March 13, 2015
Scouting With Plate Discipline
Not long ago, in the empty hours that fill scouts’ time between batting practice and the first pitch, I was speaking with (read: listening intently to) to an experienced scouting director. The topic of plate discipline came up. As anyone who closely follows the BP prospect team has figured out, I fall at the “extremely important” end of the spectrum when it comes to using plate discipline to evaluate prospects and predict their future success. Because of that, it didn’t take much prodding to get me to rave about a prospect I had recently seen from this scouting director’s system. The prospect, along with many strong skills, had a fantastic eye.
The scouting director’s response surprised me. He informed me his organization values plate discipline extremely highly, to the point of actually considering it a sixth tool (something we’ve seen the old-school camp dismiss in the past). I love the concept, but was stupefied because the organization in question is not one we typically think of as sabermetrically friendly. I assumed I was at a more extreme end of the spectrum on this topic, but here my views were accepted with arms open wider than I had anticipated.
Plate discipline, of course, is not a new concept in scouting. The way we view player value has brought about the current on-base renaissance within the big-league game, and that has had a trickle-down effect into the scouting world. With a better understanding of the value of on-base ability, we have changed how we evaluate prospects in some regards, at times excusing larger holes in a hit tool, for example, so long as it comes with strong on-base skills. The ability to avoid outs with better plate discipline can make up for some flaws in one’s ability to hit his way on base.
But beyond the direct value of a walk or a count well worked is the way that plate discipline affects the likelihood of a player’s other tools manifesting at higher levels. If a player has a great hit tool but absolutely no plate discipline, he’s less likely to make the most of that hit tool. These skills don’t exist in isolation. Which means plate discipline isn’t just a tool (like speed or power) for the player; it can be a tool (like a stopwatch or a radar gun) for the scout.
So while exceptional plate discipline isn’t nearly so fun to evaluate as the flashier tools—scouts rarely rave about the guy with a 40-hit tool who makes up for it with a good eye at the plate—it is perhaps as relevant to the final product as, say, elite bat speed or amazing feel for the barrel. Those two can be blinding, especially in regards to the hit tool, generally regarded as the most difficult tool to judge.
With so many variables that go into evaluating the hit tool—pure bat speed, swing mechanics, bat path, approach, pitch recognition, among others—even the most veteran scout would be hard-pressed to speak in certain terms when talking about the future of any prospect and his ability to hit major-league pitching. Simply put, it’s tough to be sure that any prospect will hit at the major-league level. To help, we sometimes look at numbers to help paint a bigger picture. With limited looks at players, it can be helpful to use stats to support in-person looks. If a player’s numbers support what a scout has seen for himself, it’s safer to say that what he saw was an authentic representation of a player’s true ability. If the numbers say something else, a deeper look is required.
Given the notorious difficulty of evaluating hit tools under any circumstances, might numbers help us here, too?
The biggest problem with minor-league numbers is noise, a term used in statistics to describe when numbers are affected by so many additional factors that they stop reliably describing what they attempt to describe. A statistic like K/BB ratio typically does a good job of representing plate discipline and approach at the major-league level. But in the minors, there are so many additional factors—so much noise—like age/level combinations, extreme variance in hitting factors, a lack of advance scouting, different pitch selection than in the majors, and so on, which makes it less descriptive than we’d like. Any number of these factors can distort a player’s K/BB ratio.
Our projection system here at Baseball Prospectus, PECOTA, goes in-depth to account for these factors and quiet down the stats. It’s not a simple task, but understanding its methodology can help scouts put the value of plate discipline as a scouting tool in perspective.
