Fun With Arbitrary Statistics

There are a lot of great statistics in baseball these days. They measure everything! Swing path tilt? Contact point against center of mass? Exit velocity? Hard-hit rate on opposite field line drives? You can get a report on any of those things at the wave of a few fingers. Expected wOBA? Barrel rate? There are acronyms and stat blends and projection aggregates for days. It’s analysis paralysis; the hard part isn’t finding stats to describe hitters, but figuring out which ones not to use.
I don’t have a foolproof method for seeing through all the noise for you. But I do have a fun experiment in how to work with statistics in such an overwhelming environment. I want to suggest a method that I call, These Things Are Good. Basically, it boils down to this: Come up with some statistics that you think are good without pre-determining which players you’re looking for, measure how good the entire league is at your statistics in question, and see what shakes out.
How should we do this in practice? I started by saying that I think pulling the ball when you hit it in the air is good. That gives me the first statistic I care about: air pull rate. I turned that into standard deviations for every player in baseball so that we could compare apples to apples across multiple statistics, then started looking for who the standouts were across all categories.
Air pull rate is nice, but it’s certainly not the only statistic that I’m interested in. Next, I wanted guys who hit the ball hard when they put it in the air. After all, exit velocity matters far more when batters can elevate. I did the same thing and put that in z-score terms via standard deviations. If you’re wondering what kinds of hitters these two categories identify, Kyle Schwarber, Griffin Conine, and Junior Caminero are the top three.
I might like power, but I also hate grounders rolled over to the pull side, so my next category was an inverse one: percentage of grounders that aren’t pulled. Schwarber, for example, pulls his grounders at a huge clip, 2.55 standard deviations higher than league average. Now Caminero shows up as the hitter who most exemplifies the skills I’m looking for – tremendous damage in the air, hard to shift against on the ground. Conine and Jorge Barrosa round out the podium among batters with 150 or more plate appearances, for the record.
That covered all the contact quality I was interested in looking at, but I can’t in good conscience come up with a filter for good hitters and ignore plate discipline. In keeping with the statistics I’d picked so far, all of which focus on process instead of results, I added two more process plate discipline statistics. First, I looked at swing rate on pitches over the heart of the plate. Corey Seager is the best hitter in baseball at this particular skill – SEAGER and all. Jeff McNeil and Ozzie Albies are great too. But as an example of why heart swing rate isn’t perfect, consider Ezequiel Tovar. He measures out very well by heart swing percentage – but that’s because he swings at absolutely everything.
Thus, I also added swing rate on pitches nowhere near the plate, in the chase and waste zones. I inverted this one as well; fewer swings at bad pitches is better. Tovar, for example, swings at pitches in the chase and waste zones 2.3 standard deviations more often than league average – abysmal. Seager swings at pitches in the chase and waste zones less than average – masterful.
These five stats were enough for me, so I first set out to see who’s above average in all of them. Bo Naylor and Max Schuemann have accomplished both in tiny samples. Four players have managed it in reasonable playing time: Tim Tawa, Zack Gelof, Tyler Soderstrom, and Matt Olson.
That’s a strange list, and it shows one weakness of this first way of looking at things. If you just ask for a binary yes/no, you’ll get guys who are ever so slightly above average in these skills. But that’s not really the mark of a great player. Great players either do many things at a well-above-average level, or do a lot of things OK and have one standout carrying tool. Tawa, for example, is an eighth of a standard deviation above average for air pull rate, an eighth of a standard deviation above average for air exit velocity, and 0.03 standard deviations above average in bad swing rate (above here being the good way). Technically, he’s above average at everything. But in practice, he’s basically average at everything, and those two things aren’t the same.
Gelof, Soderstrom, and Olson each have things to recommend them. They all lift and pull the ball with force, with better numbers than Tawa there, and they all also have solid plate discipline. But I want my metric to show how good players are at a skill, not just whether they’re above average, so I decided to add up all the standard deviations so that I could account for magnitude.
