What would we learn if we could see what an MLB team’s scouts saw? For the first time, we can: A former member of the Cincinnati Reds front office provided The Ringer with a copy of the Reds’ scouting database from between 1991 and 2003, consisting of more than 73,000 reports. Throughout this week, we’ll be using this newly declassified scouting gold mine to analyze old-school scouting’s strengths and weaknesses, profile players who defied the scouts’ expectations, and examine how scouting has evolved in recent years. In Monday’s Part 1, we crunched the numbers on how well scouts projected players. In Wednesday’s Part 2, we talked to four players about how they defied the scouts’ expectations and asked the most unerring Reds scout to explain his success. In Friday’s final installment, we explore how the scouting side of the industry has changed to address the deficiencies of two decades ago.
In his postscript to the paperback edition of Moneyball, Michael Lewis fed fears that the sabermetric movement would soon make scouts extinct. Discussing J.P. Ricciardi, a Billy Beane lieutenant in Oakland who landed the Blue Jays’ GM job in late 2001 and brought on Baseball Prospectus’s Keith Law, Lewis wrote, “Ricciardi, the new GM, had done what every enlightened GM will eventually do: fire a lot of scouts, hire someone comfortable with statistical analysis … and begin to trade for value, ruthlessly. … What happened was that the Jays went, overnight, from being a depressing group of highly paid underachievers to an exciting team. They were younger, cheaper, and better.”
Lewis seemed to be suggesting that firing scouts was smart. But with one exception—more on that in a moment—teams haven’t acted as if that were true. In 2016, Rob Arthur and I traced the rapid expansion in the size of front-office analytics staffs in a piece for FiveThirtyEight. Although we were focusing on the increase in statheads, we noticed that teams hadn’t actually let go of scouts to make room for more quants. In their insatiable pursuit of all kinds of data, they had hired more of both. Using team media guides and websites, we determined that the typical team had added about 10 scouts between 2009 and 2016.
I’ve updated the scout counts for 2019 in the table below, which omits the latest figures for the Orioles, whose media guide is still in production. These numbers represent approximate tallies of full-time, traditional scouts, excluding department directors, assistant directors, administrative staff, and video analysts but including crosscheckers, supervisors, and some special assistants and coordinators, along with rank-and-file amateur, pro, and international scouts.
Although the scout counts haven’t increased nearly as quickly as headcounts in R&D and, more recently, player-development departments, the average team has added about three scouts to its payroll in the past three years and now employs 32 percent more scouts than it did a decade ago. Fifteen years after Moneyball came out in paperback, teams still haven’t forsaken scouts. They have, however, changed how they use them. And there’s still some chance that Lewis’s forecast could come true.
In Part 1, we went over a host of problems that plagued scouting reports from the pre-Moneyball era. Habitual grading differences between generous and conservative scouts. An inherent risk aversion that compressed players’ scores, leading to bunching toward the center of the scouting scale and a loss of analytical value. An ill-considered (or unconsidered) weighting of tool grades to produce an overall rating. And plain old primitive player evaluation, which coupled a poor appreciation of the predictiveness of certain stats with the myopic perspective of an era when batting average still drove decisions and catchers were valued more for their arms than their receiving skills.
Law, who has evaluated prospects for ESPN since leaving the Blue Jays in 2006, invokes the example of Bobby Witt, who walked five batters per nine innings in the majors after being drafted with the third overall pick in 1985. “I got Witt’s numbers from college,” Law says. “They’re comically bad. He couldn’t throw strikes at all. … There’s no way that guy gets drafted in the top 10 today, maybe even lower. He could just throw really hard. It was sort of, ‘Wow, this guy throws harder than anybody we’ve seen before. Take him.’ That just doesn’t happen anymore, because we recognize that the failure rate in taking guys like that and teaching them how to throw strikes is not very high.”
