Sabermetrics are founded on the idea that intelligent people can use a wide base of data to learn more about baseball. Sabermetrics also require a forward-thinking mind that is willing to test the limits of the impossible and can laugh at the notion that everything has truly been discovered in this world.
Last summer, I had the benefit of working as an intern at the SABR 42 conference in my second home town of Minneapolis, MN. While there, I got to sit in on a small analytics meeting that led off with a statement similar to, "Well, we've got offense pretty much figured out," and the discussion immediately moved to defense. My only response to this was to think, "No. We absolutely have not gotten it all figured out, and we never will."
Let me explain. About a year and a half ago, I was a 20 year old kid starting his senior year in college. In my university's Sport Management program, we were required to log field hours working with a sports entity. I had the great fortune of working with Inside Edge Scouting Services, and this meant the rare opportunity to work with very intimate baseball data. When talking with one of my supervisors, we got on the topic of offense and what the company records. He told me that they logged what they refer to as "Well Hit" balls. This is where the story begins.
I've always been someone who is interested in the process of baseball more than the results. I like to know what a hitter is doing to get on base, I like to know what a pitcher is doing to get batters out, and I like to know what teams are doing to build rosters and win games. When I was told that "Well Hit" balls were being recorded by Inside Edge, I began what became my practicum work: 500 hours of research into what significance "Well Hit" balls (from now on referred to as "WHB") could hold. Before I can talk about any more, I have to explain exactly what a WHB is:
A WHB is a ball that is hit hard within a certain range of trajectory and is hit off of the barrel of the bat. It is a subjective manner of data collection that allows the scouts that collect the data to take into account velocity, trajectory, and what the batter is doing in the process of hitting the ball. The core of my research became the assumption that a hitter has three basic goals at the plate in non-situational at bats (so any play where a manager hasn't put a play on): hit the ball hard, draw a walk, and avoid striking out.
My initial hypothesis was that a hitter who does these things should have great offensive success, which meant a good on-base percentage, good power, and sustainable success. Therefore the next key in my research was to develop a metric that would help explain what exactly could be measured with WHB.
The first thing I did with WHB was to treat them as the ideal contact outcome. Essentially, I took out end results completely, so the fundamental aspect of the metric meant that I would be looking at became a "DIPS theory" for hitters. A hitter can control how hard he hits the ball, how often he walks, and how often he strikes out. This meant the creation of my first metric: WHOBP
WHOBP is "Well Hit On-Base Percentage" and it breaks down like this:
There are a few key differences between WHOBP and OBP. The first is, again, that WHB have replaced hits in the batting average component of the metric. The second is the fact that HBP types are not factored in. While hitters can do things to increase the likelihood of being hit by a pitch, they cannot actually control whether or not the pitcher throws a pitch badly enough to hit them.
The final major difference is that sacrifice flies are not a part of the equation. They are situational outcomes that, in my opinion, shouldn't credit the hitter with anything. If the hitter had hit the ball far with nobody on third base, there wouldn't have been any positive result, and giving credit to him for that effort doesn't say enough about the process of being a good hitter. While a positive outcome came out of the SF, better outcomes could have been had (in later posts, I will talk more about the other categories of ball-in-play data: medium and weakly hit balls).
With this statistic created in its most raw and basic form (it gets much more complex in later posts), I was now able to do some basic comparisons to traditional OBP. My own time, efforts, and resources allowed me to test two full years of data: 2010 and 2011.
In 2011, the league average OBP was .324, and the average WHOBP was .285. With HBP and SF taken out of the equation, this outcome was expected. Now, the biggest thing I did in this research was regression analysis. Man, I did a lot of simple regression analysis. When I ran the relationship between 2011 OBP and 2011 WHOBP among the population of 115 players I selected (who had met minimum qualifications), the resulting R value was 0.71 (.505 R-squared).
I took this relationship as an indicator that WHOBP, when adjusted, could serve as a future predictor of on-base percentage, and that is where I eventually took my research.
The start of my research came out of the simple fact that I don't like end results, and I never have. In school, I was always more interested in what questions I got right/wrong on a test more so than how I did on the test itself. If I was getting hard questions wrong and easy questions right, it was a suggestion I needed to study more. When I got hard questions right and easy questions wrong, it was a sign that I needed to be more careful and take my time.
I apply the same philosophy to whenever I watch baseball. Hopefully you have seen Spencer Schneier's post about how offense perhaps should be evaluated by the process and not the results. Well, I am a firm believer that it should be. In my mind, a hitter didn't have a good at-bat because he hit a home run; he had a good at-bat because he hit the ball well enough to see the home run as a result. It is unfair, in my opinion, to treat that outcome differently than a ball hit equally hard in another ballpark that doesn't go for a homer.
In the series of posts that will be released in the coming weeks, I will go more in-depth regarding this research project I have done. Hopefully by the end you'll have a new perspective on what it means to be a good offensive player.
Credit to Inside Edge Scouting Services for helping me get my start in analytics and allowing me to present it in a public format.