I love prediction season. Right now, every sports media outlet in the country is running endless NCAA brackets, bracket-picking advice, and studies of past bracket upset patterns–and while I’ll take it, I’m still scouring baseball pages to see what writer was foolish enough to put his name to the fortunes of only 30 teams, predicting the outcome of the 2003 baseball season. We do it every year here at Prospectus, and getting my predictions is like trying to get me out of the bar before I’ve finished my beer.
I’ve done pretty well at this the last couple years, despite my yearly gut feeling that this is going to be the time I really, truly embarrass myself. I tend to be boring, and do a boring little estimation for each team where I guess best case (everyone’s healthy, except for bad veterans, who have painless season-ending injuries, allowing cool rookies to have blockbuster seasons), normal case, and worst case (no one is healthy except for bad veterans), and then put ’em through the Riskotron 2000 to get a final number and compare that to their division. I try not to wish-cast, though that’s still not necessarily a guarantee for success.
This reduction of possible outcomes to a final number is interesting to me because it is the same issue PECOTA projections aim to address. Diamond Mind Baseball‘s pre-season predictions do something similar; they run many seasons and then base their predictions on the cumulative outcome. But you see strange things in there: a 2.5% chance a cellar dweller captures a Wild Card, for instance. And that’s where, I think, pre-season predictions encounter their first problem. Just as there’s fame and fortune to be made for correctly predicting that Unknown Player X will hit 90 home runs and become the MVP, many writers try to peg the surprise and collapse teams so they can say: “Hey, I knew the Angels would win the World Series when no one else did!” It’s hard to make predictions you know are going to be just like everyone else’s. More than that, it’s not all that fun.
One of our Pizza Feed attendees had a clear understanding of this, last week in Portland. He turned in a sheet with some wild picks on it, and when I started to razz him about it, he said that it wasn’t necessarily that he’d really thought the Orioles would finish ahead of the Blue Jays, but that he figured while many people would turn in essentially the same predictions that would all be equally wrong when the season ended, and the winner of the T-shirt would be the guy who took some risks that paid off.
Case in point: When the Diamondbacks won it all in 2001, people combed the Web looking for anyone who’d gotten that right, and the only person I know who even came close was D-Backs ex-manager Buck Showalter. For a couple of days, he was gold. He could have started a sports betting service and Pete Rose would have been called up in under a minute. You see the same thing happen after Super Bowls.
The other problem is a depth-versus-width problem. Each BP book author writes at least one team chapter, and most cover a couple. Each author knows his teams in-depth. Many know at least a couple of others as well. But only the Chris Kahrls of the world know what’s going on deep on the bench of all 30 teams, or the heated competition to be Ottawa’s set-up man this year.
One of my favorite columns of all time is a piece Rob Neyer wrote on ESPN.com in which he talked about discarding high school players entirely when scouting for the draft, and concentrating solely on college players. He cited the conclusion of a study Bill James did, concluding: “At all levels of the study a little bit of knowledge in depth proved to be of more value than a great deal of knowledge in breadth.”
That column, like many fine pieces of writing, gave me words and analogies for something I’d never been able to clearly state myself. Through it, I started to realize that the concepts discussed were more widely applicable than in just the draft, or in scouting. When you’re investing money, for instance, you’re much better off having in-depth knowledge of a few companies and finding one quality investment candidate than you are knowing, say, the names and stock symbols of every NASDAQ-listed company across the board.
When it comes to the teams I covered in BP2003, I have a great idea of how this season could play out. I know how well-stocked they are for pitching if one or two starters go down; I know which position players they can lose for a couple of weeks, and which they don’t have replacements for, resulting in an 8-20 swoon. I have ideas on who in the minor leagues could be ready for a mid-season call-up, and I can talk your ear off about the team’s top 10, 15, or whatever prospects.
Want to see me look like an idiot, though? Ask me who’s in Pittsburgh’s rotation. It would take me a while to get to the back end if I’m not in front of a computer.
Of course, all of this is a result of where I direct my efforts. I spend all year paying attention to my teams, going to and watching their games, taking notes, talking to anyone I can corner at the park, getting thrown out of minor league press boxes (Uh, ignore that), and reading anything I can find printed about the organization. As a result, my baseball knowledge is massively lopsided, stacked along the pyramid in the following manner:
- Teams I’m writing about for next year’s annual, including their minor-league teams (Know tons)
- Teams in the same division as teams I’m covering (Know well)
- Teams I like but am not covering (Know decently)
- Other teams (Bad–comparable to a particularly forgetful fan in that city) “B.J. Surhoff is our DH? What?”)
This is why I put off doing predictions. While I can easily come up with a good best/normal/worst case scenario for each team I know, I’m equally bad at doing the same thing for teams I don’t know.
In the same way, I think the smart baseball writers in each city are probably the best predictors of their teams’ fortunes. They’re going to be overly optimistic in print, naturally, especially if their city has a crush on the team.
But I would bet that if we were to survey good baseball heads for each team and ask them to predict only their team’s final record–either straight up or by asking them questions to establish a range of outcomes–you could put them all together (and normalize for optimism if you have to) and come up with a picture of what the season looks like that would be better than any single writer’s idea.
But hey, if nothing else, each person’s predictions offer an insight into how he’s approached the project and how he sees the league. Mine, for example, seem to indicate that I’ve spent too much time doing risk-mitigation work. It sure beats getting thrown out of the press box, though, so what are you going to do?