May 27, 2016

Raising Aces

Digging Under The Gun

by Doug Thorburn

A month ago, I was lamenting the modern development of today's pitchers, and a theme that was central to my position was that the present-day obsession with the radar gun has created an unhealthy paradigm shift in the ways that pitchers are bred, developed, and used at the highest level. I then spent the past couple weeks combing through the group of pitchers who have lost the most velocity, highlighting their radar-gun readings and the excellent resources at Brooks Baseball to better understand the trials and tribulations of some of the game's biggest hurlers.

So what gives?

When it comes to pitch development, I list velocity as the third most important ingredient, behind command and movement. However, the radar gun forms the bridge between modern-day sabermetrics and scouting, providing an objective measurement of on-field tools that can be analyzed from a computer thousands of miles away from the field of play. My biggest issue with box-score sabermetrics has always been the limitations of the inputs themselves – so long as a laser up the middle is measured as equivalent to a bunt single, the system is inherently flawed at the most basic level.

Now we have data that is currently being collected in stadiums across the country which is specific to on-field skills, with velocity leading the charge. We are still working on the command angle (COMMANDf/x is getting there but has its flaws) and pitch movement is best understood on a case-by-case basis, but velocity is well understood and, thanks to PITCHf/x, can be used to track a player's pitch-speed over time or to look at in-game velo trends. The future of sabermetrics is in the study of these on-field events, and though we are left to wonder the extent to which data from StatCast, HITf/x, TrackMan, etc. will be available to the public, these systems of measuring data represent remarkable progress by moving the collective focus away from personalized recipes for extrapolating wins via archaic box score stats and toward measuring the on-field process that begets those outcomes.

I mentioned this in last month's velo manifesto, but the impact of the radar gun should not be understated. The gun is a rite of passage for young scouts, some of whom will save their peanuts for months or years from a low-wage job just to purchase the overpriced symbol of inclusion among the game's gatekeepers—it takes a lot of pizza-delivery tips to pay for a Stalker. Stopwatches, charts, notebooks, none of these carries the weight of the radar gun. Sit at a game behind the plate with a radar gun, and folks around you don't just think you're a scout, they assume it.

The radar gun has made it easier to communicate just how hard a pitcher throws, and though there are other variables that influence the effectiveness of a pitch beyond raw velocity, there is no standard of measurement. Saying a pitcher hit 60-percent of his spots today doesn't say much, especially to the layman, but knowing that a pitcher sat 92-94 mph and touched 96 on the gun is something that all can relate to—even those who weren't at the game. And therein lies the value, as a scout has something tangible to pass along to his superiors, so that they can make judgments of ballplayers from the comfort of an office.

I certainly am not immune to the bridge-forming powers of the gun. The ability to be the connection between the subjective viewpoint of scouting and the objective viewpoint of sabermetrics makes the tool extremely important in the baseball culture of the modern world, as the meeting place of the two sides. The resource at Brooks Baseball is amazing for exactly this reason, that we are tracking objective data for on-field events at such a granular level and a bunch of sabermetric concepts are being flipped on their heads (ie DIPS theory) thanks to the integration of this new data. The world of baseball analysis is evolving along with the appreciation of these non-traditional aspects of the game, the ignorance of which has driven much of the disdain for sabermetrics from the scouting side of the ledger.

A perfect example of this is the data for catcher pitch-framing that has taken off in popularity in recent years. The impact of pitch-framing was well-known among baseball people for generations, but widespread acceptance of just how critical pitch-framing can be didn't occur until we had a means of measuring the skill and labeling those catchers who were good or bad at it. The objective measurement of pitch-framing helped to quantify this impact and spread awareness to the field of baseball-obsessed analysts who had failed to appreciate this critical part of the game.

Another example is the distance of a pitcher's release point. Engineers much smarter than myself have designed ingenious applications for the training of baseball players, yet it is overwhelming how often I have seen the fundamental mistake of assuming that pitcher's release point is 60 feet, six inches from the plate. That might be the distance from rubber to plate, but no pitcher releases the baseball right on top of the rubber, and the ability to extend one's release point effectively increases the perceived velocity of his pitches. A more realistic design would incorporate this reality, or at least assume a release that conforms to the average pitcher (approximately 54-to-55 feet from the plate) rather than a pitching machine placed on top of the rubber, yet such aspects are sometimes overlooked by those who are years removed from their last meaningful moment on a baseball diamond.

