My fitness signature seemed pretty accurate recently given the time of year and riding I have been doing. Went for a ride yesterday that encountered three tough, but short efforts which led to a Gold trophy. My fitness signature jumped significantly including my TP to a number that would be reasonable in June, not in November. (~350->390). Reasonable? I am having a hard time linking a good 1min CP number to a large increase in TP.
One thing to pay attention to at this time of year is drift in your power meter calibration. There is a big jump in overall power on that ride and it doesn’t seem plausible as you say. If you input your previous signature, it’s no where near what you performed on that day. I it possible for it to have gone up, but not likely as much. Also, the extraction is picking up just one point too, make the increase perhaps more dramatic. I would suggest you return the fitness signature to what it was before (or you can tinker with the data and see if there is another way to explain what happened) and monitor your data. Be sure to zero calibrate when you’re outside, not inside. If this isn’t a data anomaly, then get ready for the state road race championships next year! 
Thanks for the response.
#The calibration note is a good point. I will confirm that but I haven’t noticed the SRM drifting much
#The higher power could be due to riding with my teammates and we had a nice spirited ride. A bit different than most of my weekday workouts on the trainer
#I am surprised that a single point can raise the numbers so high without any warning or confidence interval
#I also want to note that the FTP app reported ~330W if I recall
There aren’t any warnings since by-and-large the occurrences are rare and in many cases they are caught. The big jump could have a warning and we can look to add this, if mostly for catching errors in the data that sneak past.
possibly related…I tried to build a workout today using an 180min MMP interval. That power number showed up very close to the TP. Seems like this may not be reliable for times > 60 minutes? Wondering how good, short duration numbers are throwing off longer numbers…