(Zontar, buddy, this one’s for you.)
(Also, a warning: this is super-long, but given how many different areas there are to explore on this topic that aren’t even handled here, it made sense to group all of the stuff below into one post, at least.)
The minor leagues are weird. Actually, scratch that, baseball is weird. Baseball players are seemingly subject to inexorable laws like regression to the mean, but are also all special snowflakes with unique talents and developmental paths. If baseball were a poorly-programmed video game, there would probably be a system where every player that enters an organization’s minor league system would spend a year at each level, before finally being promoted to the major league team after a full season in AAA. However, we know things don’t work that way — even if no empirical evidence on this were available, the spate of rancorous debates about whether different minor league prospects are promoted too soon, or not soon enough, across the various minor league levels and eventually to the Show should prove that point handily.
One point in particular that piqued my interest was the idea of AAA as a repository for the types of players not quite good enough to play in the majors, but not necessarily vaunted prospects. If you’ve heard a similar line of thinking — great! I probably don’t need to spend more time convincing you. If you haven’t heard this line of thinking before, then you’ll probably be confused as to why I’m bothering to spend my time looking into whether the claim has veracity. Read on anyway, or don’t. However, consider that in recent Braves history, Dansby Swanson skipped AAA altogether, making his Gwinnett debut well after his Atlanta debut after a demotion, while Jason Heyward and Evan Gattis got only a smattering of PAs at that level prior to ascending to the majors.
Unfortunately, this isn’t really a topic that has been explored much, and as such, I had to pull together my own datasets to look at it. Here’s what I focused on:
- All rookie seasons, from 2010 through 2017, the latter of which is not yet fully complete, where the player got 200 or more plate appearances (PAs).
- Within those rookie seasons, I focused specifically on wRC+ (as an aggregate measure of hitting quality), and fWAR, as well as fWAR per 600 plate appearances (as aggregate measures of player production).
- For each such rookie season, I looked at how many plate appearances the player had gotten in AAA, through the conclusion of the rookie season.
That last bullet sadly requires a bit more explanation. In short, there is no easy or facile way to determine AAA PAs before promotion, as opposed to after. Given that many players ride the shuttle to and from AAA more than a few times, the data here are somewhat marred (or tainted, if you prefer) by the fact that my solution was simply to look at total AAA PAs accumulated through the point at which the player used up his rookie eligibility. So, as a result, in the dataset, Dansby Swanson has 45 AAA PAs, even though we know he was called up before those PAs took place; meanwhile, Alex Avila has zero AAA PAs because his stints in AAA came well after his rookie season (2010), i.e., 2012, 2013, 2015, and 2016.
With this data in tow, I sought to answer the following questions:
- What proportion of rookies actually spend time in AAA?
- How do rookie performances differ based on the time they spent, or didn’t spend, in AAA?
- Are there any general correlations between PAs in AAA, and rookie season outcomes? What about thinking about player careers, moreso than rookie seasons?
- Are there any general trends in how teams have or have not assigned players to AAA over time, over the past few years?
- How does player age factor into all of the above?
Note that there are two big things not handled here. First, this is only position players, not pitchers. Pitchers will be handled in a subsequent part of this exploration, as it doesn’t make sense to mash them together. Second, this is not an exploration, directly, of the promotion paths of top prospects. Yes, I am planning to do that, but it’s a separate exercise that requires its own targeted examination. At present, I’m looking at the general effects of AAA on position players, not on said effects for an artificially-designated set of players favored by prospect evaluators.
Part 1: What proportion of rookies actually spend time in AAA?
This one is easy to answer. There have been 338 rookie seasons with 200 or more PAs between 2010 and the present. Of those, only 27 belong to players that accumulated zero PAs in AAA through said season. So, the basic answer: 92 percent of all rookies with substantial playing time since 2010 spent at least some time in AAA. So, the next time someone tells you that no eventual major leaguers spend time in AAA anymore, laugh at them.
However, there’s a difference between taking a few hacks at AAA, and staying there for the long haul. If we consider the same 200 PA cutoff as a minimum for “was in AAA for any real length of time,” our count of players goes up to 98. (Just making the cut: Tommy La Stella, with 199 AAA PAs through his 2014 rookie season.) In other words, over 70 percent of rookies with substantial playing time since 2010 spent got at least 200 PAs in AAA. Continue laughing when you hear or read the aforementioned statement.
Rather than list every single cutoff and associated percentage, consider the following chart.
The way to read this chart is as follows: “[Value on x-axis] percent of players had [value on y-axis] PAs in AAA or fewer through their rookie seasons.” In other words, 50 percent of players had 416 PAs in AAA or fewer through their rookie seasons, and therefore 50 percent had more than 416 PAs in AAA through their rookie seasons. 70 percent of players had 600 or fewer PAs in AAA through their rookie seasons. That poor, sad, 2,236 figure? That represents Ivan De Jesus, who has had a very strange professional career. Just look.
