Officially, my baseball writing career kicked off on April 1, 2008,* the day after I signed up for Bleacher Report. But the seeds of it go back way earlier than that–maybe to when I read Moneyball in the summer of 2004, or stumbled open Baseball Prospectus 2006 in a bookstore two years later, or when the ravenous reading and re-reading of said annual prompted me to try to create an even more exhaustive sort of BP-style book.**
*I’m not positive if this was the first one or if I wrote another one the day before, but close enough. I cannot believe this still exists on the interwebs. And yiiiikes, cringey title, past me.
**I think this got as far as like a 15-page Word document and then I lost interest, of course. I mean, I was a 17-year-old who knew the names of probably like 2500 professional baseball players and had some basic working knowledge of sabermetrics, but boy did that give me immediate delusions of grandeur.
But the first piece of actual baseball analysis I can recall writing happened about a year before I hit “publish” on that first blog post. It was an essay in my AP English class, dated April 10, 2007, in response to the prompt “Make an argument for some sort of change. It must be at least two pages.” It was April of my senior year of high school and I was at least partially in the throes of senioritis, but I wrote a five-page* document entitled “The Myth of Quadruple-A and the Minor League Sluggers.”
*As a personal aside, English was my least favorite subject growing up, mostly because believe it or not, anytime I was supposed to write anything, it was pulling teeth to get a paragraph out of me, let alone an essay. I just had no clue where to start, or how to articulate anything in more than a word or two. I had a great vocabulary and knowledge of grammar, but I just couldn’t ever think of anything I wanted to say, really. So this paper was a real turning point for me, because it was actually *easy to write.* It came naturally to me in a way no piece of writing in all my previous years of schooling ever had. The start of my baseball writing “career” less than a year later, and my decision to major in Writing, Rhetoric, and Technical Communication (that’s one thing, not three) in undergrad (a decision I made at the beginning of my college sophomore year a year and a half later, and would never in a million years have considered remotely possible before writing the paper in question), owe a lot to that moment.
In the essay, I articulated my most fervent baseball belief at the time: that there was a class of players, perhaps 25 or 30 at any one time, who had been dominant at the Triple-A level for a year or more and would be above-average players in the big leagues, yet who were systematically not being given chances at the major league level because of trivial scouting concerns. This was 2007, so we were only a couple of years post-Moneyball, many front offices still had little to no focus on analytics, and the “scouts vs. stats” debates raged everywhere. My essay was very a much a product of its time: I thought the numbers held the answers, and that the “eye test” was inherently flawed and added little.
But if I reflect on the actual mentality I had at the time–one which, unfortunately, persisted into the first few years of my baseball writing time, which thankfully (hopefully?) have mostly been forgotten and lost in the sands of server migrations–it wasn’t so much that I was inherently dismissive of the idea of scouting concepts adding value, it was that I couldn’t logically work out how they’d fit into the picture. As luck would have it, my parents were clearing old stuff out of their attic this winter, including a bunch of my school stuff that they had saved, and had me go through it to see if I wanted anything; I didn’t want much, but I did want the essay, which I still had fond and vivid memories of writing and was surprised and delighted to see again, even as I'd long moved past the ideology it espoused. I’d like to quote a particular section of it for you:
“The problem–and the reason [Jeff] Bajenaru will be pitching for the Tucson Sidewinders instead of the Arizona Diamondbacks this year, and Scott McClain will be hitting for the Fresno Grizzlies instead of the San Francisco Giants–is these nebulous concepts of ‘major league players’ and ‘minor league players.’ There are not two distinct sets of pitchers or hitters. Players go up and down from the majors to the minors and back, all the time. The difference between the best pitcher on a PCL team and the worst pitcher on an MLB team is slim to none. The difference between the average player in the higher levels of the minor leagues and an average player in the major leagues is a slight bit of talent and polish.
