Don’t miss Orioles players, John Means & Paul Fry, as they guest host at our Brews and O’s event!

Baseball Prospectus makes predicting future thing of past

THE BALTIMORE SUN

Bill Pecota bounced around to three teams and just about every position on a baseball diamond. He retired after nine major league seasons with a .249 average and a whopping 22 home runs.

His is not a name you would seize on when peering into the future for strains of baseball greatness. Or so I thought.

You see, Pecota's name lives on as the acronym for a 3-year-old projection system used by the folks at Baseball Prospectus. The Prospectus is a sort of baseball think tank that cranks out books previewing each season and serves as an Internet clearing house for cutting-edge thinking about statistics, team management and such.

PECOTA, which stands for Pitcher Empirical Comparison and Optimization Test Algorithm, may be the collective's most ambitious toy. It's the brainchild of Nate Silver, who studied economics at the Nobel-rich University of Chicago.

Stat-heads have worked on systems to predict the baseball future for years. Some of these work well, especially for established major leaguers. Take the average of a guy's past three seasons with extra weight given to his most recent, and you get a decent approximation of what to expect.

But Silver and his Prospectus mates wanted better answers, especially for notoriously tricky groups like prospects and starting pitchers. So they crafted a program that looks at a player's past three seasons and then scours every baseball career since World War II for similar patterns. Silver feeds everything from a player's strikeout rate to his height and weight into the program.

PECOTA then comes up with a list of comparable players for each current major leaguer (many prospects, too) and projects the future based on the past performances of like players.

The system looks at Miguel Tejada, for example, and sees Brooks Robinson, Steve Garvey and Kirby Puckett, among others.

Silver found PECOTA to be not only accurate but effective as a tool of entertainment because it links baseball's future to real players from its past.

"I like a system around which you can tell stories," he said.

Unlike others in the projection business, Silver refuses to boil a player's prognosis to a single line of statistics. He presents a range of possibilities, from the low end to the middle to the high. Based on what we know of Brian Roberts, say, he could follow a Jim Gilliam track or he could follow a Roberto Alomar track. But he's very unlikely to follow a Joe Morgan track. Silver says this is a more realistic way of looking at players.

PECOTA is not a static tool. Silver hopes he'll be able to feed more specific injury information (databases on historical aches and pains don't exist yet) and data on platoon splits (batter and pitcher performance vs. righties and lefties) into future versions. Anything to sharpen the comparisons between past and present.

When I first gave PECOTA a serious look, I found the system off-putting because it doesn't have a very optimistic view of the passage of time. Basically, it recognizes that most players who are great will get worse and that few prospects will turn out as well as fans hope. It looks at the current major leaguers and sees very few players who are likely to be stars in 2010.

Young pitchers get hurt. Wild ones rarely develop control. Hitters who'll swing at anything keep swinging at anything. If PECOTA projected Christmas, there clearly would be no Santa Claus.

Or, as Silver put it: "Anytime you're predicting that a human being is going to change behavior, there's a good chance you'll end up disappointed."

But I've come to appreciate PECOTA's realism.

Pick up a Baseball Encyclopedia and flip back to a random year. Make a mental list of the existing stars and potential phenoms. Now, flip forward five years and see how many stars remain at the same level and how many phenoms panned out. I did this recently and left with a new appreciation of Silver's system.

So what does PECOTA think of the Orioles?

It sees productive years left in Tejada, Roberts and Melvin Mora. But it sees Javy Lopez as out of the league in 2010.

It sees significant star potential in Erik Bedard but remains skeptical that he'll ever pitch 200 innings in a season. It says there's a 5 percent chance Daniel Cabrera will become a superstar. But it says there's a far better chance that because he's failed to develop command as he nears age 25, he never will.

It sees in Nick Markakis a player who will never develop enough power to be an elite corner outfielder (though Silver offered the caveat that he likes Markakis' broad base of skills).

It looks at Adam Loewen and sees control problems so bad that he's more likely to be out of baseball in five years than to be a rotation regular.

It doubts that Hayden Penn will ever overpower major leaguers enough to be better than a middling starter.

On the bright side, it looks at McDonogh graduate Brandon Erbe's 48 strikeouts in 23 1/3 innings of Rookie-level ball and sees a guy with a 70 percent chance to be a star in five years.

It also likes outfielder Val Majewski, though it worries about his health.

Orioles officials can offer scads of reasons why the pessimistic projections are wrong. They'll probably be right in some cases.

But Silver has examined the great wash of baseball history. And he's convinced that "even 'can't miss' prospects miss more often than not."

If you want to know more about PECOTA, check out Baseballprospectus.com or the Baseball Prospectus 2006, which should reach bookstores in about a week.

childs.walker@baltsun.com

Copyright © 2019, The Baltimore Sun, a Baltimore Sun Media Group publication | Place an Ad
91°