Analytics and Anecdotes is a four part series whereby I attempt to examine the growing influence of statistics and analytics in some sports -popularised by the Brad Pitt film, Moneyball- while it’s apparent inability to permeate existing structures of assessing players in others. In this second part, we take a look at baseball- a sport which has taken almost a 180 degree turn from focusing on scouting to focusing on analytics. In the previous part, we examined the basic foundation and aims of analytics. Here, we see how it has transformed baseball.
Destiny’s Child- Baseball’s Tryst With Analytics
With the success of the Oakland Athletics during the 2002 Major League Baseball season with under-the-radar players like Scott Hatteberg and Jeremy Giambi, there was no question that the statistical revolution in baseball that A’s general manager Billy Beane brought was the way forward for baseball.
Traditional scouting reports that included things like batting form or pitching action now gave way to percentages. General managers now have started trusting numbers more than what they see with their eye. The reason for this is simply that numbers are solid proof and not mere observation. As I mentioned in the previous segment, observation is subject to human error. Numbers are not. And baseball was a numbers game.
One key reason, for baseball’s successful enterprise with analytics is it’s ability to create and exploit matchups. Often boiling down to simple pitcher versus batter matchups, the general managers and coaches are able to exploit the individuality (I’ve explained this concept in my previous post on the subject) to their advantage. For example, if a team has more of right handed batters, my pitching lineup is going to be- say line-up A, and if a team has more left handed batters, it’s going to be line-up B.
Coaches these days are further able to exploit specific weaknesses of players because of statistics. For example, if a team finds that a batter is unable to handle a curveball, his slugging percentage comes down. The slugging percentage represents the total number of bases divided by the number of times the player is at bat, or simply, at bats. Slugging percentage is a representation of a player’s ability to hit the ball with power, as it doesn’t take into account a walk-on base. Statistics in baseball have advanced today to the point where general managers and coaches can look at slugging percentage for the different types of pitches that a person faces. And this can help identify a batter’s weakness. This can further translate into preparation for a game and subsequently it’s execution within the game.
The effect of analytics on the other side of the ball- that is, pitching and fielding, still needs some work. Pitching is traditionally assessed by the Earned Run Average. This is simply calculated by dividing the number of earned runs allowed by the number of innings pitched and multiplying by nine because of the nine innings. This statistic provides the number of runs that a pitcher allows per game. However, this doesn’t take into account the quality of fielders a pitcher plays with. This has led to the development of advanced statistics that negate the effects of a pitcher’s teammates in the field. These include the Defense-Independent Pitching Statistics System (DIPS). The DIPS system works by focussing on statistics which are almost totally controlled by the pitcher, for example, home runs, strike outs and walk ons. More recently, statistics like fly ball percentage and ground ball percentage have been added to it. By eliminating the effects of a fielding unit, Voros McCracken- the man credited with the development of DIPS – was able to prove that there is little difference in the quality of the pitchers. However, McCracken’s approach was not entirely fool-proof with findings by Tom Tippett showing that while home runs conceded remained fairly constant no matter who the pitcher was, differences arose annually in the Batting Average On Balls In Play (BABIP) which accounted for all hits- including any singles, doubles or triples.
Today, statistics have further advanced to include or exclude variables like the dimensions of the ballpark (all baseball fields have different dimensions), the strength of the fielding unit and so on. In fact, baseball has become so dependent on analytics that it now has an entire field of study that deals with statistics in baseball called Sabermetrics named after the Society for American Baseball Research (SABR).
Today Sabermetrics has evolved from algebraic algorithms to include higher mathematics. The most popular of these is the Value Over Player Replacement which helps determine how much a player contributes to the team. Others include the use of calculus (yes, calculus) to determine things that cannot be explained by algebra. For example, calculus is being used today to determine how often and when a team should attempt to steal bases, either in general or against a particular team.
Sabermetrics have grown today as a tool for teams to evaluate
- Past performance of a player
- Tracking player’s performances during the course of a season to notice improvements or regression
- Predicting the future development of a player and the future performance of a player
In addition, sabermetrics is today used to determine who should win awards like the MVP award, especially in close races. It is used to determine which player to draft from the annual MLB draft. It can be used to predict if a good player in college means a good player in the MLB. There are many more uses but I am not going to go into that.
