Process vs Results: The Need for Mechanical Data

Today’s Question:

How can we measure/integrate/use metrics to quantify and better understand a player’s mechanics rather than their outcomes?

The essence of this question regards our modern data collection. Although our technology, like TrackMan, StatCast, and Rapsodo, is tracking more baseball data than ever before, there’s still another frontier that hasn’t been nearly as explored.

In statistics, we often aim to separate process from results. The thinking here is that results (the effects of a certain combination of causes) can be unpredictable. As inconvenient as this is, sometimes unlikely things happen. Randomness is often one of the causes for these unlikely events, and randomness is no bueno.

No, we don’t want to focus on results, because what happened in the past may not happen again! But here’s the thing. Most of the data we have in baseball (and in the world for that matter) is information regarding results. Something happens. We write it down. The first baseball statisticians did this when they first recorded hits and errors, and we do it today when we measure velocity, spin rates, and launch angle. These are the results of processes being conducted.

But what are these processes and how can we measure them? Baseball coaches are very familiar with the essential process of the game: mechanics! The way that a player’s body moves during a pitch or a swing is the cause of whatever happens next, whether it be a home run, a strikeout, or anything in between. There is much less randomness influencing a player’s mechanics on any given rep, and that is why we should focus our quantitative efforts on it.

We already track some of these process-oriented statistics. Some examples are a pitcher’s extension, a hitter’s bat speed (with the less mainstream Blast Motion sensor), and a baserunner’s sprint speed. But in the year 2018, I am not aware of a public tool that could help an analyst or coach quantify a player’s mechanics in a way that would be beneficial.

Bat speed is a great start, but the sensors are still fairly expensive and cannot be used in games. But there are so many other possibilities! I’m talking about a pitcher’s torque in degrees/second during a delivery or a hitter’s bat angle throughout a swing. In theory, we’re heading towards a reality where players don’t have to wear these in order for teams to collect data about the minute details of their processes. Statcast can already capture a catcher’s pop time or an infielder’s exchange time on a double play, so it isn’t hard to imagine a device in each dugout focused on the hitter and one trained on the pitcher to get the kind of data only currently captured in labs.

Though I’m sure MLB teams are doing their due diligence on this subject, I’m not aware of the state of public work as far as making this kind of data available. I do know that Driveline Baseball, one of the industry’s most innovative training facilities, is way ahead of the curve when it comes to biomechanical analysis, especially for pitchers.

Having this information available would be revolutionary. It would help coaches and players better identify mechanical weaknesses or oddities and fix them before a slump begins or bad habits form. Today, this information is mostly monitored by coaches’ sight, or by examining trends in the data we have. (For example, teams can often sense a change in a pitchers’ mechanics if his velocity dips during an outing, but because they don’t have biomechanical data, they can’t be sure that the mechanics are to blame.)

And isn’t that the point of modern analytics anyways? If this data allows players and coaches to collaborate more efficiently, players will improve at a greater rate than they otherwise would have. In addition to benefits on the player or team level, this data would help analysts to identify macro trends in mechanics that would likely unearth some major findings and could ultimately lead to creating the optimal mechanics for future athletes.

It seems like the game has been revolutionized just in the last 20 years alone, and it has, but there is so much more technology and understanding on the horizon. And beginning to affordably and accurately quantify player mechanics is a step towards that future.