Projo Fantasy Sports Blog

By Michael Salfino

What 2008 shares with every other season, to the chagrin of fans and especially the managers and general managers they hound, is the seemingly sudden and dramatic performance decline of numerous stars.

Baseball provides a great chance to understand randomness. There's an interesting and lively book out right now, "The Drunkard's Walk: How Randomness Rules Our Lives," by Leonard Mlodinow, that gets into this in much more detail. But I'm going to do my best here to explain how randomness relates to baseball and, especially, our task of predicting player performance.

Mlodinow cites the Roman statesman Cicero for his belief "that an event could be anticipated and predicted even though its occurrence would be a result of blind chance." I note that 2,000 years later, Yogi Berra famously said, "It's tough to make predictions, especially about the future."

That's the fine line we walk: recognizing there's a large element of randomness in all player performance while still projecting as narrow a range as possible going forward, assuming health and continued opportunity.

Consider Carlos Pena of the Rays, who homered once every 19 at-bats before last year, when he went yard once every 10.6 at-bats. Could this dramatic power boost have been a completely random event that Pena was just lucky enough to have happen to him?

His season was similar to Roger Maris' in 1961, when he went from a guy with good power to one who broke Babe Ruth's single-season homer record. Mlodinow, a physicist who teaches about randomness at Caltech, notes that Maris, before that 61-homer season, homered about once every 15 at-bats. If you had a Strat-O-Matic card of Maris that resulted in a homer 1/15th of the time and gave him 590 at-bats, and then played out seasons over and over, Mlodinow says you'd get 61 homers one time out of every 32 tries. He suggests that Maris had less to do with his 1961 season than did Lady Luck.

The lesson is to not be seduced by such random, outlying performance and expect it to repeat. So this year, we should have expected Pena to homer every 19 or so at-bats (his pre-2007 average). And in 2008, prior to his injury, Pena did exactly that (once ever 18.8 at-bats).

This is called "regression to the mean." In other words, the default projection is to expect a player to do what he's done before. That also holds true for guys like those below who are having terrible years. You just as easily "progress" to the mean assuming -- and this is the big assumption -- you get the same level of opportunities. Often, slumping hitters get benched and thus lose the chance to prove that their bad performance was just a random stretch of bad luck.

Be careful not to apply these principles of randomness blindly. Age, injury and lack of a suitable big-league track record can provide a reasonable basis for us to assume that current performance levels are less random and thus more sustainable. But beware of our human tendency to see patterns where none exist.

Now let's make some related recommendations.

Buy

Adrian Beltre, 3B, Mariners: His line-drive rate is higher than it's ever been: 21.1 percent. But he's hitting .236 on balls in play (.294 in '06 and '07). His .233 average is incredibly unlucky and will correct.

Carl Crawford, OF, Rays: Like Beltre, a higher line-drive rate and career-low average on balls in play doesn't make sense. Expect Crawford to hit well over .300 going forward.

Andruw Jones, OF, Dodgers: He's due to return to the active roster around the All-Star Break. We have to call a bottom on Jones. Blame the bum knee. Yes, he's fat, but the more relevant weight he's been dragging is all those random factors working against him.

Roy Oswalt, P, Astros: He's giving up homers on 20.6 percent of fly balls, twice the average rate. Yet his strikeout rate is up, arguing against a loss of stuff. Expect a major resurgence.

Hold

Derek Jeter, SS, Yankees: Still an extreme ground-ball hitter, but now without power (homers on just 7.1 percent of fly balls). At age 34, it's not coming back. The line-drive rate (16 percent) is also subpar, but you can be more bullish about that (and thus his average).

Howie Kendrick, 2B, Angels: Expectations remain too high. Even when healthy, which is rare, Kendrick is a high-average hitter with below-average speed and power. Steeply discount the minor-league slugging, which came in hitter's parks against weaker competion.

Sell

Francisco Liriano, P, Twins: Few discounted the injury risk. Sometimes the fastball doesn't come back after Tommy John surgery and it hasn't yet even in the minors for Liriano.

Fausto Carmona, P, Indians: If he wasn't lucky in giving up homers on only 3.2 percent of fly balls, the results would be far worse. Walking twice as many as he Ks, Carmona has the profile of a guy hiding an injury.

---