The scouts claim this slight bit of talent and polish makes minor league sluggers ineffective. But this makes no sense. To illustrate, let’s set up a scenario. The average Triple-A pitcher has a fastball of maybe 90 mph, a curveball with 10 inches of curve,* and decent control of his pitches. The average MLB pitcher has a fastball of maybe 92, a curve that curves a foot, and slightly better control. Are those slight increases–2 mph, 2 inches of curve, and slightly better control–going to transform Scott McClain, Jon Knott, Jack Cust, and Marshall McDougall from MVP-caliber players to benchwarming liabilities?”**
*Pitch F/X, let alone Statcast, didn’t exist yet. Gimme a break.
**As a postscript to my personal note above, I want to quote the first part of my teacher’s comments on this paper: “This text includes a good deal of comfortable writing & an unexpectedly powerful closing paragraph…[in] part it’s just nice to [see you write] about something you clearly enjoy. Your treatment here is unusually full & comprehensive.” He noticed. To have my own feelings of “wait, writing isn’t an impossible task?” validated was meaningful, and it meant a lot to see that validation all these years later, given all I’ve gone on to write about minor league baseball and achieved academically. Teachers can make a huge difference in our lives by recognizing what we as kids only *think* we’re recognizing about ourselves, even in what can seem like throwaway moments like mid-April of senior year.
As you can see, the point I was kind of stuck on was an elaborate form of “MLB can only be so different from Triple-A.” The player universes of the two levels overlapped significantly within a given year, and the skillsets of the players at the two levels seemed, as far as my relatively-poorly-informed brain could understand, to differ by a present but fairly insubstantial amount. Why would Jon Knott’s long swing wreak havoc on Triple-A pitchers but then be reduced to flailing against slightly better major league pitchers? I mean, maybe it would, but it seemed presumptuous to think that without having a big sample of big league data.*
*Honestly, even all these years later, I do retain *some* sympathy for this kind of viewpoint, particularly when applied to players in those days where front offices were much less analytically-oriented.
As I proceeded to trumpet the major league merits of players like Knott, Joe Koshansky, Brooks Conrad, and Jeff Natale on Bleacher Report in the late 2000s, I felt fairly alone in continually advocating this idea that the “Quad-A” designation was inherently misguided and that there was an MLB team’s worth of 2+ WAR players just sitting around as forgotten upper-minors depth. But to this day, I see seeds of this kind of mentality in online baseball discourse, when fans ask about players they’ve noticed are putting up good statlines, yet not cracking prospect lists. Why won’t this guy be good in the big leagues, when he’s killing it in Triple-A? Why won’t this player adjust well to the upper minors, when he’s been dominating A-ball? Why won’t this short-season stud burst onto the scene with a full-season assignment? When people ask these questions, they often seem to stem from the same kind of standpoint I had in the essay: Who are you (scouts, prospect writers, etc.) to say it can’t happen for this guy? You say it’s the swing, or the stuff, or the approach, but it sure seems to be working so far. How different can [the next level] possibly be? Sometimes guys translate seamlessly across these promotions, or even improve! Surely then, the boundaries between levels are permeable.
Well, when I actually realized how dead-end of a venture trying to project players based merely on their statlines was,* I more or less decided to start from the ground up again, this time working off actually going to minor league games and evaluating players that way. I learned many valuable lessons from those experiences, but one of the most important ones was that this idea of levels as permeable is half right, but half very wrong. And though I think most prospect evaluators working off extensive experience with game footage (whether in person, on video, or both) probably have a similar understanding of the differences between levels as I do–knowledge that doubtlessly is used and implied in their analyses of players–I’ve never really seen anyone actually lay that knowledge out clearly. Since it’s an important underpinning of how prospect analysis works, I wanted to take the time to walk through my observations of how levels actually differ, how this might affect how players translate across them, and a couple of brief examples that play around with some of the ideas more quantitatively and concretely.
