I came upon The Hidden Game of Baseball when I was in high school at our local small town library. At the time I was mainly into baseball cards with a growing interest in following Major League Baseball games, usually in the summer when I had time to listen to games and read boxscores.
This book intrigued me with the concept of finding out what Ty Cobb would hit in 1985, which led to a discussion of normalized statistics. From this book I learned that historically a base stealer hurts his team when he isn't successful 2/3 of the time with his steal attempts. The book was built around the basis of runs scored, and runners making outs on the bases took away opportunities to score runs. There was in depth statistical analysis of this, with historical examples provided.
This is a book I wish I had a copy of today. Some of the concepts have since been passed over, some expanded on, but it formed the basis for much of what baseball statisticians use today for evaluation.
Friday, July 17, 2009
Thursday, July 16, 2009
Initial Schedule
For those so inclined, click here for the list of all the pairings for the play-in round of the tournament.
It's a very long list, unformatted and basic.
It's a very long list, unformatted and basic.
Which is the best major league baseball team ever?
Which baseball team is the best ever? Is it the 1927 Yankees? The 2004 Boston Red Sox? The 1994 Montreal Expos? It's a question that has been debated ferociously for years among hardcore baseball fans like myself.
I plan to put these teams to the test. Inspired by an early 1990s series by (the sorely missed) Baseball Weekly I intend to play out a bracketed tournament to determine which team comes out on top. Through the use of a baseball simulator named Out of the Park Baseball 9 I will pair up teams that will play a 162-game season against each other. I will post the results here in the blog on a series-by-series basis. I'll also provide some historical context for the teams playing.
"But how did you decide which teams make it into the simulation?"
Good question, for which I think I have a great answer: I didn't decide.
There is a difference between Baseball Weekly's simulated series and mine. Baseball Weekly used 32 teams (if memory serves, I may be wrong). My simulation will include every major league team ever, 2,604 teams (through the completion of the 2008 season). That makes this the largest sports bracket ever played out.
With such a high volume of teams I will begin posting numerous series per day when I'm underway.
Fair warning, what follows next is very geeky and could cure any insomnia you may have. Proceed at your own risk!
This orientation creates a number of challenges when creating and seeding the brackets.
Challenge #1: To do it evenly the bracket needs to have either 2,048 teams or 4,096 teams. 2,048 isn't feasible because I have made the decision that all teams make it into the simulation. 4,096 isn't feasible because we fall 1,492 teams short.
Solution: 1,112 teams will play in a pre-tournament series. This will eliminate the 556 teams I need gone to create a 2,048 team bracket.
Challenge #2: How to decide who the 1,112 teams were that would have to play into the tournament?
Solution: I ranked all 2,604 teams in a spread sheet. I ranked them using the following criteria:
1. Winning percentage (because schedule lengths differed over time)
2. Number of wins broke any ties in winning percentage
3. If a tie still existed at this point it was broken by considering team ranking within divisions
I then started at the bottom of the list and traveled up by 1,112 teams to find my cutoff point. This meant almost any team with a winning percentage of .488 or below had to play into the tournament.
Next I paired them up best-with-worst. So the team #1 (1963 Detroit Tigers) was paired with the team #1,112 (1875 Brooklyn Atlantics), team #2 was paired with the team #1,111, and so on.
This is quite an undertaking just to get the tournament to the point where it can begin. But I feel that this is the solution keeps closest to the spirit of the simulation.
Challenge #3: The bracket is just too large visually to portray every team.
Solution: To be honest, I don't have one here. I'm working on one in one of my layout programs, but it's absolutely huge already and I've only completed 1/4 of it.
There's much more on the seeding of the entire tournament, including how I randomized things a bit to help deter any pairings like the 1998 Yankees vs. the 1999 Yankees. The idea is to compare teams across history, and hopefully there won't be too many pairings like the example just mentioned. But I'll delve into all that rigmarole once we've reached the appropriate point in the tournament.
Next post I will list the teams in the initial round.
I plan to put these teams to the test. Inspired by an early 1990s series by (the sorely missed) Baseball Weekly I intend to play out a bracketed tournament to determine which team comes out on top. Through the use of a baseball simulator named Out of the Park Baseball 9 I will pair up teams that will play a 162-game season against each other. I will post the results here in the blog on a series-by-series basis. I'll also provide some historical context for the teams playing.
"But how did you decide which teams make it into the simulation?"
Good question, for which I think I have a great answer: I didn't decide.
There is a difference between Baseball Weekly's simulated series and mine. Baseball Weekly used 32 teams (if memory serves, I may be wrong). My simulation will include every major league team ever, 2,604 teams (through the completion of the 2008 season). That makes this the largest sports bracket ever played out.
With such a high volume of teams I will begin posting numerous series per day when I'm underway.
Fair warning, what follows next is very geeky and could cure any insomnia you may have. Proceed at your own risk!
This orientation creates a number of challenges when creating and seeding the brackets.
Challenge #1: To do it evenly the bracket needs to have either 2,048 teams or 4,096 teams. 2,048 isn't feasible because I have made the decision that all teams make it into the simulation. 4,096 isn't feasible because we fall 1,492 teams short.
Solution: 1,112 teams will play in a pre-tournament series. This will eliminate the 556 teams I need gone to create a 2,048 team bracket.
Challenge #2: How to decide who the 1,112 teams were that would have to play into the tournament?
Solution: I ranked all 2,604 teams in a spread sheet. I ranked them using the following criteria:
1. Winning percentage (because schedule lengths differed over time)
2. Number of wins broke any ties in winning percentage
3. If a tie still existed at this point it was broken by considering team ranking within divisions
I then started at the bottom of the list and traveled up by 1,112 teams to find my cutoff point. This meant almost any team with a winning percentage of .488 or below had to play into the tournament.
Next I paired them up best-with-worst. So the team #1 (1963 Detroit Tigers) was paired with the team #1,112 (1875 Brooklyn Atlantics), team #2 was paired with the team #1,111, and so on.
This is quite an undertaking just to get the tournament to the point where it can begin. But I feel that this is the solution keeps closest to the spirit of the simulation.
Challenge #3: The bracket is just too large visually to portray every team.
Solution: To be honest, I don't have one here. I'm working on one in one of my layout programs, but it's absolutely huge already and I've only completed 1/4 of it.
There's much more on the seeding of the entire tournament, including how I randomized things a bit to help deter any pairings like the 1998 Yankees vs. the 1999 Yankees. The idea is to compare teams across history, and hopefully there won't be too many pairings like the example just mentioned. But I'll delve into all that rigmarole once we've reached the appropriate point in the tournament.
Next post I will list the teams in the initial round.
Subscribe to:
Posts (Atom)