Pitching is about keeping hitters guessing — and about walking the line between overusing certain pitches to the point of predictability and underusing others that have quietly confounded opponents in limited doses. Now more than ever, each MLB pitcher’s repertoire is scientifically calibrated, from the shape of the ball’s arc as it approaches the plate to the spin it carries and how it looks coming out of the hand. Modern pitchers take their pitch selection as seriously as a Michelin chef planning a gourmet menu.

But even with all of that sophistication, there are inefficiencies in how pitchers deploy their stuff. Many years ago, I dove into the game theory behind pitch selection, and specifically which pitchers were throwing their different pitch types in an optimal way versus those who could stand to tweak their pitch mix a bit to achieve better results.

The thought process went like this: We know from Statcast data how frequently each pitcher throws each type of pitch, and thanks to websites such as FanGraphs, we also know how effective each pitcher’s pitches have been at preventing runs. (We now even know how good each pitch should be based on its characteristics, such as velocity, movement, spin and other factors.)

From this data, we can then find cases where there are mismatches between a pitcher’s most effective pitches and the ones he uses the most.

Of course, not every pitch can be scaled up without diminishing returns. But in general, pitchers who lean more heavily on their best pitches are likely getting more out of their repertoire than those who don’t.

I then developed what I call the Nash Score for pitchers (so named for the Nash equilibrium of Game Theory, which describes a state in which any change in strategy from the current balance would result in less optimal results). Nash Scores work by comparing the runs a pitcher saves with each pitch in his arsenal to the average runs saved by all of his other pitches combined.

Pitchers with low (good) Nash Scores have achieved a close balance in effectiveness between their most-used pitches and the rest of their repertoire, which implies that any change in pitch mix would make them less effective overall. Meanwhile, pitchers who have high (bad) Nash Scores are either using ineffective pitches too much or not using their best pitches enough, suggesting that a reallocation might be needed.

Now is a good time to update Nash Scores for the current era of MLB pitchers.

Let’s highlight the top-15 qualified starters and relievers who have achieved the greatest balance according to their Nash Scores over the past three seasons (with recent years weighted more), as well as the 15 who might be leaving performance on the table.

But first, here is a chart showing all qualified MLB pitchers — using a three-year weighted pitch count — with their Nash Scores plotted against their Wins Above Replacement:

Explore the full, interactive chart.

Now, let’s get to the rankings, starting with the most balanced starters in our sample:

Irvin, Crochet among most optimized starters

Note: Listed rates for pitch types are usage share over the past three seasons and run values per 100 pitches for that pitch, relative to the average for the rest of their pitches combined.

The award for the league’s most balanced starter belongs to perhaps an unlikely name: Washington Nationals righty Jake Irvin. Irvin has been an average pitcher at best in his three MLB seasons, with an ERA of 107 (100 is average and lower is better) and a FIP (Fielding Independent Pitching) of 114, and he has never even had 2 WAR in a season yet. But in terms of maximizing his repertoire, the case can be made that no pitcher is getting more out of what he has to work with.

Over the past three seasons (again, with more weight on more recent data), Irvin has almost exclusively used three pitches: four-seam fastball, curve and sinker. Each was within 0.2 runs per 100 pitches of the average of his other offerings, meaning he found the mix where basically all of his pitches are equally effective — the whole point of this entire exercise.

Now, Irvin has drifted a bit away from equilibrium in 2025, using more of his curve (and less of his fastballs) despite them being more effective, so it’s worth keeping an eye on whether he continues to optimize his Nash Score. (Especially since his best-shaped pitch is actually his slider, which he almost never uses!)

Among the rest of the top 15, several other pitchers showed a knack for maximizing their stuff. Garrett Crochet — the nasty left-hander who broke out last year and was dealt from the Chicago White Sox to the Boston Red Sox — pairs an elite fastball with an even more dominant cutter (plus a bit of a sinker-slider), giving him one of the game’s best (and most equalized) pitch mixes.

Fellow Red Sox hurler Kutter Crawford follows the same template, with similarly effective four-seamers and cutters making the bulk of his repertoire. Others strike the balance differently: Jesus Luzardo and Freddy Peralta use more off-speed stuff, while Ryan Pepiot and Corbin Burnes rely on strong fastballs as their primary pitches — but only use them about half the time. And then there are guys such as Taj Bradley and Taijuan Walker, who lead with shaky main pitches, but throw them so infrequently that the rest of their pitches help equalize the overall mix.

It’s also no surprise to see Tarik Skubal, arguably the best pitcher in baseball, grace a list of hurlers who pick from their arsenals in the most efficient way. What everyone on the list has in common is a pitch selection largely in equilibrium, where effectiveness and usage are closely aligned.

Sewald, Poche among most optimized relievers

You’ll likely notice that the top relievers tend to be more optimized (with lower Nash Scores) than the top starters, which is probably an artifact of a few factors: First, relievers usually throw just a couple of pitch types, so it’s inherently easier to align usage with effectiveness when there’s less to balance. Second, those pitches are often thrown in short bursts at maximum intensity, which allows pitchers to rely more heavily on their strengths without diminishing returns. And finally, relievers don’t need to navigate a lineup multiple times, so they can lean on their best pitches more without the same concerns about stamina or predictability that starters face.

