Can you plot Beltran and Cameron on the same graph so it's easier to compare them?

With graphs like these, it's going to be hard to compare two players directly because each player has a unique "expected" curve. If you could find a way to transform a player's "expected" curve into the league-average curve, you could apply the same transformation to his "actual" curve in order to pretend that every fielder had the same opportunities. Then you could overlay multiple player graphs that were in the same "scale" to compare players much more easily.

Sky, I agree with what you are saying, but then the labeling will be wrong. It will really be "actual adjusted", which sounds weird. I think I prefer seeing it as David is laying it out.

What would help more is that:
1 - Force the y-axis from 0 to 1 in all graphs, rather than letting Excel trying to fit the data

2 - Slide the "vectors" over, so that the 0 on the x-axis is straight-away center

That dip in Cameron's chart, read in combination with the linked blog-item on anticipating rather than tracking, is very telling. It illustrates a consequence of physics related to side-spin and stability.

Balls hit to the right and left fields, or even to the right and left parts of center field always have some side-spin (this is a result of the ball not striking the bat perpendicular). This side spin causes the path of the ball to be stable; the lateral 'curving' force on the ball is constant over the flight of the ball, and thus predictable. Balls hit to dead center sometimes have negligable spin on them, and thus behave erratically. The same aerodynamic phenomenon is responsible for knuckle ball behaviour.

A fielder who tracks the ball in flight continuously is able to adjust to the unpredictable flight paths of a spin-less ball and keep his percentage up.

According to the linked post, Cameron judges the trajectory early and goes to the expected landing point without watching, so if the ball starts to deviate, he does not see and adjust. That dip in his chart is for those hits to dead center, those likeliest to have no spin and thus unpredictable trajectory.

My $0.02 worth.
David

I'm a Mets fan and see about half the games on TV. Several times last season, I saw Cameron get caught trying to come in on balls hit directly at him, then realize they were hit harder than he had thought at first and have to reverse and go back. The ones he didn't catch were extra bases, of course; I think I remember several triples. It's weird in a way, because on those, he can watch the ball, since he's facing it as he runs in. Then again, as a bad outfielder in amateur softball, I've made the same mistake too many times to remember :) Line drives look like they should start sinking, but then they stay up because they're hit harder than I realized. I guess Cameron's so sure of his instincts that he's less cautious than other centerfielders on those balls. He doesn't have that one-beat hesitation to double-check that a lot of outfielders have, he instead starts coming in right away.

This is a great idea. You might try graphing the difference, (actual - predicted), so that an average performance shows up as zero, above-average as positive values, and below-average as negative values. It's hard to make out the small differences between the actual and predicted values, relative to the large variation in actual and predicted values across different x-values.
Also, look at the Beltran-Medium Fly Balls graph. The actual and predicted curves look almost identical to the eye. But this is misleading. Beltran is actually quite a bit below average here. The eye tends to see the perpendicular distance between the two curves (i.e., the shortest distance between the curves), but what really counts is the vertical distance between the curves at each x-value. If you look at the vertical distance at x=5 and x=4, for instance, you can see that the red point is substantially above the blue point. Plotting the differences would make this more clear.