I pulled together three+ year (2008-2011) nFRAA averages for all players who have played in 2011 and have at least 150 PA's during that time span. The left side of the spreadsheet just reports those totals, and a per-season rate (per 700 PA's instead of innings, since that's how BPro lists it in their spreadsheet).
I also did a very unscientific "regression" of sorts on the data. The values reported at BPro are already regressed within each season based on their performance. However, if we're trying to infer something about talent level--which is the point of a 3-year average in my view--there is even greater uncertainty than with a performance/value estimate. Therefore, I've required that players effectively have 2.5 seasons of data here before we use their straight up 3+ year average. If they have less, I regress toward average in proportion to how much data we do have. Why 2.5 seasons? It's just a wild-ass guess, but it feels about right based on the data. If you don't like it, that's ok--this is why I'm reporting all of the data. You can use whichever per-700 PA stat you wish, or come up with your own way of making an adjustment. Just please be very wary of the raw per-700 PA numbers for guys with less than a few seasons of data.
One qualifier: I was a little sloppy in that I assumed all PT occurred at the player's primary position, as listed in BPro's table. Obviously, this is not true, and for some players it will result in erroneous estimates. For most players, however, I think it's probably close enough for my purposes.
Another: I'd hope this goes without saying, but please don't just use this as the be-all end-all estimate of a player's fielding performance. I would tend to be fairly skeptical of Jose Lopez's performances, for example, given how different they are from other measures and his reputation. But it's good data, and worth considering as you evaluate a player's fielding skills.