Witt wasn’t a total bust; although he was a below-average starter, he still lasted 16 seasons and produced roughly 15 WAR. Plenty of toolsy picks from that period panned out worse than that, including the Reds’ 14th overall pick in 1997, high school third baseman Brandon Larson, about whom one Reds scout said in ’97, “He’s going to develop into a fine hitter.” (Narrator voice: He wasn’t.) Two years before that, the Phillies had used their 14th pick on ultra-toolsy outfielder Reggie Taylor, another player Cincinnati’s scouts loved (and whom the Reds later acquired). The Phillies could have had Roy Halladay instead. “I do think teams have learned, certainly over the last 15 years … ‘Hey, those profiles, maybe they don’t work,’” Law says. “The Reggie Abercrombie, Greg Golson, ‘He’s a great athlete, we’ll teach him to play baseball’ [model] doesn’t usually work.”
During his time in Toronto, Law pushed to add plate discipline to the Blue Jays’ reports, both because the team lacked reliable pitch data for amateur players and because it’s a vital trait that was seemingly slipping through the cracks of the standard five-tool system. He hoped that merely asking the scouts to rate players’ plate discipline would send the signal that selectivity was valued by the Blue Jays brass. “If I don’t ask you to think about plate discipline, you may simply not think about it at all,” he says. Law mentions that some teams have also started asking their scouts to rate a pitcher’s fastball command and command of his secondary stuff separately, rather than using one blanket command rating.
FanGraphs prospect analyst Kiley McDaniel, who has worked for four major league teams in front-office and/or scouting roles, says that some clubs now ask their scouts to provide probability ranges in addition to the median tool grades. That extra nuance allows scouts to distinguish between a polished player with a 40 hit tool who’s already overachieved and has no chance to become an average big leaguer, and a rawer, higher-ceiling prospect who also has a 40 hit tool but has a 10 percent chance to put it all together and be an above-average big leaguer down the road.
Teams are also going to greater lengths to adjust for scouts’ tendencies. “We’ve always tried to squeeze every ounce of information from the scouting reports,” says Orioles assistant GM Sig Mejdal, former right-hand man of Jeff Luhnow in St. Louis and Houston. “Even though they’re asked to use the same scale and you define the scale, they’re human beings, and so some are going to use the entirety of the scale while others will be clustered around the center. … It’s a straightforward exercise to … normalize the scouting grades and then from there try to see what predictive ability they have.”
According to Hank Sargent, the former Reds scout profiled in Part 2, some scouting reports were still handwritten and transposed in the office when he arrived in 1997. He believes the team switched to full computer entry the following year. Even when Law was in Toronto a few years later, he says, “We had very little in usable electronic form.” Given those conditions, what Mejdal describes as a “straightforward exercise” was, according to Law, “just unthinkable at the time.”
Mejdal notes that grades won’t be visibly bumped in a team’s internal system from, say, a 55 to a 59.7 because a scout is a tough (or tentative) grader. Rather, that adjustment occurs automatically when the report is incorporated into a valuation or projection. This process isn’t universal, though, and some teams still prefer to keep stressing the importance of using the whole scale rather than resign themselves to getting garbage in and doing their best on the back end not to get garbage out.
It’s common today for teams to take more care than they used to with the way individual tool grades are combined into one rating, even before that rating is tweaked to account for each scout’s numerical leanings. “When I was in Toronto, they just took the five tools and averaged them together, and that was the overall grade,” Law says. “I’m like, ‘That’s the dumbest thing I’ve ever seen,’ especially because J.P. made it clear he didn’t care if they could run. Could they hit? Could they hit for power? Could they get on base? It was basically all we wanted, so why are we averaging on the speed score?”
Mejdal says the Cardinals and Astros didn’t pay much attention to OFP during his time with those teams, but he notes, “People from other teams that I’ve spoken to, they all have adjusted OFPs of some sort, whether it’s based on position or maybe the weighting is more sophisticated.” It’s not unheard of for teams to weight by title or role or, depending on the report, to place extra importance on the input of a scout who specializes in hitting or pitching.
One might think the latter approach would be an even more fruitful source of scouting insight. Maybe scouts’ skills are as varied as those of the players they evaluate; a certain scout might not just be better at projecting pitchers or hitters, but a savant at rating curveballs or catchers or players from a particular system where he’s well connected and can do a better job of judging off-the-field factors. In theory, it would be beneficial to know a scout’s true talent at assessing every trait. In practice, though, it’s difficult to draw such fine distinctions statistically.
