Some power data does come from Powercal, but most from a Cycleops PT300 Pro stationary and some from a Powertap SL hub equipped wheel which does both trainer and road duty. The Powercal is not accurate for me at all, consistenly underreporting, but I don't know if it's prone to wildly high spikes. I've looked at the raw data a few times and don't recall any glaring issues, so I don't know if that's necessarily the source of the outlier data point (again assuming the straight line from 1600w up to 2300w represents the anomoly), but it could be. Since it relies on HR monitoring to calculate power, it's certainly very bad at short intervals where there's an HR lag behind the effort.

I don't agree my curve looks odd, however. It is very consistent because the bulk of the power data comes from programmed training routines on the cycleops PT300 Pro bike, almost none of which is maximal effort training. As I said before, I rarely ride the Powertap on the road (mostly trainer duty, but some winter/spring rides) both because it's not my preferred wheelset and because it's an old wired unit that does not integrate with GPS data, which is far more interesting to me for road rides since I do my power targeted training on the stationary (and some on Powertap+Fluid 2 trainer) anyway.