So, there you have it. Don’t let anyone tell you that AAA is a purgatory in which non-major leaguers spend their days. Sure, AAA has got a lot of that. But making a stop at AAA is standard for most eventual major leaguers. There is no great “skip AAA” directive. (Again, whether such a directive exists for “top prospects” will be handled at a later date.)
Part 2: Does spending time in AAA affect rookie season performance?
Let’s get something important out of the way: any relationship between data on rookie performance and AAA plate appearances will be incredibly complex. The reason is because there are two competing explanations in play:
- Playing in AAA affects a player’s ability to acclimatize to the majors (in either a positive or a negative way), and therefore players who play in AAA will have different outcomes during their rookie seasons than those who do not; and
- Teams systematically vary who is and is not allowed to spend time in AAA by player type, prospect quality, or other factors, and therefore players who do (or do not) play in AAA differ from those who do not (or do).
Why is this problematic? Because the combination of these effects can drown out the ability to view a statistical relationship. If AAA systematically helps a guy during his rookie season, but teams only skip the better players over AAA, then it is possible that despite both of these effects existing, no evidence will appear in the data. (I won’t belabor this point, but can explain more in the comments if it is unclear.) It’s important to keep this in mind as you look at the other information in this section.
The table below separates players into a few buckets, and sub-buckets (mini-buckets?), by quantity of AAA PAs through their rookie season. Specifically, we have:
- Little to no AAA exposure — further separated into players with 0 AAA PAs, and players with 1-199 AAA PAs;
- Under, or just about a season of AAA exposure — further separated into players with 200-399 and 400-599 AAA PAs; and
- A season or more of AAA exposure — players with 600+ AAA PAs.
The results speak for themselves, I think, but I’ll touch on the things I find particularly interesting (which is, like, everything!).
You can see that by these buckets, pretty much the same number of players have 1-199, 200-399, and 400-599 AAA PAs. Clearly, teams are not really choosing any one of these natural breakpoints as a heuristic for AAA exposure or MLB promotion decisions. In addition, about 30 percent of MLB rookies are former (and future?) AAA journeymen/veterans.
Playing time is distributed in what is probably an intuitive sense: the less time a player spends at AAA, the more PAs they tend to get in their rookie MLB seasons. This is kind of a no-brainer: earlier midseason call-ups will lead to more MLB playing time. But, you can also see a second trend that reinforces this: the more a guy marinates in AAA, the less he’s relied upon to be a starter in the majors. This is consistent with the general idea that guys with a ton of AAA PAs tend to be role players, rather than presumptive starters if and when they are promoted.
Hitting is where it gets interesting. There is, for all intents and purposes, almost no real difference between any array of players with at least one PA in AAA. There’s a slight indication that those players not spending a bunch of seasons in AAA are marginally better in terms of offensive outcomes, but that’s about it (compare 96-97 to 94). The real kicker, though, is that the small handful of players with no AAA exposure tends to be appreciably better with the bat than the rest of the rookies examined here. Why does this occur? Some of it is baked in to the success of foreign imports who skipped AAA, namely Yasiel Puig, Jose Abreu, and Jung Ho Kang. However, even without the professionals from another league, the pattern still holds (though only at about half its strength): Giancarlo Stanton, Miguel Sano, Christian Yelich, and Matt Duffy are players who skipped AAA entirely and hit very well as rookies. Of course, this isn’t always the case: for every Stanton or Sano, there’s a Derek Dietrich, Rougned Odor, or rookie season Jose Altuve (80 wRC+!) who balances this pattern out. But, there’s more balancing altogether within the other buckets. I think this makes sense: some of the guys pushed aggressively are the guys generally expected to handle the majors more suitably.
What’s interesting, though, is that even the next group no longer really holds this pattern. In aggregate, by the time you get to a cutoff of around 150 PAs in AAA, the group with fewer than that has about the same batting line as the group with more than that. In other words, this effect is pretty unique to the guys with very little or no AAA exposure. Another way to look at it is with 50-PA tranches: the 100-150 PAs in AAA tranche already looks a lot like the other tranches with more AAA exposure. The only particularly noteworthy groups appear to be those including hitters with fewer than 100 AAA PAs, and even then, the pattern is more suggestive than definite. For almost every rookie Michael Taylor (69 wRC+) there’s a rookie Alex Bregman (112 wRC+), and vice versa.