The formula that goes into the PECOTA projection system is complicated, and requires more depth than scouts need to understand in order to use it, but I’ve asked Andrew Koo of our stats and research department to help explain what goes into the portion that takes plate discipline into consideration. —Jeff Moore
In everyday boxscores, numbers represent outcomes: 2 BBs means a batter walked twice. But outcomes can be noisy: Maybe a bird flew in front of the pitcher just as he threw one of those ball fours, or maybe first base was open and the pitcher was on deck, or maybe the umpire was mad at the catcher, or maybe the batter is a 26-year-old in A-Ball and scares the wits out of the pitcher. What we want to do in sabermetrics is measure talent, or skills, based on those outcomes. Then, as best we can, we want to predict future outcomes based on the skills. Two walks in one game don’t mean the batter has a good eye, and they don’t mean he’ll repeat the feat the next day. But if we can say the batter has plate discipline—and that the plate discipline was the underlying skill of those two walks—then we get more confident.
But skills are also abstract. What exactly is a hit tool? What exactly is plate discipline? We can explain these idea in words, to an extent, but not with the clarity of numbers with which we can say “the batter walked twice.” BB:K ratio is one attempt to quantify the skill of plate discipline, but walks and strikeouts are not solely a reflection of that skill: the skills of pitch recognition and contact ability (i.e. swing-and-miss propensity) would also show up in a batter’s walk and strikeout numbers. Connecting outcomes to skills as confidently as possible is the challenge of any projection system, especially PECOTA, which attempts to quantify multiple skills for each player. The ultimate goal would be projecting players solely based on their skills. If we knew, for example, a batter’s ability for discerning balls from strikes, both precisely and quantitatively, then we could project his plate discipline (i.e. O-Swing and Z-Swing rates) with high certainty.
(In the major leagues, we can evaluate these skills much more accurately with PITCHf/x. In the minor leagues, however, BB:K is all we have; complete minor-league PITCHf/x isn’t available publicly, so walks and strikeouts become our less-granular proxy.)
PECOTA contains the primary elements of most projection systems: regression, aging curves, and park and league adjustments. But PECOTA also distinguishes itself with its Comparable Players system. To project Johnny Orangeseed, PECOTA isolates and quantifies several of his skills, then compares his rating in such skills to those of Kenny Pearbear, Lenny Grapehead, Mike Lemonstand, and thousands of others. (It also factors in attributes such as height, weight, age and handedness.) Once PECOTA finds the most similar players* to Orangeseed, those comps help determine his projection. The key here is that comps are based on skills; two players could both hit .300 with a .900 OPS, but not necessarily with the same skill set. We believe players with similar skills will produce similar future performances more reliably than two players with similar raw outcomes will. This idea is, in a lot of ways, what scouting is based on, too.
* If you require a mathematical definition of “most similar,” PECOTA measures Euclidean distance in n-th dimensional space. So if there were only two skills per player to compare, envision a 2-dimensional space (a square) with thousands of points plotted as (x-skill value, y-skill value), each point representing a player. A player’s comps are the other dots closest to him. Now generalize that concept to, say, a 9-dimensional space.
So what are these skills? Plate discipline is one, approximated with walks and strikeouts. Contact rate is another, which can be approximated closely with strikeouts per plate appearance. Other skills include power (quantified by slugging on contact—SLGCON) and speed (triples per extra-base hits, and other metrics). Essentially, PECOTA tries to quantify the hit, power, and speed tools, then project a player based on comparably tooled players. After all, two players with the same skill set and the same degree of competency in those skills should perform similarly. (This assumes, of course, we are measuring the correct skills. Some predictive skills--like makeup--might be beyond our capabilities.)
The ultimate goal here is identifying current prospects who have a high probability of succeeding as major leaguers. Plate discipline plays a significant part here: a prospect with poor plate discipline skill, as regarded by PECOTA, will likely generate comparables who also had poor plate discipline. If those comparables went on to have disappointing careers, PECOTA will know it and will adjust expectations downward.
