So Many Soothsayers, So Little Time
February 20, 2007
If you are a huge baseball fan, chances are you also love the stats the accompany the great sport. It's also a likely assumption that you are involved in some kind of fantasy baseball game.
With the advent of fantasy baseball, many publications love to issue projections for the coming year, all purporting to use complicated formulas to deliver the most accurate information about future events. While I enjoy looking at these projections as much as anyone, their accuracy largely goes unchecked once the season begins. It's like weathermen (excuse me, METEOROLOGISTS): they throw out some numbers for tomorrow's weather, but when they are 10 degrees off and you are carrying an umbrella around on a sunny day, no one goes back and challenges their guess (excuse me, carefully crafted hypothesis).
Some respectable sites, such as insiderbaseball.com, do their own internal auditing of past predictions, always attempting to improve their success through formula tinkering. The real question becomes: Is any of this necessary? Specifically, couldn't anyone look back at the past few years of performance and just estimate a player's future BA, HR, RBI, etc. without the use of a supercomputer?
I decided to answer my own question and test the theory. Before the season starts, I will select 25 hitters and 25 pitchers. Most of them will have 3+ years of experience, but at least 5 of each will have just 1 or 2 years of MLB experience. I will simply look at their previous stats (imagine staring at the back of a baseball card) and make my projections in the categories listed at the bottom of the article. My projections will involve no calculator, no paper, and no advance knowledge other than what I already know. In other words, I know Barry Zito is now on the Giants so I may lower his ERA projection, but if I'm not sure what team Ronnie Belliard is on, it will just have to be a guess on my part.
Obviously, players with little or no experience are next to impossible to predict without plenty of comparison data, so I wouldn't expect to compete in predicting performance of rookies and young players. My biggest question is whether the performance of the majority of established players can be better forecasted through formulas than a quick look at past stats.
I will choose my players in a semi-random basis, attempting to get a wide range of players with my 50 choices. Future Fantasy Focus articles will include a spreadsheet of my predictions and updates along the way. Keep in mind, Bill James himself says that people take his work too seriously and that the evidence that the formulas work is almost non-existent. I guess James and I are on the same page, so the end of the season will give us a better indication if we are correct. I will use Sportsline.com as a reference for comparing predictions. They are one of the biggest fantasy sports sites (if not the biggest), so most readers should be familiar with their predictions. I'm not sure what the results will be, but if my predictions turn out better than Sportsline.com's, you can bet your baseball card collection the Marcinkus Handbook will be out in 2008!
Hitter: BA, OBP, R, 2B, HR, RBI, SO, SB
Pitcher: W, L, S, IP, ERA, WHIP, K