*This slowly gnawed at me starting in 2010 or so, and by the end of 2011, I was convinced that stat-based approaches clearly were missing a lot of valuable information (Now, over a decade later, some people are taking it on more skillfully than I did then, no doubt aided in part by the richer data that has come into public view in the years since, so this isn’t a dig at those folks). Happily, this coincided with when I had graduated from college, had a car, and had moved to an area where there was a lot of minor league baseball around. So there wasn’t much I could’ve done before that anyway–milb.tv was way less useful for analytical purposes than it is now.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
First, let’s talk about the part that’s half right. Yeah, levels have a reasonably high degree of overlap between them. Just last year in the A’s system, Max Schuemann hit .224/.347/.363 in High-A Lansing but .320/.398/.416 in Double-A Midland, Mickey McDonald hit .273/.378/.337 in Midland and .333/.423/.438 in Triple-A Las Vegas, and Garrett Acton added 15% to his strikeout rate while shaving 3% off his walk rate in being promoted from Low-A Stockton to Lansing, for instance. The fact that it’s not all that rare to see performances jump across promotional lines, let alone hold relatively steady, indeed reflects the truth in the idea that “High-A pitchers” and “Double-A pitchers,” for instance, are not entirely distinct from one another. There isn’t a neat dividing line that sorts players into rigid tiers.
I was really fortunate in my live-look days to live in areas that allowed me to see prospects at all levels: I’ve seen quite a few games at every level of the minor leagues (including the now-defunct Rookie-Advanced and Short-Season-A) other than the rookie complexes, and so I got a firsthand look at what the level of baseball is like every step of the way up to the big leagues. As I’ve talked about before, one of the first aspects of minor league baseball that defied my expectations when I started frequenting games was just how ubiquitous MLB(ish)-level skills are. You don’t go to even a Low-A field and immediately rule out the majority of the players on it from MLB relevance; in fact, the differences between them and big leaguers are pretty subtle, all things considered. I mean, right now, the Stockton Ports have a relief pitcher with an 11.50 ERA at age 24, Alexis Cedano. With a statline like that at such an advanced age for the level, you’d think he’d look miles away from even the worst big league pitcher, but Cedano brings a 92-94 mph fastball with decent plane that has touched 96 mph, a hard overhand curve with decent shape, and a functional sinking changeup with reasonable arm speed. His command isn’t very good, but he doesn’t have the yips or anything. The difference between Cedano and, say, Jake Lemoine just isn’t all that apparent to the untrained eye, and Lemoine has been four levels and several runs of ERA above Cedano.
So when I watch levels directly adjacent to one another, I absolutely see a ton of overlap in the current abilities of the players in them. Even with the A’s recent promotions of a few of their most dominant players at various levels–Jonah Bride, David Leal, Brett Harris, and Blake Beers–there are still a good number of players at each level of their minor league system who I could easily see slotting into the next level right now. I won’t trouble you with an exhaustive list, but it probably runs into the twenties for the organization as a whole. That’s a fairly usual amount, even perhaps a bit low–I’ve seen minor league teams where the majority of the players already look next-level-ready (Midland, at least earlier in the year when everyone was healthy, wasn’t far off from this).
The problem is that in focusing on how levels aren’t entirely different from one another, we* can easily minimize what differences do exist. Just because the boundaries of levels are permeable doesn’t mean that levels don’t present very different challenges for players. More importantly, the differences can often tell us a lot about what sorts of players might handle promotions well, and which might have struggles.
*An oft-discussed problem in persuasive/editorial sorts of writing is the author arguing against a position that nobody actually holds, thus undermining their own points because they demonstrate a complete inability to understand what the other side even is. I’m not alleging here that any/many folks out there necessarily are thinking of levels in this way. I just know that I did once upon a time, and I’m sure others who haven’t taken the time to consume a ton of minor league baseball might have some parts of/similarities to my old views.
Once I had settled into live-game analysis in the 2013 season and stopped oohing and aahing at how almost every player didn’t look like a complete scrub, I did start to notice that despite all of the similarities in raw talent that were evident at the various levels, there were some differences. It turned out that my old insistence of there “not being two distinct sets” of players at different levels was kind of missing the point. The differences between levels, I realized, mostly centered on who wasn’t there.