As Branch Rickey associate Fresco Thompson wrote in 1964, “Just as it takes some time to learn whether or not a player is going to make the grade, so it takes time to find out a scout’s value.” It takes years to determine which players will get good, and a random bad break can kill a player’s career in a way that would appear to reflect poorly on a scout who actually nailed the predictable part of the player’s career path. Despite the possible uses of a more sensitive scouting of scouts, Mejdal says that “in my experience it has been difficult to do that in a data-driven way.” He adds, “I think everybody through experience has their own subjective confidence in different scouts relating to different aspects of the game. … My guess is that many teams incorporate that in a less systematic way.”
McDaniel concurs with Mejdal’s assessment. “The scouts we needed info on the most were covering lower levels,” he says via Slack, “so we needed 4-5 years to grade their reports—probably more, but that’s the minimum. And by the time you can grade their first year of reports, they’ve gotten four new contracts and a promotion or been fired, and their evaluating skill has probably changed or their biases/specialties may have shifted.” On top of that, the high-level executives who may make decisions based in part on scouts’ input generally already have a decent experiential sense of the strengths and biases of each evaluator, so the analytical lift may be of limited utility anyway.
“The scouts add tremendous predictive ability when we’re done squeezing what we can out of the numbers,” Mejdal says. But the “numbers” are coming closer all the time to replicating—and in some cases, improving upon—the information that scouts have historically supplied. For teams, that’s an opportunity. For at least some scouts, though, it may prove to be a threat.
Royals special assistant to the GM Mike Toomey, who started scouting in the 1980s, remembers watching pitchers alongside longtime Senators and Rangers scout Joe Branzell, who was one of the first scouts to use a radar gun. “If he was throwing 83 on Joe’s gun, the guy could be throwing up to 90,” Toomey recalls. “And I would say, ‘Hey, what’s this guy throwing, Joe, about 88, 90?’ ‘Ah, no way, shit, here it is right here, 83.’ They calibrated those guns, so if you were looking over his shoulder, I’m thinking to myself, ‘Shit, Joe, there ain’t no way this guy’s throwing 83.’” Even that imprecise system was better than what was available before; Toomey says the best means of judging velocity and late life before radar guns was noting whether the batter swung and missed. That may have worked well with high-level hitters, but it made it more difficult to get a good read on whether amateur arms were actually good or just feasting on inferior competition.
Precision isn’t a problem for modern scouts. TrackMan radar is installed in virtually every minor league park and 54 Division I schools, silently recording the speed, spin, and trajectory of pitched and batted balls with greater accuracy than any human could hope to. The Astros have actually encouraged their scouts—at least those they haven’t let go—not to bring their radar guns to games. Recording the redundant readings could only distract the scouts from the dwindling ways in which the eye can add value.
Minor league pitcher Michael Plassmeyer, a 22-year-old lefty who was traded to Tampa Bay last November in the Mariners-Rays swap of Mike Zunino and Mallex Smith, is a case study in the potential and peril of present-day scouting. Plassmeyer was part of the first University of Missouri squad to play in front of the school’s newly installed TrackMan system in 2017. Collectively, its pitchers posted a 3.67 ERA, nearly the lowest figure for a Mizzou staff since the Max Scherzer–led 2006 unit, but Plassmeyer wasn’t one of the pitchers responsible for that impressive run prevention. The 6-foot-2 southpaw, a St. Louis native in his sophomore year, finished with a 4.83 ERA, which was an improvement upon his unsightly 5.12 as a freshman. But even that lower mark was misleading: The lefty’s ERA against SEC opponents was 7.92, and he lost his rotation spot midway through the season.
“I didn’t have a good put-away pitch,” Plassmeyer says. As a sophomore, his heater hovered near 90, and he allowed seven homers, too many for his 54 innings pitched. Yet Plassmeyer clearly had control (as indicated by his team-low walk rate) and some ability to miss bats (as demonstrated by his batter-per-inning strikeout rate). He just needed the right instruction and technology to unlock his latent potential.
