Curve Ball: Baseball, Statistics, and the Role of Chance in the GameIn its formative years, from the 1970s through the 1990s, sabermetrics was p- marily an amateur undertaking. Publications were aimed at a relatively small audience of baseball fans. To be sure, this ever-growing group of aficionados brought a lot of sophistication to baseball analysis, and were constantly looking for statistical insights beyond the listings of the top ten batters found in popular newspapers and magazines. But their influence on the baseball profession was very limited. A few consultants like Craig Wright developed temporary relati- ships with various teams, but none were able to stay long enough to create a p- manent sabermetrician staff position. (See Rob Neyer’s November 11, 2002, arti- 1 cle on ESPN. com. ) All of this changed, however, in 2002 with the hiring of Bill James by the Boston Red Sox. With that move, we have seen the admittance of the foremost proponent of sabermetrics into the top echelon of professional ba- ball management. The art and science of careful statistical analysis, it now seems, had made it into the big leagues. Since the publication of the first edition of Curve Ball in 2001, we have been overwhelmed by the positive responses from readers and critics. We’re pleased with the reception, of course, but we don’t want to rest on our laurels. Like a pitcher refining his repertoire, we’ve revised, expanded, and updated the book for its publication in this paperback edition. Several readers and critics took us 1 http://espn. go. com/mlb/columns/neyer_rob/1456664. |
Contents
SIMPLE MODELS from Tabletop Baseball Games | xvii |
Model Assumptions of AllStar Baseball | 6 |
Introducing the Pitcher | 7 |
The Independent Model | 13 |
The Interactive Model | 18 |
Which Model Is Best? | 22 |
EXPLORING Baseball Data | 25 |
Simple Graphs | 27 |
OnBase Plus Slugging OPS | 162 |
Batters Run Average BRA and Scoring lndex DX | 166 |
Runs Created RC | 167 |
More Analytic Models | 170 |
AVERAGE RUNS per Play | 173 |
Least Squares Linear Regression LSLR | 174 |
Adding Caught Stealing to the LSLR Model | 180 |
Adding Sacrifice Flies to the LSLR Model | 183 |
Typical Valuesthe Mean and the Median | 29 |
Measures of SpreadQuartiles and the Standard Deviation | 30 |
Interesting Values | 32 |
A FiveNumber Summary | 33 |
OBPs of Offensive and Defensive Players | 35 |
Relationships Between Batting Measures | 36 |
Relating OBP and SLG | 37 |
What about Pitching Data? | 39 |
Strikeouts and Walks | 40 |
Looking at Strikeout Totals | 41 |
Defining a Strikeout Rate | 42 |
Comparing Strikeout Rates of Starters and Relievers | 45 |
Association Between Strikeouts and Walks? | 46 |
Exploring Walk Rates | 47 |
Comparing Walk Rates of Starters and Relievers | 48 |
INTRODUCING Probability | 49 |
Looking for Real Effects | 51 |
Predicting OBPs | 53 |
Probability Models | 55 |
Observed and True OBPs | 57 |
Learning about Batting Ability | 60 |
Estimating Batting Ability Using a Confidence Interval | 64 |
Comparing Hitters | 66 |
SITUATIONAL EFFECTS | 69 |
Surveying the Situation | 70 |
Looking for Real Effects | 72 |
Observed and True Batting Averages | 73 |
Batting Averages of the 1998 Regulars | 76 |
Two Models for Batting Averages | 77 |
Do All Players Have the Same Ability? | 78 |
A Model Using a Set of Random Spinners | 79 |
Situational Effects | 84 |
Turf vs Grass | 85 |
Scenario 1 No Situational Effect | 87 |
Scenario 2 Situational Bias | 88 |
Scenario 3 Situational Effect Depends on Ability | 89 |
Finding Good Models | 90 |
What Do Observed Situational Effects Look Like When There Is No Effect? | 91 |
The Last Five Years Data | 93 |
The No Effect Situations | 94 |
The Bias Situations | 94 |
The Ability Situations | 97 |
How Large Are the True Ability Effects? | 102 |
Game Situation Effects | 103 |
A Lot of Noise | 104 |
STREAKINESS or The Hot Hand | 107 |
Thinking about Streakiness | 108 |
Moving AveragesLooking at Short lntervals | 112 |
Runs of Good and Bad Games | 115 |
Numbers of Good and Poor Hitting Days | 116 |
Mr Consistent | 118 |
Mr Streaky | 122 |
How Does Mr Streaky Perform During a Season? | 125 |
Mr Consistent or Mr Streaky? | 128 |