The takeaways for pro-rated fWAR are similar in some ways, but not others. The less time spent in AAA, the more valuable the player, on a rate basis. Again, this is largely driven by higher-caliber players being promoted more quickly, but interesting, this pattern holds better than just hitting. Notably, the swing isn’t quite large — the difference in medians between guys with 200 or fewer PAs in AAA and guys with 600 or more is less than a win, and the difference in averages is roughly half a win, or the rule-of-thumb margin of error as far as WAR goes.
So, as far as conclusions go: players who spend less time in AAA tend to be better. This isn’t necessarily because AAA sucks out a player’s life force, but probably just because better players get promoted more aggressively and/or skip AAA altogether. Any effect of “getting more practice” at AAA seems to be negated by the better players getting more aggressive promotions. You could perhaps point to the slight differences between players with 200-399 and 400-599 AAA PAs, but those differences are so small as to be basically negligible. In short, we shouldn’t assume, as a blanket rule, that players need more seasoning in AAA. That may be true for a specific player with specific issues, but as a rule of thumb, a lack of AAA PAs does not appear to harm a player’s rookie production, after taking into account that better players may potentially spend less time in AAA to begin with.
Part 3: Are there general correlations between AAA PAs and rookie season outcomes?
Not much is different with regards to the potential answers to this question and the one immediately above.
In case you’re like, “What the hell am I looking at?” for this quarter of charts, well, that’s kind of the point. The left charts are wRC+ plotted against AAA PAs; the right charts are fWAR/600 plotted against the same. The top is all rookie seasons; the bottom is only rookie seasons with fewer than 1,000 AAA PAs, to exclude the most egregious of the AAA journeymen, who make up about 10 percent of total rookie seasons over this period.
Basically, there’s no correlation. This actually looks like it was generated by a random number generator — that’s how little evidence of any hard-and-fast relationship between AAA exposure and rookie season production there is, at least when we are thinking about lumping all the data together like this.
Again, this might be due to competing effects: better players spend less time in AAA, but players may also improve if they spend more time in AAA, and the two effects cancel each other out to some extent. Or, it might be because spending time in AAA is not generally really relevant for MLB performance. In general, though — don’t cry over missing AAA opportunities for a given position player, at least not without specific cause to do so.
This pattern is also true if you look at a player’s entire career since that rookie season. In fact, the numbers are eerily similar to the table in Part 2, above. Specifically, as you can see below, when using career numbers, you only see a few changes, which are basically that players with 400 or more PAs in AAA experience a greater slide in hitting ability than their peers. Essentially, this is not particularly novel as a finding: players not rushed through AAA are less-solid bets, long-term, for major league success than those who are.
The correlations when comparing career wRC+ and fWAR/600 to plate appearances spent in AAA are similar to those focusing only on rookie wRC+ and fWAR/600, that is, essentially nonexistent.
Part 4: Are there trends in teams assigning position players to AAA over the last few years?
When doing this analysis, a constant tickle in the back of my mind was that doing this was all well and good, but the statement that “up-and-coming players don’t spend time in AAA anymore” wouldn’t be fully repudiated if my analysis were heavily weighted or biased by stuff that happened in 2010, as opposed to more recently. To that end, it was important to look at whether trends in AAA PAs have changed over time.
Hopefully, the charts below provide a decent look at that.
By raw count, you can see that the number of rookies overall hasn’t varied too much each year (2017 is low because it’s still partial, and the 200 PA cutoff has not been relaxed), and there aren’t any notable trends in the number of rookies within a specific bucket bucket of AAA exposure. It’s not even that it’s cyclical — it’s just noisy. Or lumpy. Whichever you prefer. The percentage basis chart helps clarify this.
So, there you have it. In case you’re still with me through nearly 3,000 words, what have we learned?
- Most rookies (~90%) spend at least some time in AAA. In any given year since 2010, at least 60% of rookies have spent at least 200 PAs in AAA. There is no support for the overarching suggestion that MLB players now routinely skip AAA.
- Spending time in AAA does not have a clear-cut effect on rookie season production. The best young players tend to be in situations where their organizations feel comfortable with them skipping AAA, so the small handful of guys that never appear in AAA tend to have the best outcomes. But, beyond that, there is no clear pattern. You could say that the AAA veterans tend to be role/bench players in the majors, and that the longer a guy spends in AAA, the less likely he is to be a major league regular. But, as far as major league acclimatization and the like, AAA exposure doesn’t seem to have much of, or any, effect, when considered through the lens of the complete body of rookies. (Of course, AAA may help any given individual player, and there’s no natural experiment that can assess how well a given player would play in the majors with and without a set of PAs in AAA. However, for the complete population of rookies, there doesn’t seem to be any AAA effect.) This pattern also extends to overall major league production.
- There is no overarching trend of AAA becoming less popular as a stop for eventual major leaguers over time. The proportion of rookies skipping AAA altogether, or getting only a few PAs therein jumps around, but teams aren’t limiting exposure of their promotees to AAA in any systematic way that’s growing over time.
