ATL CTL equations?

[marcovelo125]

Does anyone know what specific equations the (online) Performance Manager at TrainingPeaks uses for ATL/CTL? (Or is that proprietary math?)

I've been using a couple of versions of the equations in an Excel spreadsheet which have been close to the online Performance Manager, but I'm starting to get diverging values. The difference between the CTL values is approaching more than 10% of current CTL and I would like to figure out what's going on before the models diverge even more.

Anyway, if anyone knows the specific formulas, I would be grateful to have them,

Thanks,

marco

 

[daveryanwyoming]

I don't work for TrainingPeaks nor have I seen their source code but FWIW here's the Excel formulas I use for ATL and CTL and get good agreement with WKO+ but I only use the spreadsheet to plan tapering or short term CTL builds so haven't looked out more than a few weeks with it.

CTL: =ROUND(C2+(B3-C2)*(1-EXP(-1/$C$34)),1)
ATL: =ROUND(D2+(B3-D2)*(1-EXP(-1/$C$35)),1)

Where:
C2 = Cell containing previous day's CTL
B3 = Cell containing today's TSS
C34 = Cell containing CTL time constant in days (default 42)
D2 = Cell containing previous day's ATL
C35 = Cell containing ATL time constant in days (default 7)

Maybe this diverges from WKO+ over longer time frames, I can't say but it follows all the original TSTWKT discussions from the wattage group where the PMC concept was hashed out based on Bannister's impulse model prior to the WKO+ implementation.

-Dave