Team Play | 130 |
A Consistent Team | 134 |
A Streaky Team | 137 |
Thinking About Streakiness Again | 139 |
MEASURING Offensive Performance | 141 |
The Green Quest | 142 |
Runs Scored per Game | 144 |
Batting Average and Runs Scored per Game | 149 |
Slugging Percentage and OnBase Percentage | 153 |
Intuitive Techniques | 161 |
The LindseyPalmer Models | 185 |
Palmer Enters the Picture | 195 |
Comparing the LSLR and LindseyPalmer Models | 198 |
THE CURVATURE of Baseball | 203 |
The DLSI Simulation Model | 204 |
The Probability of Scoring Two Runs | 205 |
The Probability of Scoring No Runs | 207 |
A DLSI Example | 211 |
Lessons from the Simulation | 215 |
DLSI and Runs per Play | 220 |
Where Do We Stand? | 222 |
Additive Models | 223 |
Product Models | 224 |
Player Evaluations in the Best Models | 226 |
Player Evaluations on an Average Team | 229 |
Sorting Out Strengths and Weaknesses | 236 |
MEASURING CLUTCH PLAY | 239 |
Clutch Hits | 241 |
Leading Off an Inning vs Not Leading Off | 245 |
Two Outs vs NoneOne Out | 247 |
Situation Evaluation of Run Production | 249 |
A New Criterion for Performance | 255 |
The Calculation of Win Probabilities | 262 |
Player Game Percentage PGP | 268 |
World Series Most Valuable Players | 275 |
Looking to the Future | 278 |
PREDICTION | 281 |
Guessing | 282 |
Predicting 1999 Game Results | 283 |
How Good Were Our Predictions? | 285 |
Predicting the Number of McGwire and Sosa Home Runs | 287 |
Whats Wrong with This Prediction? | 288 |
A Spinner Model for HomeRun Hitting | 289 |
How Many AtBats? | 290 |
Binomial Probabilities | 291 |
What If We Dont Know Sosas True HomeRun Rate? | 292 |
Revising Our Beliefs about Sosas HomeRun Probability | 294 |
One Prediction | 295 |
Many Predictions | 298 |
Predicting Career Statistics | 301 |
Sosas HomeRun Probabilities | 302 |
How Long and How Many AtBats? | 303 |
Making the Predictions | 305 |
DID THE BEST Team Win? | 307 |
The Big Question | 308 |
Describing a Teams Ability | 310 |
1871 to the Present | 311 |
Explanations for the Winning Percentages | 313 |
A Normal Curve Model | 315 |
Team Performances over Time Revisited | 317 |
A Mediocrity Model for Abilities | 319 |
A Normal Model for Abilities | 320 |
Weak Average and Strong Teams | 321 |
A Model for Playing a Season | 322 |
Simulating a Season | 323 |
Simulating an American League Season | 324 |
Simulating Many American League Seasons | 328 |
Performances and Abilities of Different Types of Teams | 329 |
Simulating an Entire Season | 333 |
Chance | 336 |
POSTGAME COMMENTS A Brief Afterword | 339 |
Bibliography | 343 |
Other editions - View all
Curve Ball: Baseball, Statistics, and the Role of Chance in the Game Jim Albert,Jay Bennett Limited preview - 2003 |
Curve Ball: Baseball, Statistics, and the Role of Chance in the Game Jim Albert,Jay Bennett Limited preview - 2007 |
Curve Ball: Baseball, Statistics, and the Role of Chance in the Game Jim Albert,Jay Bennett No preview available - 2003 |
Common terms and phrases
1993 World Series Alomar American League at-bats Average Team baseball statistics batter batting averages boxplot calculation career Carter's chance Chapter column compute consistent hitters error estimate evaluation fans Figure getting on base graph hitting data home runs home team home-run probability large number look Loss Points Major League Mark McGwire National League number of games number of runs OBP values observed On-Base Percentage on-base profiles percent Phillies pitcher plate appearances play players plot Pr(0 runs prediction probability of scoring RC/G reach base RMSE run value runners in scoring Runs Created Runs per Game runs scored sacrifice flies scatterplot Scoring Position simulation situational effects Slugging Percentage Sosa spinner standard deviation stemplot Strat-O-Matic streaky hitter strikeout strikeout rates Table team's Todd Zeile Tony Gwynn Toronto true batting average walks weights win probabilities winning fractions winning percentages Yankees Zeile
Popular passages
Page i - The difference between a ballplayer's ability and his performance, and how probability is used to model the role of chance in performance.


