I came to realize that the powermeter I was using (Watteam) was giving incorrect results (about 12% high) and my trainer KICKR was also giving me incorrect results (about 10% low). The Watteam folks reinstalled the sensor and that fixed that. I also discovered, after a journey through Wahoo customer support, that there was a secret calibration option for the KICKR and that fixed that. In the meantime I also added the new Vector3 pedals (which have been great out of the box). Now, I have three power meters that all agree almost perfectly. Unfortunately, this means that I had a lot of records that were incorrect. The outside records overstated my output and the trainer records understated them. So, I decided to just delete all those old records and just retained the records that used the calibrated power meters. These retained records start in mid-December. I decided to seed the progression by putting a new record in at Dec 1 and setting the PP to 500W, the HIE to 6.0 and TP to 200. These should all be under my actual values. At this point I have 17 additional (actual) records and the progression seems to be giving me reasonable results for my last couple of rides but I don’t see any stars (but the Training Pacer is giving me a Freshness rating). So, the question is: are there some number of minimum data points before the stars appear?
Check the glossary on Training Status. Also, it might be good for you to replace your deleted activities with manual entries, to ensure you have a properly reflected XPMC. See the FAQ for more details on them.
OK, so if I have this correct, the stars value is based on an exponentially weighted moving average of XSS across 42 days (I guess this value is from physiological research) using a proprietary “decay” coefficient, correct? So, when I see TL exceed 25 I will see the first star(s). Do I have that right? I was assuming that stars had something to do with “converging” on a reasonable signature (which the algorigthm has done) but really it’s all about load. Freshness is related, I suppose, and is based upon the currency of the load. Starting to make sense.
Oh, and you have recently added the ability to set the “decay” coefficient (indirectly through the drop down list in Account Settings | Profile). Starting to make sense…
The time constants correspond to how your training load increases/decreases. The Signature Decay Method describes how your signature parameters decline over time without breakthroughs. A blog is being written to describe it. Best to leave at Original for now.
If you’ve got the .fit files from your deleted activities it might be possible, if a little time consuming, to adjust the power values using https://www.fitfiletools.com/#/top , and then reload them.
Thanks Cary but I think I will just ride more…