To illustrate this for current prospects, our lead stats developer Rob McQuown ran UPSIDE scores, which are heavily based on the comp system, for all prospects on the BP Top 101. Here are the worst-projecting hitters on that list (min. 650 minor-league PA):
According to PECOTA inventor Nate Silver’s old definitions, prospects with an UPSIDE below 25 are “average prospects,” with “some chance of a meaningful major-league career, but more likely to end up on the major-league fringe.” These are batters PECOTA believe may fail as major-league hitters (but not as defenders!) because of a poor minor-league showing in plate discipline, power, contact ability, speed, or any combination of. Indeed, many on this list garner a low UPSIDE because of plate discipline and contact ability—staples of the hit tool. Jeff will delve into the potential downfalls for a few of them. —Andrew Koo
The goal of using PECOTA, and more specifically isolating UPSIDE, is to give scouts a predictive tool at their disposal besides just their eyes and experience. Not in place of those two things, by any means, but a historical context to go along with what their instincts are telling them. We’ve all seen reports on players that say something along the lines of “the tools are there if he can just refine his approach.” Take these quotes, taken directly from our Top 101 scouting reports, for example:
So we all intuit the connection between approach and hit tool (and success). But what kind of odds are we talking about?
Despite their risk, all the players in the table above are talented, and in some cases, extremely talented. Otherwise they wouldn’t have landed in our Top 101. But as we know, often times tools and raw talent can overshadow likelihood of success, especially in rankings. It’s also not fair to say that just because a player’s UPSIDE score falls below 25 that he is destined to fail. What it’s telling us is that it’s worth a closer look.
*UPSIDE scores are easily accessible to subscribers on BP player pages, along with player comparison scores.
With a lot of these players, a closer look shows mitigating factors in the player’s defense. PECOTA penalizes players who have struggled and have had limited time to make up for it (McKinney), players who profile to play a corner position but don’t have big-time power (Piscotty), players whose only power showing comes in a hitter-friendly league (Renfroe, Robertson) or players with massive strikeout rates (Frazier, Alfaro).
Some of those attributes are quite correctable. McKinney, for example, has a limited resume, so if he comes out strong next year his score could rise more rapidly than a player with a larger sample. Renfroe has only had success in the Cal League and didn’t hit for as much power after a midseason promotion to Double-A, but a few early-season home runs to add to his limited resume could get him out of the danger zone in a hurry. Others, like Clint Frazier’s extreme strikeout rate, will be much more difficult to correct.
For a few of these players, however, this score fits with the trends we see both on the field and in the more traditional stat lines we look at.
Of all the prospects with low UPSIDE scores, Mondesi has perhaps the least projected power. With just a 6 percent walk rate and below-average power, a potential 40-50 hit tool (.250-.270 batting average) makes for a potentially poor offensive player.
Mondesi is an extreme case in many regards, but the most obvious is his age, which even if he repeats a level (a likely scenario) will still make him one of the younger players in the Carolina League. But despite his youth, he is approaching the point where it is unlikely he’ll improve his plate discipline.
Below is the average decrease (weighted by plate appearances) in walk rate (in percentage points) from minor league age/level to the majors:
For players who are young for their minor-league levels, the drop-off in walk-rate is not nearly as steep upon reaching the majors, which is one positive indicator for Mondesi. However, the drop-off is still there, as it is for virtually all minor-leaguers of all ages at all levels. Extrapolating, then, we see in Mondesi player who simply might never have major-league-quality plate discipline. This is a large part of why PECOTA penalizes players with poor plate discipline numbers, even when those players are young for their levels. Despite his youth, the numbers strongly suggest that what we see from him this season is more or less what we’re going to get from him in the majors.
Even without an offensive game, Mondesi is a prospect, and as one of the youngest prospects in our top 101 (and in his leagues) there is plenty of developmental time remaining for improvement. But the historical track record of players with his offensive profile is frightening.
At bat, however, Almora’s free-swinging approach is perhaps the reddest of flags on this list. Not only does he sport the lowest walk rate of any of the top hitting prospects in the game (by a notable margin) but it’s a lower rate than any current major leaguer. More strikingly, there are literally no current major leaguers who walked as infrequently in the minors as Almora has as a professional. Lastly, it’s trending the wrong direction, declining at each full-season level he’s reached, from 6.3 percent in Low-A in 2013 to 3.1 percent in High-A at the start of 2014 to 1.4 percent in Double-A after his promotion.
