This is a not-particularly-scientific study of the effects of luck on a baseball team's chances, using the 2003 Diamond Mind Baseball season disk as data.
After the events of the last two seasons, baseball fans may be forgiven for thinking that luck, not talent or skill, is the primary determinant of success in baseball. (Fans of the Chicago Cubs are more than willing to discuss this topic at great length.) I became more aware that luck plays a factor in baseball, at least in simulated baseball, when I was replaying the Jays' 2003 season using Diamond Mind.
My simulated Roy Halladay was born under a bad sign: he always pitched in bad luck. In a typical start, the digitized Doctor would get 11 straight ground ball outs, put two men on base with bloop singles, and give up a three-run bomb. Or, he'd pitch a complete game, allow two runs, and lose 2-1. By August, he was leading the league: with 15 losses. This led me to do two things:
- Fire myself as manager (i.e., stop playing the season).
- Wonder: how much luck is there in baseball?
To try to test this, I used Diamond Mind to automatically replay the 2003 season twelve times. (I'd've done more than 12, but I have a short attention span.) Here are the Jays' won-lost records, the division winners, and the wild card winner for these seasons:
1 83-79 NYY CHW OAK BOS
2 85-77 NYY MIN OAK CHW
3 79-83 NYY CHW SEA BOS
4 91-71 NYY CHW and MIN (tie) OAK BOS
5 96-66 NYY CHW SEA TOR
6 93-69 NYY MIN OAK BOS
7 85-77 NYY CHW SEA MIN
8 87-75 BOS MIN SEA NYY
9 79-83 BOS MIN OAK NYY
10 87-75 BOS CHW OAK NYY
11 87-75 BOS CHW OAK NYY
12 80-82 BOS CHW OAK SEA
Notice that the Jays' win total ranged from 79 to 96, which is about a 24% difference. Of the 14 American League teams, six made it to the postseason five or more times, seven didn't make it at all, and the Jays managed to sneak in once. Each division title was shared by two of the eligible teams. I have no idea why the Yankees won the first seven divisional races and lost the next five. Neither, I assume, does George Steinbrenner.
As for luck on an individual level: here are Carlos Delgado's batting lines for the twelve seasons, with batting average, home runs, RBI's, and (for the fun of it) hit by pitches listed. Yes, I know these are the traditional counting stats, but we're not trying to analyze players here, just their horseshoes.
.334 39 148 16
.276 30 99 19
.304 32 103 9
.310 42 121 16
.331 49 149 21
.317 50 150 15
.311 39 109 14
.306 40 125 15
.273 46 135 13
.314 43 134 19
.317 49 154 21
.285 49 154 10
Here, computerized Carlos compiles batting averages ranging from .273 to .334. There is a 61-point difference between best and worst, which is about 22%. The home run totals range more widely, with extremes of 30 and 50, or a 66% difference. The RBI totals are almost as widely spaced, ranging from 99 to 154, or a 56% difference.
Oddly enough, the strongest correlation here between team success and individual success is Delgado's HBP total. When he gets plunked 13 times or fewer, the Jays finish under .500. Carlos, I guess this means you have to take one for the team every now and again.
Here's Roy Halladay's seasons (except for season 5, which I stupidly forgot to save pitching stats for). The numbers are ERA, wins, losses, and home runs allowed:
3.52 18 10 27
3.75 14 13 37
3.06 20 10 23
3.68 16 11 30
2.68 18 6 21
3.21 18 10 25
3.45 17 12 24
3.47 18 11 33
3.98 16 13 24
2.95 18 10 23
3.32 15 11 32
Notice that the good doctor only won 20 once in his 11 simulated seasons. While his ERA fluctuated between 2.68 and 3.98, his win total was relatively stable: he won 18 five times. His range of wins, measured as a percentage, was actually smaller than his range of ERA.
What conclusions can we draw?
Not many, to be honest - this is a small sample size, the simulation may be flawed, etc., etc. But here's what I conclude:
- Luck does play a factor in the pennant race, but only up to a point. If you're armed with popguns, and your opponent has tanks, I'd bet on the tanks. The Yanks are likely to win this year; the Jays could; the Tigers won't.
- There's no point in fine-tuning a player rating system. If a player's batting average (to use one metric) can swing as much as 60 points based on luck alone, it's better just to put the guys hitting about .260 in one pile, and the guys hitting about .300 in another pile.
