensity^3.5; }  -> 71

[TT]

Ah... I think the "problem" is that the BikeIntensity is significantly less than than 1.    My Bikeintensity was only 0.916 in the above example (I assume the test button uses the latest Activity)

[Ale Martinez]

El domingo, 2 de enero de 2022 a la(s) 21:56:28 UTC-3, TT escribió:

Ah... I think the "problem" is that the BikeIntensity is significantly less than than 1.    My Bikeintensity was only 0.916 in the above example (I assume the test button uses the latest Activity)

Unless the activity is pretty short relative intensity likely will be < 1. 

Test uses the selected activity in Activities View sidebar  

[TT]

Unless the activity is pretty short relative intensity likely will be < 1. 

Yeah I understood that part.  Just not the part where the Stryd model really has running LESS stressful per watt than biking when just a small amount under threshold.  This was unexpected for me.  In my experience, the running actually takes more recovery than biking at the same intensity.  But that might be because for me it's all up/downhill.  The legs take quite a pounding.  Downhill running is obviously not in their model.

Test uses the selected activity in Activities View sidebar  

Good to know.  Thanks Ale!

[Ale Martinez]

El lunes, 3 de enero de 2022 a la(s) 04:54:12 UTC-3, TT escribió:

Unless the activity is pretty short relative intensity likely will be < 1. 

Yeah I understood that part.  Just not the part where the Stryd model really has running LESS stressful per watt than biking when just a small amount under threshold.  This was unexpected for me.  In my experience, the running actually takes more recovery than biking at the same intensity.  But that might be because for me it's all up/downhill.  The legs take quite a pounding.  Downhill running is obviously not in their model.

There are 2 issues here:

1) I suggested to compare with RSS values computed by Stryd software because I don’t know how they apply the RSS/min for different intensities they published to the full activity (may be piece way?) and normalized power is never referenced, the full algorithm is not  public, AFAIK.

2) If you plan to compare or combine stress scores across sports you should check they are designed for that purpose, otherwise they may need adjustments, see for example the Event relative values in: https://www.triathlon.org/about/download_file/training_load_quantification_in_triathlon