I wouldn't equate Garcia or even Colon to Zumaya, who was basically a lock to be miss a large part of the season on the DL. As it turns out, that number is 100%. But it was going to be above 50% in any case, at least based on his past history.
GMs buy reclamation projects because your options are limited; you can't just go out and find a proven, cheap closer or corner outfielder most of the time. Why not take a shot on Zumaya for less than a million dollars?
IMO, Cherington only has Aviles and Sweeney that kind of fit the Zumaya mold. Ross is a known quantity, as are Punto and Shoppach. The rotation is a big question mark, but not because we signed an Oswalt for cheap, but because we are trying to convert one or more good relievers to starters.
I think it's true that performance in MLB has a lot of discontinuities. But that works both ways. For every Jose Bautista who suddenly figures out the game after scuffling for years, there's a Grady Sizemore who can't get it together consistently after looking like he would be all-world. I would bet that the standard deviation of performance in MLB is much higher than in the stock market.