Another issue, that Winkelman skips, is different teams have different developmental philosophies.
Phillies seem to stress skill development over performance, at least below AAA ball.
That is, they'll take pitches away from pitchers to force them to throw changeups, they'll change batting stances even if it takes a while to adjust.
I think this is more prevalent the last few years, as developing minor league players has taken precedence over "marketable" statistics.
This also impacts how fast you promote a player, which can skew some models, conservative philosophy will have older players putting up better performances on average since they're not rushed.
I think he's right that the best use of these models is as a screen to identify overlooked players worthy of a second glance - not to project performance.