Fangraphs' WAR vs. Smith's WAR for pitchers

[STLCards2]

Since Fangraphs uses FIP from 2002-2009 to calculate WAR, and Smith uses some variation of RA adjusted for park and defense to calculate WAR, is it safe to say that Fangraphs' WAR is probably better to use on a seasonal level, but Smith's is better to use for "career" (or at least over a big chunk of 2002-2009), for those who apply?

It seems like 5-9 years is enough to establish somewhat reliable BABIP ability, significant situational split "ability, significant running game prevention, significant XBH reduction, and significant DP inducing. If those skills are somewhat "true" after 5-6 years, it seems Smith's WAR (which includes those things), would be better to use.

I guess the question is, for how many seasons is Fangraphs' WAR more reliable than Smith's WAR? 2? 3? When should one start looking at say Lincecum's WAR on Smith's database as opposed to Fangraphs?

[Tango Tiger]

Excellent question, and well-considered.

This is a great example of what works at say the seasonal (or even game level... like ignoring non-HR hits, or heavily discounting them) may not be so good at the career level.

To get a good answer to a great question, you need to do some careful studies. Until then, like most things, take a bit of both, and you'll be close enough.