In doing so, I had to make a decision about how heavily to weight each skill. I don’t think it’s fair to say that each is equally important. Hitting the ball hard, for example, is much more correlated with success than not pulling it on the ground. So I made a few common-sense adjustments to the weightings, like so:
| Statistic | Inverted? | Weight |
|---|---|---|
| Air Pull | No | 1.00 |
| Air EV | No | 2.00 |
| Ground Pull | Yes | 0.50 |
| Heart Swing | No | 1.00 |
| Bad Swing | Yes | 1.50 |
The “Inverted?” column just notes whether higher is better; “No” means a higher z-score is better, “Yes” means that I flipped them before using them. Why these particular weights? Eh, because they make sense to me. But that’s not all that rigorous, so I also checked the correlation between this statistic and wRC+ for various weights of each statistic. This blend is pretty good! It’s the best I could find without just running some kind of solver to maximize r-squared. But these weights explain 25% of the variation in wRC+, which is pretty good, and also double the amount of variation explained by equal weights, so I decided that these are good enough for me.
Here’s a quick rundown of how I picked these weights. I first determined that hitting the ball hard had to be the biggest weight and set that at 2.0. For me, not swinging at bad pitches is second-most important, so I decided that had to be the only other weight above 1.0. I personally favor air-pull guys over tough-to-shift groundball types, so I made that the third-most-important weight, tied with heart swing rate, which also appeals to me. (An initial cut of this had heart swing rate lower. I’m unsure what’s best, and again, this is a very unscientific process.) Finally, not hitting pulled grounders matters, but it doesn’t matter that much unless you’re awful at it, so I gave it the lowest weight.
The best hitter in baseball based on this very specific way of measuring things? That’d be Munetaka Murakami. He does the two skills I find most important – crushing the ball in the air and not swinging at bad pitches – extremely well, and while he’s not great at the other three skills I listed, he’s not killing himself there either:
| Player | Air Pull% | Air EV | GB Pull% (-) | Heart Sw% | Bad Sw% (-) | Weighted Score | wRC+ |
|---|---|---|---|---|---|---|---|
| Munetaka Murakami | -0.09 | 3.02 | -0.78 | -0.49 | 1.14 | 6.77 | 150 |
That table looks lonely, though. Let’s add the next nine players (min. 300 plate appearances here, I was picking minimums somewhat arbitrarily throughout the process):
| Player | Air Pull% | Air EV | GB Pull% (-) | Heart Sw% | Bad Sw% (-) | Weighted Score | wRC+ |
|---|---|---|---|---|---|---|---|
| Munetaka Murakami | -0.09 | 3.02 | -0.78 | -0.49 | 1.14 | 6.77 | 150 |
| Nick Kurtz | -1.25 | 2.77 | -0.36 | 0.30 | 1.25 | 6.29 | 144 |
| Heriberto Hernández | 0.86 | 1.58 | -0.79 | 0.79 | 1.21 | 6.22 | 107 |
| Juan Soto | 0.08 | 1.58 | -0.23 | -0.09 | 1.70 | 5.58 | 159 |
| Miguel Vargas | 0.86 | 0.94 | -1.00 | 1.02 | 1.43 | 5.41 | 132 |
| James Wood | -1.45 | 3.19 | 0.06 | -1.27 | 0.93 | 5.10 | 151 |
| Spencer Torkelson | 1.21 | 0.80 | -0.93 | 0.55 | 1.38 | 4.97 | 107 |
| Ronald Acuña Jr. | -0.25 | 0.98 | 0.16 | 1.85 | 0.84 | 4.89 | 118 |
| Max Muncy | 1.61 | 1.40 | -1.94 | 0.13 | 0.82 | 4.80 | 128 |
| Kyle Schwarber | 1.93 | 1.68 | -2.55 | -0.24 | 0.67 | 4.77 | 143 |
There are a lot of very good hitters on this list. In fact, it fits the classic rule that a good statistic is 80% stuff you already knew and 20% surprises perfectly; this list is seven 2026 All-Stars, Ronald Acuña Jr., and then two guys who you might not have thought about even once this year.
Plenty of ink has been spilled about Torkelson’s long and winding major league career. Suffice it to say that the skills that made him the first overall draft pick in 2020 are similar to the kinds of skills that get you on this list. I’m more interested in Hernández, though, so the remainder of this article is less about this interesting-yet-fundamentally-unscientific method of finding good players, and more about one of the good players I found.
Hernández has been a player of note since 2020, when he was traded from Texas to Tampa Bay in the Nathaniel Lowe deal. He was a Top 100 Prospect for us at the time, and after a convoluted path through the minors, he debuted for the Marlins in 2025 with a solid season, putting up 1.3 WAR and a 118 wRC+ in 87 games. He scuffled to start the 2026 season and briefly returned to Triple-A, but since returning, he’s played his way into an everyday role. He’s hitting .250/.314/.529 since his May 7 call-up, good for a 126 wRC+, and has been one of the best hitters on the Marlins as they’ve surged back into playoff contention after a rough 12-game skid last month.
