AI FTP Prediction vs Detection

How have people found the alignment (or not!) between FTP Detection and FTP Prediction? Do they gradually converge on detection day? Or can they be wildly different?

My current prediction is a 1w drop, but I’m hoping that is wrong (see my 7w drop thread!) and the detection will show a few watts increase!

AI FTP for me I think predicts a little too high but the prediction when I was doing almost exclusively TR workouts seemed fairly constant other than the 1w fluctuation and the final detection agreed with it. Now more of my riding is in groups outside or roadbike TT’s (and I’m missing TR workouts for them), its naturally taking a bit longer to converge and the prediction circa 5 days out agrees with the detection.

In almost all the cases I have had so far, it predicts something that seems too high… it drops then slowly over my training window a little or a lot (sometimes saying no gain or even loss…) then as I approach detection date… it creeps back some, eventually being pretty close to prediction.

Which scares me cuz it said on new prediction… 350w (from my 336 PR, lol) it’s already now saying 344 (last night) after just 6days, I bailed on morning run and now it says 345 at 1wk from prediction start.

I wouldn’t mind it just getting to just 340w, yesterdays vo2 6x2.5’ was like WTF… why 400w! lol :scream:

But… my guess if I stick to training, outperform my workouts (WL higher than starting WL)… take it easy on my running… it will drift back to 345-350 in a few weeks (heck one of the weeks is a recovery week…easy!!) So just have to nail tomorrows over unders… rest week… then a hard week, plus one more vo2 workout and I’d be around 350!!! seems too fast/high.
But–sets me up for some good endurance and tempo work now that its nice outside on the weekends (at higher FTP)!!!