While they may do a good job of looking at trends in general, there will always be outliers and exceptions when looking at a sample of 1. For example. Carl Crawford did not overachieve. His OPS was higher than predicted because he is FAST...Super-Duper fast. Something that is not considered when calculating prOPS. Crawford can turn singles into doubles and doubles into triples and often turns outs into hits. All of these things will increase his SLG which increases his OPS. and because he is fast, his actual OPS is higher than predicted based only on where the ball was hit. That does not mean he "overachieved". It just means he took advantage of his speed.

Who knows what factor is influencing Pena's numbers. But for anybody to think he underachieved in 2007 is ludicrous.

I looked at the Rays data, then looked at the Yankees. What immediately struck me was the difference in accuracy of the 2 data sets. More of the Yanks predictions were more on target. I take this as an indication that a predicitive tool using past performence as a guide is likely to be more accurately predictive when there's a larger pool of data. Note that most of the largest outliers in the data sets are among players with limited stats actual stats in 2007, particularly on the downside.

Strangely, the upside performers more tend to be players with regular play in '07 and deeper history. The Yanks outliers were Cano, Jeter, and Posada (difference + of .05 or more), the Rays include Upton, Crawford and Iwamura. What really struck me in the case of Crawford and Jeter were low prOBP's compared to their histories.

I guess my point is this - predictions / projections are just that. Performance is was matters. Saying a player was lucky / unlucky, underperformed / overperformed gives way too much credence to the predictive tool. In baseball, it's reality that matters.

Let's consider Upton and DY to show another flaw in this stat and other stats that look at every batted ball on a chart.

Upton and DY generate an enormous amount of bat speed. More bat speed means the ball is hit harder. In a lot cases that means the ball travels farther. That can be calculated by looking at things like home runs (a ball hit very hard).

but what about ground balls?  Let's say Upton hits a groundball 6 feet to the left of the shortstop. Now compare that to a groundball hit to the same spot by  Josh Paul. Upton's groundball is likely to be travelling MUCH faster. there is a better chance of that ball getting through the infield for a single. Paul's groundball is likely to be fielded. On top of that add Upton's speed and even if the ball is fielded (moving towards the hole), there is a good chance he beats it out. Not so for Josh Paul.

SO even though each batter hit the ball to the exact same spot on the field and in each case it was a ground ball (in influencing factor in this formula) Upton and DY get on first base with a single. Josh Paul is out by 5 feet. Was Upton or DY lucky? Did they "overachieve"? Or were they just more skilled than Josh Paul?

When looking at a number like prOPS, it can help predict an increase (or decrease) in OPS to the next season. It works about 75-80% of the time. So it is a good indicator. But there are still 20-25% it does not work for. Some of those players, and where these stats fail, are the highly skilled players. They are not lucky. They just have the ability to get hits on balls others would not get a hit on. That is why they have higher OPSs than "predicted".

http://mlbfleecefactor.com/09-free-agents/