It’s not hard to understand why Hernández is succeeding this year. As I’ve already laid out in the premise of this article, he basically does all the things that I like hitters to do: He chases less than average, swings at pitches in the zone more than average, and does plenty of damage when he connects.
You’ll probably notice that I left one key part of hitting out when constructing my formula: Making contact. I did that because I couldn’t think of a great contact number to use, and also because I don’t think it’s particularly correlated to overall success after you’ve already controlled for things like making good swing decisions and hitting the ball hard when you connect. Hernández is a good example of what I mean here. He’s coming up empty fairly often when he offers; Baseball Savant has him in the sixth percentile for whiff rate. But, well, Schwarber is in the fifth percentile for whiff rate. Kurtz is in the first percentile. Plenty of the very good power hitters identified by this method swing and miss a lot; they just take good swings and hit the ball hard when they make contact.
Hernández is well on his way to doing just that. His patience isn’t in question; he posted double-digit walk rates in every year of his minor league career. But when he was striking out 30% of the time in the minors, it was easy to imagine how this might all fail. In 2022, he posted a 31.4% strikeout rate in High-A; that’s not the kind of line that translates well against big league opposition.
As it turns out, though, a change in approach was all he needed. That version of Hernández was too passive; his 42% swing rate was way below big league average, and he posted numbers in that range in 2021 and 2023 as well. Since then, he’s become less selective, chasing more but also getting more high-value swings off. His swing rate climbed to 46% in Triple-A in 2024, 45% in the majors in 2025, and 46% in the majors this year.
There are good and bad ways to increase your swing rate, but Hernández did it the good way. He’s still quite patient outside the zone, but he’s swinging at strikes more often. He’s not making an incredible amount of contact against those strikes – his 78.6% zone contact rate is eight percentage points below average. But he’s swinging enough, and with enough force, that pitchers can’t just chuck it down the middle early in the count and then spam secondaries late.
Another way of thinking about it: You can’t control where pitchers throw you the ball, but you can certainly give them incentives. Hernández’s old combination of low contact and low in-zone swing rates made pitchers’ jobs relatively easy; he wasn’t going to chase, but he probably also wasn’t going to take early swings at pitches in the strike zone. Now, though, the incentives are confusing. He swings more than average at pitches over the heart of the plate, and even if he misses a few, tossing cookies to a guy with 80th-percentile bat speed and a penchant for lifting and pulling is a bad idea. On the other hand, he doesn’t chase a lot, and if he gets ahead in the count, he’ll especially look to do damage. It’s a tough spot for opposing arms.
This isn’t a style of hitting that just anyone can adopt. Hitters with less power on contact would find themselves overwhelmed by a stream of pitches in the strike zone. Hitters without Hernández’s excellent sense of the strike zone would drown in sliders below the zone. But he’s sitting in a happy medium right now; the obvious solutions against him basically aren’t working. Even with his bottom-shelf contact rate, he’s been excellent since returning to the majors, and from my perspective, he might be even better than his results. Neat finding for a very silly, very ad hoc statistic that I think nonetheless captures a lot of the best hitters in baseball – a group that Heriberto Hernández might be a part of.
Ben is a writer at FanGraphs. He can be found on Bluesky @benclemens.
I assume Conine misses the top 10 list only because of the 300 PA min? His plate discipline looks solid to me and you already said he was top 3 when just looking at the batted ball metrics.
Fun fact: Conine has at least a .380 wOBA in every month he has played (he missed most of April and June and all of May with a hamstring tear but has hit extremely well since coming off the IL in June)
The Fish have an interesting roster setting them up nicely for the next few years.
Forget Heriberto. Conine is the real breakout star of this piece!
Heart swing
Air Pull
Bad Swing
Air Evo
Ground Pull
I christen your new stat HAirBAG.
As far as I’m concerned this is canonical now
Love it! I worked at EPA in the 1980s. We entitled our guidance manual the “Superfund Health Assessment Manual”, just for the acronym😎.
This is a really fun rabbit hole Ben!
Heriberto so far looks like a lefty crusher who really struggles with righties. He could be destined for a short-side platoon role. Tork is a bit like that as well, though not quite as extreme. I’m guessing that guys on the border of an above-average wRC+ are enriched for platoon candidates.