Crucially, one more element that makes baseball different is it’s lack of positional interdependence to the level that is seen in other team sports like soccer. Apart from the relationship between a pitcher and his fielders- which is also far less impactful than one might expect- there are no real interdependent relationships on the baseball field. And this sort of system allows for more independent function. This in turn means that when a coach is deciding on various schemes – defensively or offensively- he is looking at a series of individual matchups with confounding factors (to use a statistical term). The outcome can be more easily affected by simply manipulating each individual matchup than in other sports. For example, basketball has the concept of help defence and American football has a concept of double coverage and both sports have zone defence in them. But baseball lacks an equivalent.
So, is good old-fashioned scouting really dead? Bob Johnson, a former scout with the Washington Nationals says in an interview with NY Daily News, “The job is disappearing. Only about a half-dozen teams still use an advance scout. Most teams use video and analytics instead. That’s the trend, and it’s fine — unless you’re trying to win.”
For the layman, an advance scout in baseball is tasked with compiling reports on upcoming opponents by watching them when they play a game. He is usually away from home most of the season.
“Video is great but there is a lot it doesn’t tell you. It doesn’t show how the defense, how outfielders react to the ball off the bat, or why a hitter isn’t handling a certain pitch because he has a thumb injury. There are a lot of things you pick up by being at the ballpark. You watch the catcher’s feet. Last year there was a catcher who tipped every pitch by the way he set up with his feet — a good advance scout sees that and reports it.”
“Most teams use video and analytics instead. That’s the trend, and it’s fine — unless you’re trying to win.”
Johnson was let go by the Nationals at the end of the season before last when the Washington ball club said they were “going in a different direction”. They hired three analytics guys to replace him.
However, to say that scouting is dead is not accurate. Johnson himself, for example, helped the Kansas City Royals when they reached out to the Nationals scouting department for reports on their two playoff opponents in the 2015 American League Championship Series and the World Series- the Toronto Blue Jays and the New York Mets. Johnson, according to the article, helped identify patterns that the Mets pitchers had gotten into, which allowed Kansas to run better. Kansas ended up winning the World
Series in 2015.
Johnson also says that it allows him to identify players who continually change and modify their game through the season. Speaking of then-Mets and current Nationals second baseman Daniel Murphy, Johnson says, “There were about five different Daniel Murphys last year (2015). He makes adjustments during the season as well as anybody, and he’s a completely different hitter when he’s looking away compared to when he’s looking in. You need to see his last 20 at-bats coming into a series.”
From what Johnson says and with the evidence of the Royals’ World Series win in 2015 to back it up, scouting, it seems, has a real place in today’s world. But with the advent of video analysis that can cover every part of the field (including the fielders’ and pitchers’ tendencies as Johnson points out), and with statisticians becoming more and more equipped to deal with the modifications that players make during a season coupled with more real-time analysis, it is hard to explain the commercial sensibility in paying a guy (not to mention paying for hotels and food and so on) to scout teams in far off places when one could have much more information readily available within the bullpen of a clubhouse.
Scouting, in today’s baseball, however, does find a major role in evaluating a player’s intangibles. His attitude to the game, his leadership qualities as well as questions like how he responds to adversity can be answered purely by a scout- who may look at him in practice, look at his interaction with peers and look at his value to the team from a non-baseball point of view but from a personality standpoint. This is especially useful when evaluating draft prospects out of college or out of other pro leagues like Japan. Scouting is also used these, in conjunction with science to evaluate a player’s medical condition. This is especially important to see if he can withstand the rigours of an MLB season that comprises of 162 games followed by a post season as well as assess any existing chronic injuries.
In the end, baseball has become the darling child of sports statisticians whether it intended to or not, thanks to a combination of inherent factors within the game, as well as various success stories. But so far, it has failed to analyse minor flaws in a team that only an opposing scout with a deep knowledge of the game can catch. But going by current trends, this may not last for long. The evidence is pointing overwhelming to a scenario where one day, possibly every on-field aspect will be covered by video analysis and data collection, leaving only the off-field aspects to scouting.