Part 5: Does player age have anything to do with the findings above?
This is a great question, and one that didn’t occur to me initially, even though it seems “duh”-level important in retrospect. (Thanks, siegeface.) After all, your Manny Pinas (a 30-year-old rookie having a good season for the Brewers this year) are very different from your Dansby Swansons and your (eventual) Ronald Acuñas, and not just because your Swansons and Acuñas have little to no AAA exposure, while your Pinas toiled for 1,000 or more PAs in obscurity.
The average age of a rookie position player with 200+ PAs from 2010-2017 is 24. (This is the median, as well.) If you go across the AAA PA buckets, you get slight upticks in age: 23.2 (0 PAs in AAA) —> 23.7 —> 23.6 —> 24.2 —> 25.6 (600+ PAs in AAA). So, younger rookies are more likely to skip AAA than older rookies. If this feels tautological to you, (“they’re older rookie because they spent time in AAA!”), I don’t disagree in the least. Also, “congrats” to Ed Lucas for being the oldest non-foreign import rookie in this sample: he finally made the majors after 1,425 PAs in AAA, at the ripe, blooming age of 31.
Here’s a set of stuff I did to try to examine age, and the effects:
- Chop up the data by age at rookie season, and examine whether there were any relationships between AAA PAs and wRC+ or fWAR/600 for both rookie seasons and careers within a given age range. Answer: nope. This is actually a pretty important finding, because you would probably assume, given no other information, that perhaps younger players would either be “readier” or “less ready” for the majors for a given amount of AAA PAs, such that there would be a more clear-cut relationship between AAA exposure and performance for players in the same age range. However, that’s not the case: there is too much variation across players to say that AAA exposure has anything to do with outcomes, even when controlling for age. The only point it really starts to matter is for 30-plus-year-old rookies, who appear to be worse players the more time they spend at AAA. But that makes intuitive sense, and isn’t really a huge finding.
- Chop up the data not by age itself, but by a range of ages: 19-23, 24-27, and 28+. The results here were interesting.
- For the 19-23 group, rookie season performance was best among players with 1-199 AAA PAs, followed by 0 PAs, and then with a mishmash of the other AAA PA totals. But, for career performance, there wasn’t much difference between the 0 AAA PA and 1-199 AAA PA groups, and a much clearer pattern of much worse performance for 600+ AAA PAs for these young guys. In other words, there may be some evidence that for young players, having a handful of AAA PAs may help with the adjustment, but in the long run, the aggressively-promoted tend to bear out as better players.
- For the 24-27 group, you kind of get the reverse. For rookie production, far and away the best comes from the 0 AAA PAs group. These are essentially your late bloomers, though there are only a handful of them in the sample. Meanwhile, the worst rookie season hitting performance comes from the 1-199 AAA PAs group. By fWAR, though, they still tend to be better, suggesting that they were perhaps promoted for non-hitting reasons. For career outcomes, the small cluster of 0 AAA PA players still tends to be the best, and the remaining players fall into the comfortable pattern of more AAA PAs equals worse career. So, this kind of “stalling out” in AAA, even for guys who debut well before 30, isn’t necessarily a good sign.
- There aren’t that many rookies that debut at 28+, and most of them manage to accumulate 600 or more PAs in AAA when doing so. As such, it’s hard to draw any real trends, especially since foreign imports (Jung Ho Kang, Hyun Soo Kim, Nori Aoki, etc.) tend to make up the vast majority of the players with limited or no AAA exposure in this group.
(A variety of data and graphs are available for this on demand, but they’re overly confusing to include straight up, given the amount of data sifted through.)
So, there you have it. After all that, I can conclude that most rookies are still playing in AAA, and the virtue of playing in AAA (or not) does not really portend success, failure, disaster, or any particular role for any specific player. There may be some limited evidence that younger players might benefit from at least a bit more AAA exposure before their debuts, but given that these players tend to rebound later in their careers, even the effect of this additional exposure may not really be that relevant or useful for a team trying to put wins on the field immediately. Meanwhile, guys that could potentially be late bloomers (we’re talking your 23-26-year-olds in AAA) tend to have worse major league outcomes the longer they languish in the minors, though a small handful of these late bloomers do go on to make impactful debuts and settle into good careers, often regardless of their previous AAA exposure or lack thereof. (For example, see Matt Carpenter and Todd Frazier.)
In short, I’m not really sure that I learned anything particularly earth-shattering from this exercise, but without doing it, I could only suspect that the relationship between AAA and major league performance was thorny and not straightforward, rather than being certain of it.
As always, happy to extend this in the comments, if the interest is there. Data is available too, if that’s all you want in lieu of me continuing to blather.