Here is a short break from the rule 4 draft. We have been getting so many hits lately, I thought a reintroduction to my perspective might be in order. Many think I am a sabermetrician, but if that is the case . . . I guess we all are.
Baseball is a counting sport. It is one of those things that is so special. We know what 56 means. We know what 2632 means. We know 4256. Someday, we’ll have to suck it up and remember 762. Baseball is special because so much of it is quantitative and we try to find standard ways of measuring everything we observe in baseball. It is why arguing about Babe Ruth and Willie Mays is far easier than arguing about Johnny Unitas and Dan Marino. This process began with Henry Chadwick’s cricket-inspired box scores as he tried to find a succinct way to describe the game in a newspaper column with limited space. Quickly, one imagines, statistics were not only used to describe what happened, but also try to predict what will happen. From our own respective fields, it is easy to see how useful it is to know what went right or wrong. It is far more useful if we can use past experience to encourage good things or prevent bad things. This need to be able to predict probable outcomes spurned advances in statistical analysis to this day. Baseball is less of a black box and more like a poor english translation. It kind of makes sense and you know what went into it, but the words are just a little off.
A little off is more valuable than having no clue, so predictive analysis garnered some steam. Branch Rickey and Al Roth dreamed up two quite useful statistics in the 1950s to fill a need. They wanted to better characterize power and the ability to get on base. The prevailing methods, slugging percentage and batting average, were considered by them to be somewhat misleading. In turn, they created isolated power and on base percentage. For those who do not know why these statistics are important, I’ll quickly explain. If you do, skip to the next paragraph. Isolated power is basically slugging percentage except you do not give credit for first base. This is a tool that attempts to measure power more accurately than slugging percentage does. Singles are typically not as effective scoring runs or creating run scoring opportunities than multi-base hits. Additionally, a player with an absurdly high number of balls that fall in for hits is often in a large sample of at bats is often benefiting from an unsustainable singles rate. The tradeoff is that you do lose a lot of information by ignoring the singles. On base percentage was also an improvement on the existing statistics because it corrected this commonly held fallacy: walks are given, not earned. Walks were, and still are by some, considered to be a failing of the pitcher, but the batter deserves some credit here. Rickey and Roth were smart enough to challenge the status quo and determine new ways to view the game.
In the past thirty years, we have seen the statistics side of the game develop more and more. We’ve fallen down foolish paths (i.e., saves, holds, game winning RBIs) and have developed some truly astonishing tools (i.e., wins shares, VORP, WARP, DIPS, fielding metrics). The way in which we can observe the game is also changing with the spread of Pitch f/x and the eventual availability of similar systems for hitting and fielding. It is within this exciting time that we are beginning to understand just how things work in baseball. Of course, the more and more we learn, the more and more we realize how little we actually know. The first step in citing any statistic is knowing the framework in which that number was generated. Numbers can be deceiving, but they also hold a lot of promise in predicting future performance if you know what they mean. From this perspective, I watch and have great passion for the game. It is a perspective that I have used throughout my, there are things that we know and things that we don’t. Using the best knowledge I have on hand, I make a prediction. This brings us to Chad Bradford and his release point.
The first game of the year that I attended was on May 31. I had seats 26 rows behind home plate and off centered toward third base. Very good seats for an Orioles-Red Sox game that could potentially feature Manny Ramirez becoming the 24th player to reach 500 home runs and the second one to do it in Camden Yards. In a bittersweet way, I brought my camera hoping to see Manny launch one. His first two at bats were against Garrett Olson. Garrett Olson is a prime guy to hit a homerun off of. His groundball rate is 43.1%, which is on the lower end for the Orioles. He does seem to keep batters off guard a bit as he achieves a rather low HR per fly ball rate (7.1%) for how many fly balls he actually gives up. Those two at bats resulted in a shot to left that was caught and a groundball to Melvin Mora for an out.
As the night wore on, number 500 was looking to be more and more elusive. Manny’s chances of hitting a homerun decreased in his third at bat as Lance Cormier came in. Lance induces grounders almost 60% of the time and his mistake pitches are hit for homeruns at an average clip (9.5% HR / fly ball). Manny hit another ball hard, but it was for an out into left field. The top of the seventh saw a 3-3 gridlocked game go 4-3 as a David Ortiz sky high popup to left field and, arguably, a poor handle by Ramon Hernandez led to a 4-3 score in the Red Sox favor. To try to keep the score from hemorrhaging, Trembley called in Chad Bradford. At this point, I turned my camera off and put it in my pocket. I commented to my friend that no one was going to see Manny hit a homerun tonight. Chad produces groundballs two-thirds of the time. He gives up about one or two homeruns a year. Now, it should be known that when he does give up a flyball, it has a 1 in 5 chance of being a homerun. When he makes mistakes, they are awful mistakes. I figured with a guy like Manny at the plate, there is no way he would leave a ball up in the zone. Well . . . to the right is the Pitch f/x data from the Bradford-Ramirez at bat. A single fast ball out of his hand at 81.5mph about 2.5 inches away from his normal release point and wound up across the center of the plate, at a height of 2.8 feet, and a final speed of 73.7mph. Manny redirected the pitched 425 feet into the center-right center stands. Shortly before that outcome, I took my camera out of my pocket just to be safe and captured this:
No matter what the data may suggest, there is always uncertainty. I don't know how to predict when a pitcher will have a poor release point. The ability to predict performance is the dragon we all chase. In baseball, the manager tries to determine which player in which situation results in the best chance to win. In life, when we buy a house or sign a contract for a job . . . we are using known information and try to project how a community will change or how our property will change in value. Sometimes you get it right and sometimes you lean too heavily on a single statistic to make a decision, whether it is a ground ball percentage or a growth index.
Baseball is not a black box. Though, it is a complex series of events affected by many variables, of which we only know a few. Knowing just a few variables can greatly increase accuracy. A translator knows this. A realtor knows this. A guy who watches baseball knows this. So, sit back and enjoy. My next post will be much less in terms of narrative and more in terms of: Have we reaped what Andy MacPhail has sewn?