Computer projection systems are a useful tool in baseball analysis, but like any method of predicting the future, they're not flawless soothsayers. Statistics in baseball capture a lot of information about the players, and projection systems are good at wading through that information.
But just like scouts who make evaluations with their eyes, projection systems will be wrong. A lot.
One of my favorite preseason exercises is to run my way through the ZiPS projections and see which projections I find to be surprisingly high and low and hope that I can "beat" the computer. Sometimes, I win, such as in the case of Matt Holliday, whom the ZiPS SuperComputer seems to have a secret vendetta against.
Other times, the computer wins. For example, ZiPS was down on Josh Hamilton far more than most people (and all the other projection systems) going into last season. ZiPS was right.
Which projections do I find surprisingly low this year? Here are the 10.
David Wright, 3B | New York Mets
ZiPS projection: .276/.358/.467, 19 HR, 74 RBI, 72 R, 17 SB
Wright has been safely above this projection over the past two seasons, putting up a .307/.391/.501 combined line. While the homer projection seems to be in the ballpark, I think the batting average is too low. His .254 average in 2011 is more of an outlier than usual for players having bad seasons, caused not only by injury, but also by his pronounced uppercut swing that became harder to get results from both from age and the nagging physical issues.
Wright is hitting a lot more line drives now, an approach that's better suited in tough home park.