Sunday, April 04, 2004
Projected Standings for 2004
Each spring Tom Tippett of Diamond Mind Baseball simulates each season 100 times, based on rosters and player health as of the middle of spring training, to predict the final standings.
What makes these predictions different from the 100 or so most of us have read is the way Tippett presents the results -- identifying how often, in the course of 100 simulations, each club wins its division or the wild-card berth. Given 100 chances, only six clubs never make the postseason, and only the Yankees and Red Sox qualify more than 90% of the time. Remember that when you see some blowhard, or some Commissioner, saying that even before Opening Day, half the clubs have been effectively eliminated.
Writing now, no one knows which highly touted rookies will fizzle, which veterans will miss half the season with injuries, and which players will far exceed expectations. But even after the season, when everyone's actual performance is known, simulations show that it can still be difficult or impossible retroactively to identify the winner. That was the subject of my first-ever SABR presentation, at the 1993 convention, entitled "Tinkering Ever with Chance, Or...How Well Do Statistics Predict the Present?". I've just uploaded a copy to the Website.
My paper used an earlier version of Diamond Mind to show that even if one knew each player's actual statistics, as well as his club's batting order, pitching rotation and managerial tendencies, 10 or 20 simulations produced three or four different division winners in most six-club divisions. Even with a later version of Diamond Mind that uses each club's actual game-by-game starting lineup, 17 of the 30 major league clubs qualified for the playoffs in at least one of the 10 2003 seasons I simulated. As the saying goes, "that's why they play the games."
|
Each spring Tom Tippett of Diamond Mind Baseball simulates each season 100 times, based on rosters and player health as of the middle of spring training, to predict the final standings.
What makes these predictions different from the 100 or so most of us have read is the way Tippett presents the results -- identifying how often, in the course of 100 simulations, each club wins its division or the wild-card berth. Given 100 chances, only six clubs never make the postseason, and only the Yankees and Red Sox qualify more than 90% of the time. Remember that when you see some blowhard, or some Commissioner, saying that even before Opening Day, half the clubs have been effectively eliminated.
Writing now, no one knows which highly touted rookies will fizzle, which veterans will miss half the season with injuries, and which players will far exceed expectations. But even after the season, when everyone's actual performance is known, simulations show that it can still be difficult or impossible retroactively to identify the winner. That was the subject of my first-ever SABR presentation, at the 1993 convention, entitled "Tinkering Ever with Chance, Or...How Well Do Statistics Predict the Present?". I've just uploaded a copy to the Website.
My paper used an earlier version of Diamond Mind to show that even if one knew each player's actual statistics, as well as his club's batting order, pitching rotation and managerial tendencies, 10 or 20 simulations produced three or four different division winners in most six-club divisions. Even with a later version of Diamond Mind that uses each club's actual game-by-game starting lineup, 17 of the 30 major league clubs qualified for the playoffs in at least one of the 10 2003 seasons I simulated. As the saying goes, "that's why they play the games."
|