By this, I mean that each level had a class of players that seemed to not make it up to the next one. In short-season ball, most pitching staffs had three or four guys who sat in the mid- or upper-80s and relied on deception or patterning to get outs; most Low-A staffs had either zero or one. But those Low-A staffs had a few guys who would sit in the 89-91 range and throw just an ordinary slider and change; High-A staffs would have only zero or one. Those High-A teams’ most fungible arms would be more in the 90-93 mph range with ordinary offspeeds, or 89-91 with a plus secondary pitch. And so on. The same sort of dynamic existed with hitters: the least interesting three or four players on each team (corner guys without bigtime bats, or up-the-middle guys who really couldn’t hit) would fall into a sort of group that you wouldn’t see much of at the next level up. In my head, I termed this the One-Third Rule: one third of players at each level don’t move up to the next one. It is thus in that “worst third” where the differences between levels are readily apparent.
The One-Third Rule is itself overly simplistic and reductive in a number of ways, some of which I’ll unpack in a bit. But what I find so useful about it is that it completely reframes the question of how we should go about thinking through how players might translate to the next level. It’s no longer about something like whether the player can adjust from facing “High-A opponents,” whatever that abstract idea actually represents, to “Double-A opponents,” whatever that means. What it is about is whether the player’s strong performance at his current level is especially driven by disproportionately strong performance against players who are not talented enough to move up to the next level. In my simplistic schema, the worst third.
To think about this in a really simple way, consider a hypothetical pitcher on the 2021 Great Lakes Loons, the Dodgers’ High-A affiliate, facing off against the 2021 Lansing Lugnuts, Oakland’s High-A affiliate. The most likely lineup that pitcher might’ve faced would have looked something like this:
2B Cobie Vance
SS Elvis Peralta
C Drew Millas
3B Jordan Diaz
1B Will Simoneit
LF Shane Selman
CF Austin Beck
RF Lester Madden
DH Pat McColl
Now imagine that hypothetical pitcher was promoted to Double-A Tulsa over the offseason, and he faced off against Double-A Midland. He wouldn’t get to face Beck and McColl, who are still in Lansing this season, and Madden, who didn’t garner an Opening Day assignment. The middle third of the lineup here is indeed in Midland, however. Vance, Millas, and Peralta are in Double-A as well, but in other organizations (minor league Rule 5, deadline trade, and release/free agent signing, respectively). So if a lot of our hypothetical pitcher’s success in Great Lakes was due to his ability to easily dispatch guys like the 2021 versions for Beck, Madden, and McColl, he would be in for a bumpy ride with Tulsa against lineups like Midland’s. It’s as if he simply has to face a lineup of the 1-6 hitters over and over again. A similar thing is true with pitchers (at the next level, it’s like facing the best four starters of a six-man rotation over and over again, along with the best four or five relievers) though the frequency of season-long (or longer) injuries does make things a little less linear.
At least, that’s how I’ve usually thought about things the past few years. A quick quantitative analysis puts the year-to-year promotion rate lower than two-thirds:
Level | % Position Players at Previous Level in 2021 | % Pitchers at Previous Level in 2021 |
A+ | 45.02% | 37.75% |
AA | 46.75% | 42.99% |
AAA | 27.92% | 32.12% |
*This is data on the percentage of players at each level who have had at least 80 PA/20 IP this year (as of May 25, when I started drafting this article) who had at least 250 PA/40 IP at the next level below in 2021. I just used this year of data in the interest of capturing where things were now, post-MiLB contraction.
That analysis does show that, perhaps for injury and other reasons, the rate of advancement is lower for pitchers than hitters, and it really decreases in the AA/AAA boundary, since there are way more repeaters at AAA than any other level. But when you factor in guys returning from injury and guys who fell just a bit below the playing time thresholds I’ve set, I don’t think the one-third idea is really that far off (until AAA, which despite these low numbers is not typically thought of as the hardest level to adjust to)*. You could maybe argue that it should be the Two-Fifths Rule for batters and the One-Half Rule for pitchers, or something weird like that. But whichever of these thresholds best captures the level of talent that quickly ascends to the next level vs. that which stagnates at the previous one aside, the point is that the conceptual difference between adjacent levels is better described as the vanishing** of the lower tier of talent rather than a wholesale improvement in the baseline talent level.
*Needless to say (I think?), the AAA/MLB jump doesn’t really conform to this idea in a particularly standard way either. I think it ends up being reasonably on point in a conceptual sense until then though.