In the summer before his junior year, Plassmeyer went to P3 (Premier Pitching & Performance), a training center in St. Louis. He followed a weighted-ball and long-toss program, and he helped speed up his arm action by working on explosive movements in the weight room rather than lifting heavy. He added a tick or two to his fastball, refined his slurvy breaking ball using the facility’s Rapsodo system—another pitch-tracking technology—and learned that his fastball had a high spin rate and would be most effective up in the zone. Prior to seeing his spin rate, he had typically thrown low, as tradition dictated. “I definitely needed the numbers,” Plassmeyer says.
Before his junior season started, Plassmeyer thought the dots he’d connected and the arm strength he’d gained would give him a chance to get drafted, despite his 4.98 career college ERA coming into the year. But by heeding advice from P3 and Mizzou statistician Matt Kane about aiming his fastball higher and throwing his breaking ball more often—he estimates almost 40 percent of the time—he surpassed his own expectations. Plassmeyer made 14 starts, averaging 6½ innings apiece, and he walked only 17 batters while striking out 103, tied for the seventh-highest total in the SEC. His ERA fell to 3.05.
Coming off that strong season, Plassmeyer got a call from the Mariners on the second day of the draft, even earlier than he’d dared hope. Seattle had selected him in the fourth round and would pay him a $425,000 bonus. “Everybody was expecting Plass to be a Sunday starter, a hold-you-over-type guy,” Kane says. “And all of a sudden he is a fourth-round pick because he’s gone out there and dominated with 90 miles an hour from the left-hand side.”
Plassmeyer probably wouldn’t have posted those stats without his high-tech makeover, but even if he had, he almost definitely wouldn’t have been drafted as high as he was without what that tech revealed about his pitching process.
When the draft draws near, coaches at TrackMan-equipped colleges can advocate for their players by volunteering data that illuminates aspects of their performance that are hard to see from the stands. “Part of it is convincing that area scout around here that this is a guy that deserves an opportunity in pro ball and being able to provide that [information],” Mizzou head coach Steve Bieser says. “They’ve got the radar gun and they can see the velocity, but being able to provide some of that extra information of spin rates and even pitching tunnels of how his pitches pair” can pique a scout’s interest in a previously low-profile pitcher like Plassmeyer, a Bizarro Bobby Witt whose speed doesn’t wow anyone on its own.
Even if the scout isn’t swayed, his bosses probably will be. Although a scout may be reluctant to ask his front office for data on amateur players, for fear of looking like he’s blindly subscribing to what the numbers say, major league teams do pay for access to college TrackMan data, and many of them weight it heavily in their draft models. Some front offices even pass that info along to their scouts.
In Plassmeyer’s case, “The scouts were like, ‘We love the analytics and the stuff behind it,’” he says, adding, “I think it makes them a little more comfortable and solidifies that I can keep on getting outs at the next level.” A member of the Mariners’ front office confirms that the lefty’s TrackMan readings were part of his appeal. Plassmeyer continues, “Maybe if [the data] weren’t there, I would look like a regular 88-90 [mph] lefty.” In his first taste of pro ball, he definitely didn’t: In 13 games and 24 innings in the Low-A Northwest League before his offseason trade, Plassmeyer struck out 44 batters and walked only four, giving him the highest strikeout-minus-walk rate of any pitcher in the league with a minimum of 20 innings pitched. He threw more than 70 percent strikes, and he finished with a league-low fielding-independent pitching (FIP) mark of 0.93.
It’s much easier to evaluate performance prospects like Plassmeyer today than it once was. Some aspects of performance that previously seemed nebulous and may have taken guts for scouts to get giddy about now have objective backing. Last year, Law says, no one hesitated to hand out a 70 to eighth overall pick Carter Stewart’s curveball, which had one of the highest spin rates on record for an amateur player. “If you look at it, it’s got velocity. It looks right. Oh, the data back it up with a huge spin rate? OK, sure. It’s very easy to stick your neck out. No one’s gonna come after you and say, ‘Why’d you do that?’” Modern scouting is still largely based on comparing current players to past players and established baselines, and extrapolating performance from there, just as it always was. As Sargent says, “There just weren’t as many categories of comparison.”