DIPS theory is another example. Ask any pitcher, and he will say that he has an impact on the quality of contact on balls in play. However, the tools of box-score stats lacked the sensitivity to detect such things, so the sabermetric world threw up their hands and said that such things were out of the pitcher's control, yielding a paradigm that focused on “true” outcomes yet ignored most of the events that took place on the field of play. Now we know better, thanks to the proliferation of batted-ball data and the few sneak peeks that we've had into the world of HITf/x. The sabermetric world can finally detect these elements that were right in front of our faces the whole time, suddenly able to catch the signal from the noise because we're finally pointing better microscopes in the right places.

PITCHf/x systems and other technology is becoming more widespread in minor-league stadiums, but it is virtually nonexistent (for obvious financial reasons) at the amateur levels, so the radar gun will still carry its considerable weight in scouting circles. This is a problem, since the main injury-related issues with velocity-building and the radar-gun obsession take place at the amateur levels, where workloads are split among multiple teams and multiple coaches, and where the risk of abuse is the highest – and that's before getting into the growth aspect where the body is changing so rapidly during these pubescent ages that it is least-acclimated to the tasks being assigned to the throwing arm.

So the radar gun was, is, and will continue to be hugely important in the evaluation of baseball players, and this will be the case until we see a cultural shift that appreciates that A) velocity can be built, B) that it can be built safely later in a player's career, and C) there's a time in a player's career to ratchet up the power, and it's not during puberty. In fact, the ideal application of C) changes for every player based on mechanical and physical progression through functional strength and stability.

The max-effort-every-pitch paradigm was borne out of the modern bullpen, and it's hard to argue with the results. I'm not as worried about the approach from relievers due to the low-exposure outings and relatively light workloads (though the sheer volume of outings might be another factor), but the idea of extending that concept from the 20-pitch guys who enter the game in the late innings to the 100-pitch guys who are expected to last three turns through the batting order is patently insane to me; yet that's the norm in MLB 2016. Now trickle on down the pipeline, and you have amateur coaches trying to take cues from major-league patterns, creating a minefield of player development mayhem in Little Leagues across the country.

The greatest compliment that I receive – typically after a speaking engagement—is from baseballholics who have obsessed on the game for their entire lives yet find themselves watching baseball in a different way, opening their eyes to some of the mechanical aspects that are central to my particular approach to pitching. My goal in these presentations—as well as in these articles—is to give the audience a tool kit to use when watching the game so that they can see these aspects for themselves. It might be based in subjectivity, but I firmly believe that it takes an appreciation and awareness of some of these aspects in order to better inform our judgments of the stats, and to better understand how and why certain numbers have merit.

Sabermetrics can be a crutch for real analysis, as assumptions of “luck” and “true talent” are often used when the numbers at hand don't jive with performance, but that assessment has taken a different shape as new tools are put at our disposal. The reality that players are dynamic, with ever-changing skill sets and no real ceiling to their potential for improvement, is both complicated and frustrating for even the analytically-minded, as the human pull to provide final answers to some of these questions is in direct opposition to this rapidly-evolving reality of any athletic endeavor. We want to believe that Carlos Gomez sucks, or that we were wrong and he's actually really good, or that we were wrong again and he actually sucks, but nobody is lining up to grasp the very real possibility that he sucked once, was great once, currently sucks again and could be very productive once again at some point in the future.

The radar gun has the potential to marry the two sides that have long struggled to understand one another's perspective, giving statheads an objective measurement of on-field performance to appreciate and quantify baseball-specific skills. Right now it's the only game in town, in the sense that StatCast and other on-field measurements are still largely hidden from public consumption but PITCHf/x data is readily available, so it is critical to acknowledge the missing data rather than gloss over those aspects which don't compute, and to temper the veracity of sabermetric conclusions until better data is integrated into the system.

We are on the verge of something huge, such that theoretical debates over what constitutes a win will seem meaningless when comparing the functional fielding range, throwing velocity and transfer timing of fielders; it will seem meaningless when we're measuring the exit velocity, spray angle and angle of elevation of every batted ball; and the shape of those debates will change as we improve the data set with on-field measurements such as those made by StatCast, Trackman and Sportvision's f/x suite of baseball analytics.

The future of baseball analysis is exciting, but first we must shed the shackles of previously-accepted theories and be willing to shift the paradigm as technology reveals new truths about the game. Letting go of the past can be difficult, but it's necessary for us to fully embrace the future, and any scientist worth his salt will gladly evacuate previously-held theories in light of new evidence that surpasses the precision and accuracy of past experiments. The baseball world is forging ahead, but the pace of progress will be largely dictated by our collective ability to integrate new information with the firmly-held beliefs of the past.

Doug Thorburn is an author of Baseball Prospectus. Follow @doug_thorburn
Click here to see Doug's other articles. You can contact Doug by clicking here