**Two things on this. First, of course the lower talent level doesn’t *entirely* vanish at the next level: it just becomes very rare. Second, by “vanish,” I don’t mean that all of these lower-third players get released: Austin Beck may yet play games in Double-A and beyond, for instance. But in order for those lower-third players to move up, they have to make substantive improvements, whereas everyone else has a chance to ascend even if they basically don’t improve at all.
As I found in my study of players who repeat levels, significant (statistical) improvements from players happen more rarely than we might typically think (collectively, the hitters in that analysis improved from .251/.320/.374 to only .262/.332/.400 in the following year). They do happen, of course*–there are several players in the A’s system who have showed up in much improved form in 2022, for instance (that article’s subject Austin Beck, as well as Logan Davidson, Stevie Emanuels, Trayson Kubo, Blake Beers, Kyle McCann, etc.)–but the plurality of players more or less holds steady on a year-to-year basis. So if you’re a good player at a given level, you don’t necessarily have to make a big leap in skill to survive at the next one, unless your current success is predicated on the domination of the particular caliber of opponents who will not advance.
*And this does mean that, of course, the best players at a given level end up being better than the best players at the level below it–it’s not perpetually flat in the way that the One-Third Rule kind of implies. But again, the rule serves (for me) as more of a sanity check/thought sharpener than any sort of attempt to describe the different levels with nuance.
This is really more of a theoretical heuristic than something one needs to actually apply statistical rigor to,* but for the sake of actually demonstrating an example, let’s consider two of the Stockton Ports’ best pitchers last season, both pitching to ERAs around 3 and K/BB ratios well north of 3. But let’s split their performance out into two groups: how they fared against the worst third (by wOBA) of their opponents (who hit around .225/.310/.340 with walk and strikeout rates just over 9% and 28%, respectively) and how they did against everyone else (around .285/.375/.475 with 11% BB and 24% K (woo, inflated Cal League numbers!).
*Not that it wouldn’t be extremely interesting to have somebody come up with an opponent-weighted statline, whether in this sort of thirds idea or using some sort of weighted gradient. But that person would have to have actual database/programming skills, which I don’t. I had to hand-chart all the plate appearances in this example, which is really time-consuming (otherwise, I’d have done a lot more with data here).
Player/Opposition | BAA | OBPA | SLGA | BB% | K% |
Pitcher A vs. Upper ⅔ | .217 | .300 | .333 | 11.43% | 30% |
Pitcher A vs. Lower Third | .156 | .200 | .281 | 5.71% | 45.71% |
Pitcher B vs. Upper ⅔ | .248 | .282 | .396 | 5.45% | 24.55% |
Pitcher B vs. Lower Third | .235 | .304 | .392 | 7.14% | 28.57% |
Both pitchers were fairly effective against everybody, but you can see that Pitcher A dominated that worst third, limiting them to a batting average well under the Mendoza Line and striking out almost half of them, but his command slipped significantly against everyone else, cutting his K/BB ratio from near 9/1 to below 3/1. Pitcher B, on the other hand, was about equally effective against everyone.
Both pitchers were promoted to High-A Lansing at midseason. Pitcher A, Jose Mora, had been Stockton’s closer, but he struggled mightily in Lansing, posting an ERA north of 11. There was certainly some bad luck in that ERA, but the command problems that his best-⅔ performance hinted at became fully-fledged issues at the higher level, as his walk rate jumped to 15% and his strikeout rate declined to 24%. Pitcher B, Jack Cushing, had been Stockton’s ace, and he went to Lansing and…immediately became their ace, with a 5.4% walk rate and 26.1% strikeout rate.
In those cases, the difference between the pitchers’ performances against the lower third and the upper two-thirds seems to have been prescient. I’m not saying it always is, or that it’s even particularly predictive by itself*–in the other example I tested out, I found Max Schuemann hit .323/.450/.516 against the worst third of his High-A opponents last year (by FIP) and .187/.300/.306 against everybody else. But once he was promoted to Midland at midseason, Schuemann’s Double-A batting average was in line with the former number far more than the latter (the walks and power numbers did come down as the upper two-thirds numbers would lead you to believe, though), and it’s stayed there this season so far as well.