- I'm looking forward to the season finally starting. That's not scientifically proven, but it is, nonetheless, true.
After the events of the last two seasons, baseball fans may be forgiven for thinking that luck, not talent or skill, is the primary determinant of success in baseball. (Fans of the Chicago Cubs are more than willing to discuss this topic at great length.) I became more aware that luck plays a factor in baseball, at least in simulated baseball, when I was replaying the Jays' 2003 season using Diamond Mind.
My simulated Roy Halladay was born under a bad sign: he always pitched in bad luck. In a typical start, the digitized Doctor would get 11 straight ground ball outs, put two men on base with bloop singles, and give up a three-run bomb. Or, he'd pitch a complete game, allow two runs, and lose 2-1. By August, he was leading the league: with 15 losses. This led me to do two things:
- Fire myself as manager (i.e., stop playing the season).
- Wonder: how much luck is there in baseball?
To try to test this, I used Diamond Mind to automatically replay the 2003 season twelve times. (I'd've done more than 12, but I have a short attention span.) Here are the Jays' won-lost records, the division winners, and the wild card winner for these seasons:
1 83-79 NYY CHW OAK BOS
2 85-77 NYY MIN OAK CHW
3 79-83 NYY CHW SEA BOS
4 91-71 NYY CHW and MIN (tie) OAK BOS
5 96-66 NYY CHW SEA TOR
6 93-69 NYY MIN OAK BOS
7 85-77 NYY CHW SEA MIN
8 87-75 BOS MIN SEA NYY
9 79-83 BOS MIN OAK NYY
10 87-75 BOS CHW OAK NYY
11 87-75 BOS CHW OAK NYY
12 80-82 BOS CHW OAK SEA
Notice that the Jays' win total ranged from 79 to 96, which is about a 24% difference. Of the 14 American League teams, six made it to the postseason five or more times, seven didn't make it at all, and the Jays managed to sneak in once. Each division title was shared by two of the eligible teams. I have no idea why the Yankees won the first seven divisional races and lost the next five. Neither, I assume, does George Steinbrenner.
As for luck on an individual level: here are Carlos Delgado's batting lines for the twelve seasons, with batting average, home runs, RBI's, and (for the fun of it) hit by pitches listed. Yes, I know these are the traditional counting stats, but we're not trying to analyze players here, just their horseshoes.
.334 39 148 16
.276 30 99 19
.304 32 103 9
.310 42 121 16
.331 49 149 21
.317 50 150 15
.311 39 109 14
.306 40 125 15
.273 46 135 13
.314 43 134 19
.317 49 154 21
.285 49 154 10
Here, computerized Carlos compiles batting averages ranging from .273 to .334. There is a 61-point difference between best and worst, which is about 22%. The home run totals range more widely, with extremes of 30 and 50, or a 66% difference. The RBI totals are almost as widely spaced, ranging from 99 to 154, or a 56% difference.
Oddly enough, the strongest correlation here between team success and individual success is Delgado's HBP total. When he gets plunked 13 times or fewer, the Jays finish under .500. Carlos, I guess this means you have to take one for the team every now and again.
Here's Roy Halladay's seasons (except for season 5, which I stupidly forgot to save pitching stats for). The numbers are ERA, wins, losses, and home runs allowed:
3.52 18 10 27
3.75 14 13 37
3.06 20 10 23
3.68 16 11 30
2.68 18 6 21
3.21 18 10 25
3.45 17 12 24
3.47 18 11 33
3.98 16 13 24
2.95 18 10 23
3.32 15 11 32
Notice that the good doctor only won 20 once in his 11 simulated seasons. While his ERA fluctuated between 2.68 and 3.98, his win total was relatively stable: he won 18 five times. His range of wins, measured as a percentage, was actually smaller than his range of ERA.
What conclusions can we draw?
Not many, to be honest - this is a small sample size, the simulation may be flawed, etc., etc. But here's what I conclude:
- Luck does play a factor in the pennant race, but only up to a point. If you're armed with popguns, and your opponent has tanks, I'd bet on the tanks. The Yanks are likely to win this year; the Jays could; the Tigers won't.
- There's no point in fine-tuning a player rating system. If a player's batting average (to use one metric) can swing as much as 60 points based on luck alone, it's better just to put the guys hitting about .260 in one pile, and the guys hitting about .300 in another pile.
- I'm looking forward to the season finally starting. That's not scientifically proven, but it is, nonetheless, true.
