From a previous post by Mike Bravard:

"GB/FB weaves its way through a lot of this game because it is a KEY stat. For example a pitching staff's GB/FB ratio will have a dramatic effect on player's range stats in the outfield and infield."

Perhaps this accounts for the difference. Maybe by trying to introduce too many factors PB sometimes ends up with just "noise".

I know Mike is proud of all the factors PB takes into account, but at times it seems to result in unrealistic results.

It's like some of those prognosticators who'll come up with "well, Sosa has batted .567 between the hours of 4PM and 4:30PM in the month of June over his career." Is that meaningful?

Personally, I think platooning is overly encouraged by the game...especially because the sample size of the player stats is often weak. For instance, the first year up Chris Singleton was a killer to lefty pitching. Since then, he can't hit them a lick. Probably he never could but just got "lucky". What happens when a righty pitcher who has a great record against lefty hitters goes up against a lefty hitter who has a great average against righty pitchers? Is this logic or just chance? Possibly if CAREER numbers were used for guys with a LONG history then the righty lefty numbers could be meaningful. Right now it just encourages platooning to extreme. I know Mike thinks the major league managers don't platoon enough, but maybe they know what they are doing.

Boy...did I get off on a tangent!!