100%: you may even be able to split this approach out by opposing handedness to tease out who’s truly hopeless against same side pitching: if all these things are good then presumably doing them poorly has a lot of signal too and doing them meaningfully differently with or without the platoon advantage also feels more meaningful than observed splits. Maybe it would even stabilize sooner / need less regression?
Oooh, I want rolling 15 game window plots of this stat / things like it too. It’d be fascinating looking at things like this and SEAGER and other process composites for breakout or hot streak leading indicators
Not really. A .832 vs .734 difference in OPS is not predictive at all for a sample of only 600 career PA.
That’s fair–I can’t find his aggregate minor league splits, but they don’t seem super pronounced year-to-year. I stand by Tork being a short-side platoon bat though. He really doesn’t hit enough against righties to be worth it for a first baseman.
A fascinating way to discover talent that others might not find. I hope Chaim is the only team employee that reads this😎
Great stuff! This gets me thinking about how the “Overton window” for lack of a better term keeps getting shifted on what an MLB viable K rate upper limit is. 20%, 25%, and now 30% were formerly disqualifying numbers in prior eras. The interplay between doing damage on contact and making a sufficient amount of contact to capitalize has always been there and I’d hypothesize it was harder to adequately balance these forces via analytics before we could quantify how extreme and damage inducing the contact is.
30+% k rate guys that mash are still uncommon but not unheard of nowadays but I feel like these talents may have existed in prior eras and been overlooked or given insufficient opportunities due to a heuristic?
Flip side argument: just how historically bad was Rob Deer’s contact ability to manage K’ing 30+% pretty often in the 80’s when a league average K rate for a pitcher was 50-65% of what it is now?
I don’t know, 30% is still a pretty good barrier to success. Per FG leaderboards, there have been 67 qualified non-pitcher hitters with a 30+% K-rate (qualified looks like 1,000 PA). Excluding catchers who benefitted from the positional adjustment (Tyler FLowers and Mike ZUnino) and a 1B who last played prior to WWI (Jake Stahl), there are no hitters with a career 30+% K-rate with more than Rob Deer’s 14.3 fWAR. There aren’t even many guys at 10+ fWAR. And the list is littered with recent “he might be good if he could just strike out a bit less” notables such as Tyler O’Neill, Franmil Reyes, Mark Reynolds, Chris Davis, Chris Carter, Keston Hiura, Miguel Sano, and the patron saint of high-K hitters, Joey Gallo.
Oneil Cruz is on the list and while he’s still young enough that maybe he reins it in, we’re now 6 years and over 1,800 PA down the road of decent-but-not-great.
You’re just fighting too much math if you are an automatic out 3 of every 10 trips to the plate, even for a guy with insane batted ball metrics like Cruz.
(I note that James Wood is on the list too, but I think he’s already showing this year that he is bringing that K-rate down under 30%. Remember that Judge was over 30% his 1st few years.)
Great numbers / this also maybe makes my point in a more nuanced way: all the names you mentioned are players of more recent vintage and while most / none have had extended runs of MLB level success, they have all been MLB players of some note for some period of time while Wood and Cruz may be the first 10+ WAR guys of this career profile since Deer?!?
Put another another way: how many guys even got as much run as this cohort in the 80’s and 90’s? How many guys that had “prohibitive” K levels back then may have had early career judge / wood paths to success if the range of what “plays” was differently understood? These may not often be durably great profiles but even at a lower level more José Siri / Domingo Santana / Christopher Morel brief peaks are being realized in an MLB that is more tolerant of outcomes with a wider range of shapes that lead to the same aggregate production.
Like anything else, K rates should be normalized by that season’s MLB averages.
Totally agreed: is there an easy way to sort all time K%- or K%+ seasons with a certain threshold # of PA? Deer has to be an all time leader or right there given his Gallo-esque game in a league that had a step change less TTO vs today.
See below with the link to the query
Found it (filtered since 1960): right there with multiple Gorman Thomas and Dave Kingman campaigns, and a 7.9 WAR Mike Schmidt season?!?:
https://www.fangraphs.com/leaders/major-league?stats=bat&lg=all&qual=y&type=23&month=0&ind=1&team=0&rost=0&players=0&pos=np&sortcol=5&sortdir=default&pageitems=200&startdate=&enddate=&season1=1960&season=2026
As a Lazaro Montes believer, I believe you have created a metric for Lazaro Montes believers.