No way that is true, deviation never increases as you play
My deviation just went from 60 to 62 after a win just now. The range from your rating to the numbers it gives you is one standard deviation, which should predict your skill 68% of the time. So if you have a predictable win/loss ratio based off of a sliding scale, your deviation will be small. However, if you do something that is basically anomalous, like losing five times in a row, that falls well outside the predicted range. To keep up whatever confidence interval Shoddy uses (95%, I'd assume), it has to adapt to your record. For example, Shoddy is 95% (or however much) confident that, say I'd win at least two out of five matches against someone with a given rating. But if I lose four of the five, the uncertainty inherent in the calculations rises - to take into account my sudden failure at Pokémon, it might adjust itself to say that 95% of the time, I'd win 1.5 out of five or whatever matches against the same person.
I think I suck at explaining it, but the TL;DR is that a small deviation means you're consistent. I was inconsistent, so it shot up.
[/threadjack]
Anyways, I think that while the metagame has adapted somewhat to the presence of Cresselia, Porygon-Z just keeps getting more dangerous as people really figure out how to use it. I've had a couple matches recently where I was absolutely devastated by a scarfed set hitting and running. Boltbeam + Tri Attack (+trick/hp fighting) is a stellar moveset, and even a Timid P-Z is fast enough with the scarf to outrun and OHKO Chlorophyll Tangrowth and Exeggutor. It's reminiscient of a scarfed Typhlosion, except it has perfect (I think) coverage and can OHKO back after taking a priority move to the face, unlike Typhlosion, who peters out and has to resort to Overheat or taking one for the team. Combine that with its great abilities, and P-Z is looking very much like BL material right now.