The downside for scouts is that they’re no longer the sole source—or even the preferred source—of vital information that teams trusted them to provide for a century. “In the data-heavy world, scouts provide more value on the margins in separating players who have similar statistical profiles, similar TrackMan or Statcast-style data,” Law says. “Well, why do you like this guy better than that guy? You might not even necessarily be able to articulate it, but … you may be recognizing something you don’t know why you’re recognizing. ‘This player reminds me of this other player or this category of players that I’ve seen over 15, 20 years.’”
Another complication is that player development has become even less predictable than before. New techniques and technology allowed Plassmeyer to take an unexpected leap, and in his first two months in the Mariners’ system, his previous exposure to tech and concepts such as tunneling and spin-based location enabled him to skip steps in his development, working with the team to tweak his release point in a way that enhanced his changeup and allowed him to throw it more often. “Rather than having to start from square one, I got a head start on guys,” he says. “They can have more advanced meetings with me.”
As a consequence of teams’ intervening earlier and more actively in tweaking athletes’ technique—and of players like Plassmeyer embracing a progressive approach—Missouri pitching coach Fred Corral says, “The learning time frame is decreased. They get to their destination sooner.” We’ve entered an era when pitchers are adopting different pitches at such a scale that pitch selection has shifted on a leaguewide level, young players are producing at historic rates, size is less weakly correlated with power production than it has been in decades, and teams are deploying both pitchers and position players in new ways that affect which skill sets are preferred or required. Projecting players is getting easier in some respects, but in others it’s more of a mystery than before.
“If you don’t adapt, you die,” Sargent says. “And I think that if you have access to information that helps paint a more vivid picture, then not using it is a mistake.”
And potentially a professionally fatal one. Although scouts are more numerous than ever, that doesn’t mean their status is secure. Headcounts have increased, but not across the board: International scouting departments have expanded as the game has grown in markets that have embraced baseball to a greater degree in recent years—including Korea, Taiwan, Brazil, Curaçao, Europe, and others—but regular-season MLB advance coverage has dwindled as almost all teams have switched to stats and video to scrutinize upcoming opponents rather than sending scouts to see them in person.
The greatest threat to traditional scouting is the Houston Astros, who have laid off most of their scouts over the past few years. Between 2016 and 2019, most teams increased the size of their scouting staffs, but the Astros slashed theirs from 52 to 18, which is less than half the size of any other club’s. The Astros are the only team in baseball to have done away with its pro scouting department, which was once overseen by Kevin Goldstein, who now fulfills a broad range of duties as the team’s special assistant to the GM, player personnel. In other words, with rare exceptions, no Astros scouts evaluate pro players in person, and the amateur scouts’ assignments are often dictated by data that directs them toward the most statistically intriguing talents.
In August 2017, when the Astros informed eight scouts that they wouldn’t be brought back, GM Jeff Luhnow called the cut a “reconfiguring,” telling MLB.com, “This is not any sort of reduction in staff size and scouting,” and characterizing the news as “normal” and something “that happens every year.” What the Astros are up to is obviously abnormal, but whether it’s ill-advised remains to be seen. The Astros have replaced their pro scouts with a small contingent of “scouting analysts” and roving video technicians who capture high-speed, high-definition video of players and transmit it to Minute Maid Park for automated processing that yields insights into mechanics, pitch grips, and potential areas of improvement. Without having to travel, the scouting analysts can quickly synthesize several sources of information to provide a more comprehensive picture of a player’s performance and potential than it’s typically possible to supply from the stands.
In speaking to several ex-Astros scouts about the team’s methods for my upcoming book about player development, I heard a mixture of frustration about the aforementioned misleading messaging, concern for fellow scouts and the industry as a whole, grudging respect for the Astros’ effectiveness, and critiques about the subtleties of technique or makeup that the Astros may miss by largely dispensing with an in-person presence. The extent to which the Astros have downsized their old-school scouting presence has incited some internal dissent, and no other club has quite copied the Astros’ model. “I think a lot of teams are kind of [taking] a cautious view of what they’re doing and want to wait and see,” one former Astros scout says. Leading teams like the Yankees, Dodgers, and Red Sox continue to invest heavily in scouts even as they double down on technology, but this winter touched off a feeding frenzy for Astros personnel who could spread their scouting approach.