Almora will be 21 this season and is likely slated for Double-A, and as the above chart shows us, even 21-year-olds in Double-A can expect to see their walk rates go down upon reaching the majors. His off-the-charts aggressiveness after a promotion last season can be chalked up to small samples and adjustments, but it’s going to be extremely important for Almora to buck his current trend this season if he has any hope of turning into a .300 hitter in the majors.
Almora has yet to show much power, but flashes the ability to eventually have average over-the-fence pop. With the potential for 15 home runs and plus up-the-middle defense, Almora is a top 101 prospect, but in order to justify his lofty ranking, he almost has to hit around .290-.300. The only players in the majors who have done that over any kind of extended period of time with a similarly extreme level of aggressiveness are Howie Kendrick and Pablo Sandoval, two notoriously free swingers with elite bat-to-ball skills. Almora’s pure hitting ability might be in their class, but he will have to improve his patience and approach to reach even their low walk rates. If he falls short of their elite hitting ability by even a little bit, we’re talking about a player who makes an obscene amount of outs and will fall well below his ceiling.
So what’s the problem?
As a 21-year-old heading to Double-A, Williams is in the same boat as Almora with something to prove in terms of improving plate discipline. He also has some swing-and-miss issues to worry about, where the PECOTA comparisons are even more damning. For all of his aggressiveness, Almora manages to get the bat on the baseball more often than not, even if it’s poor contact. The same can’t be said for Williams, whose 29 percent strikeout rate in High-A ball last season is a concern, as was his 32 percent rate after a promotion to Double-A. Williams needs to not only be more patient but make more contact if he wants to have any chance to do so in the majors.
While they’re not perfect stats, Williams also compares poorly to current major leaguers in more traditional measures like K/BB ratio, K rate and BB rate, falling in the bottom fifth percentile in all three categories when compared to all major leaguers over the past three years. There are no current major leaguers (good hitters or bad) who with such poor minor-league rates in those three measures as Williams. He will get to the majors, but in order to stay there, he will need to be successful using an approach that no other current major leaguer is getting away with.
Unlike our other red flag candidates, Williams does project to have at least average power. That’s raw power, of course, and its in-game adaptation will be dictated by his ability to get good pitches to hit, but it does project to be better than the other players I’ve mentioned. Williams is also not the same caliber athlete and doesn’t project to be a defensive asset. He’s a plus runner in a straight line, but the Rangers have already shifted him to left field, which will be his ultimate destination. It’s not that he can’t refine his approach, or succeed with it as it is, but the historical odds are stacked against him.
That last line is important. It’s not that any of these prospects can’t reach their potential, specifically in regards to their hit tools. Could Raul Mondesi hit .290? Could Albert Almora hit .310? Absolutely. Having seen all three of our at-risk players in person, I’d be tempted to grade their potential as elite, too.
But much like we differentiate between raw power and game power, we need to look at the hit tool in the same light, and this is where scouts can use PECOTA to help supplement what their eyes are telling them. Nick Williams has the potential to be a 65/70 hitter, but his extreme approach, and the fact that there are no current major leaguers getting away with a similar approach, hurt his odds. Almora could be a .300 hitter, but our data tells us that he’d have to be among the most elite players in the game in terms of pure bat-to-ball ability to get away with his lack of plate discipline. And he might be. But it’s important for scouts (both in the industry and our own here at BP) to realize that that’s the statement they are making if they decide to put a high hit tool grade on him, or others with similar approach issues.
Numbers will never replace the work of scouts, no matter how in depth our projection systems go or how much minor-league data we get. But the numbers can help us tell part of the story, particularly with plate discipline and particularly with the hit tool. The can, at least, help warn us about which hitters will have more to overcome than most. —Jeff Moore