*For one thing, the sample’s often gonna be pretty small, as it is with Mora, Cushing, and Schuemann.
But let’s think about what those lower thirds actually are. For hitters, that lower third is usually either a) guys who have poor approaches and are prone to “getting themselves out” against the stuff typically wielded by pitchers at a given level, or b) good approach or not, they don’t have the strength to really punish mistakes, so pitchers can pound the zone without fear of retribution beyond a flare over the infield. For pitchers, it’s a) guys with fringy stuff or b) guys with command issues. The players who might disproportionately take advantage of those opponents might often be the players who have good approaches and savvy, but lack the physical tools to hang in well against the better opponents.
These sort of players are especially likely to run into problems as they advance because there are fewer obvious adjustments/improvements they can make to improve. Someone like Jorge Juan, to pick an A’s example, is prone to losing his release point and missing badly for innings at a time, consistently falling behind hitters. He’s not really facing the batter at that point: he’s battling himself. The physically-overmatched 5’7” second baseman with a good eye who hits ninth in the order will be more than happy to oblige and let Juan sail four 98 mph fastballs up and armside and walk down to first base. But when Juan isn’t wild, it’s like the old Steve Stone quote: “If my curveball is breaking and I’m throwing it where I want, the batter is irrelevant.” In both cases, Juan’s success or failure is due much more to his own ability to execute than the quality of the opponent he’s facing. Thus, if he can get more consistent with his command in general, his stuff will play against anyone.
Juan has a chance to do that. But what are guys like Ty Damron, Michael Danielak, and Matt Milburn supposed to do? All three are 27 or older and stuck in Midland with middling performance, despite all three of them putting up impressive numbers in the levels below it. They’ve all got average sliders that they lean on heavily and at least one other functional offspeed pitch (Damron has three more, in fact), and they work the ball around the strike zone and hit the corners. But none of them throw very hard or have a consistent bat-missing offering. They were able to carve up the low minors because if you have command, savvy, a solid slider, and a couple of other reasonable pitches, you’ll shut down the weaker hitters at those levels, who will get themselves out, and you’ll do okay against everyone else. But when those opportunities dry up, it becomes much harder for pitchers like that to miss bats, and they also can’t afford to miss their spots to even the #9 hitter in the lineup. There’s not an easy way for this sort of pitcher to adjust to opponents of higher quality, short of simply finding a way to throw harder or get better stuff in some other way (maybe, if they didn’t have one before, they could give a more deceptive delivery a try, but that’s about it), because they’re already quite polished and have optimized the stuff they have–that’s what got them to their level in the first place.
So not only does a One-Third Rule kind of idea capture a lot of (my) observed difference between levels, it also tends to favor (conceptually; I don’t know about statistically) the players who have more traditional “prospecty” attributes (tools, etc.) than players who have performed well but lack those attributes. This is why I (and I imagine that in some related sense, many other prospect writers) express skepticism about organizational-type players with good statlines translating to the next level while often remaining relatively unconcerned about toolsy prospects’ middling performance.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
So that’s my story of how I learned to stop focusing on “permeability” and love the one-third split idea. It might be an interesting idea to play around with in a more quantitative sense in the future, and I’m not sure how well splits by opponent quality would predict prospects’ ability to seamlessly translate to higher levels of competition and/or raise their prospect status or make it to MLB. At the very least, though, it gave me an education as to what we’re actually looking at when we consider performance at various levels of minor league competition and helped get me toward a more productive framework than “his performance is good, so I should expect his performance to be good at the next level.” If you’ve struggled to conceptualize why analysts make these sorts of claims about players in spite of their current performance, I hope this has been illuminating toward some possibilities as to why. Players will still defy these heuristics all the time too–nobody saw Schuemann’s breakout coming, and the one-third split didn’t catch it either–but that’s the beauty of minor league baseball analysis: there are always counterexamples to learn from, no matter your philosophy. But players’ ability to maintain or improve their performance as they advance to the next level of competition hinges on how they can adapt to their easiest opportunities for success being taken away. Promotions may not create a whole new ballgame, but they do introduce critical differences that separate players in a way that indeed dovetails well with a lot of conventional scouting wisdom.
Comments