Successful teams’ tactics tend to migrate to the rest of the league, especially if they cut costs. As another ex-Astros scout told me for the book, “It’s hard to argue because they’ve been so good. … If it goes well, then scouts in general should be worried because you aren’t going to need them anymore.” Still, he adds, “I think over a longer term that will come back to bite them a little bit.”
One front-office source says, “All the numbers see are the ends, and while this is still baseball they’re playing, it’s a far cry from the major leagues, and presumably you could have success with means that will work at the college level but [won’t] work at the major league level. The prototypical major leaguer has those correct or attractive means. I think that’s what a scout is attracted to.” But today’s technology is blurring the lines between statistical and scouting data, and tomorrow’s may erase any remaining distinction. “Maybe one day when you’ve got computer vision quantifying all that, you won’t need to scout,” the source says. “But until then, I think you do.”
The Astros still employ as many scouts as they do because the amateur and international markets aren’t as thoroughly blanketed by tracking devices as pro ball. But it’s only a matter of time until data and video coverage are complete almost everywhere prospects appear. TrackMan recently reached its first Division II school, and the system is now installed in two high schools. Data-rich amateur showcases and initiatives like the Prospect Development Pipeline League are making it easier for teams to evaluate players’ performance against high-level competition before the draft, and the Trainer Partnership Program is extending the data dragnet to international markets. Teams are also obligated to get amateur players to sign up for a new MLB-operated portal called Draft Prospect Link, where prospects upload biographical, vision, health, and medical documents for all 30 teams to access. As one scout says, “[There’s] no real such thing as hiding players anymore.”
As one might imagine, scouts are evolving to mirror their environment. New scouts tend to possess some degree of statistical literacy, and some teams have bolstered their scouting staffs with recruits from the same sabermetrically oriented websites whose analytical exports have populated the league’s R&D departments. But a desire for self-improvement and self-preservation has also inspired some veteran scouts from before the Reds database days to go back to baseball school, in some cases quite literally.
“There are a lot of old-time scouts that are looking for jobs because they just haven’t been able to adapt to what’s going on out there,” says Frank Marcos, who served as the senior director of the MLB Scouting Bureau for more than 26 years. The bureau—which was created in the mid-1970s to supplement teams’ internal scouting operations and also administered the Scout Development Program, more commonly called “Scout School”—was downsized in recent years and shuttered in 2018. That culling left a lot of longtime scouts out of work. Carl Moesche, who lost his job at the bureau in 2016, was determined not to be one of them.
Moesche has been scouting for more than 30 years, first with the Orioles and Yankees and then with the bureau. When he became a free agent for the first time in 21 years, he says, “I think the overriding feeling was, the game was changing in the scouting world. … There are people that are stubborn, old-school if you will, but I think in this business today, to be smart about it, you better understand that this is where the game is trending and I don’t see how it’s going to stop. Because when you look at analytics … you understand that these are verifiable numbers, that these are absolutes for what the player has done.”
In 2017, Moesche took an online course in baseball analytics from Sports Management Worldwide. Later that year, the Red Sox hired him to scout the Northwest, which he partly attributes to his course certification and to the mind-set it instilled. “It’s made my evaluation more open-minded and more willing to take a second look,” he says. On the reports he writes today, he continues, “You have to quantify what you’re seeing based on statistical data to back up what you saw. … You really have to see if the numbers stack up to what you believe is true.” If they don’t, it may not be the numbers that are wrong.
Maintaining one’s position as a scout in these turbulent, tech-driven times, Moesche says, is “not about age so much as it is about knowledge. … It’s really, ‘What can you do for us?’” Scouts ask that question every time they appraise a player, but they’re forced to confront it too. And the more the game changes around them, the more they need new answers.