That said, some relievers do a better job of balancing than others. Though he has been nursing an injured shoulder since April, Cleveland’s Paul Sewald had been the best over the past few seasons — the two pitches he used 99.7% of the time, a four-seamer and a slider, were both within five hundredths of a run of each other in terms of effectiveness per 100 pitches. The batter knows one is likely coming… but they’re both equally tough to hit.

This was a very common theme among the top relievers, too: Each of the next four names on the list (Colin Poche, Tanner Scott, Joe Jimenez and Alexis Díaz), and eight of the top 11, used a version of that same pitch mix, with fastballs and sliders of near-equal effectiveness making up the vast majority of their pitches. Hey, if it works, it works.

But those who bucked the trend are also interesting. Philadelphia’s Orion Kerkering, for instance, flipped the tendency and relied mostly on a slider with the four-seamer as a change-of-pace pitch. Milwaukee’s Elvis Peguero was exactly 50-50 on sliders and sinkers (though both abandoned him earlier this season, and he has bounced between MLB and AAA), while Nats closer Kyle Finnegan introduces a splitter into the equation — and there’s longtime veteran closer Craig Kimbrel with his knuckle-curve (though it hurt his Nash Score).

Not all of these relievers have been lights-out, but many were, serving as great examples of how to stay effective even when hitters have a good guess at what’s coming.

Blanco, Kelly among least optimized starters

Now we get into some truly fascinating cases, where it’s important to remember that you can still be a great pitcher while still having a deeply strange, and seemingly suboptimal, mix of pitches.

There seem to be a few ways to land on this list: First, and most straightforwardly, you could have a far less effective No. 1 pitch than the rest of your arsenal, meaning you might stand to throw it less and the others more. Both of the top two above, Houston’s Ronel Blanco and Arizona’s Merrill Kelly, have primary four-seamers that are at least 1.5 runs worse per 100 pitches than their other options, and secondary off-speed pitches that are at least 2.4 runs better than the rest — classic cases where the Nash Score would suggest bringing them closer to balanced until the difference begins to flatten out.

Then there are cases such as Joe Ryan, Michael Wacha, Dylan Cease, Chris Sale and Michael King, in which their No. 1 option is clearly the best, but they throw other, much less effective pitches nearly as much, reducing the advantage of a dominant primary pitch. Spamming the top choice might lead to diminishing returns, but there’s room to give there before it starts being a suboptimal strategy.

And finally, we have the odd case of Paul Skenes — and Gavin Williams too, but Skenes is more fun to dissect — in which somehow the primary four-seamer is less effective than the other pitches, and so is the secondary breaking pitch, suggesting the need to dig deeper into the bag more often. But how can you argue that Skenes isn’t doing the most he can? He literally leads all pitchers in WAR. The thought he could optimize his stuff even more is terrifying.

Kahnle, Bender among least optimized relievers

Finally, we get to the less optimal end of the reliever spectrum. And as stable as the opposite side was, with a bunch of guys using their boring fastball-slider combos to carefully record outs, this one contains more varied pitch mixes. Well-represented, for instance, is the phenomenon I found with R.A. Dickey the first time around — that despite his knuckleball being both his best pitch and the one he used most often, the Nash Score implied he should throw it even more because it was much more effective than the rest of his offerings.

While we don’t have any knucklers in the bunch this time, we do have guys such as Detroit Tigers setup man Tommy Kahnle, whose lead pitch is a changeup (not a fastball) so effective that it’s nearly four runs per 100 pitches better than the rest of his repertoire. Pitchers who work backwards like this must mix in fastballs to keep hitters honest — but at the same time, the fastballs are much less valuable that using them slightly less might be good even if it makes the change less effective. (Anthony Bender, Brenan Hanifee, Steven Okert, David Robertson, Greg Weissert and Cade Smith were in this category as well, among others.)

Just as odd were the cases of Ryan Helsley, Justin Lawrence and John Brebbia, whose primary pitches were far less effective than their secondary options, despite each essentially having only two pitches to work with. The numbers might be asking for those hierarchies to be flipped around.

And finally, there are guys such as Kenley Jansen, who spam one solid pitch — but they don’t have much else to work with, so any deviation worsens performance, even if the Nash Score still dings them for imbalance.

In the end, no metric — not even one rooted in Game Theory — can capture the full complexity of pitching. But Nash Scores do give us a window into something that’s often hard to pin down: How much a pitcher gets out of what they’re working with, and whether they’re winning the rock-paper-scissors aspect of the batter-pitcher showdown.

Some get the most out of average stuff through smarter allocation. Others leave value on the table despite electric arsenals. In either case, the path to better performance might be as simple (or difficult) as throwing the right pitch at the right moment just a little more often.

Four-star prospect Jett Washington, ESPN’s No. 1 safety in the 2026 class, announced his commitment to Oregon over Alabama and USC on Thursday night.

Washington is a 6-foot-5, 210-pound defender from Las Vegas and the nation’s No. 22 prospect. The top-ranked four-star recruit in the 2026 ESPN 300, he trails only five-star tight end Kendre’ Harrison (No. 11 overall) as the highest rated of the Ducks’ six top-300 pledges in the current cycle.

Washington enters the fall as a third-year starter at Las Vegas powerhouse Bishop Gorman, where he logged 38 tackles and five interceptions during his junior season in 2024. A nephew of the late Los Angeles Lakers legend Kobe Bryant, Washington helped lead Bishop Gorman to a second consecutive Nevada Class 5A state basketball title earlier this year.

Washington’s pledge is a much-needed recruiting win for Ducks coach Dan Lanning after a string of near misses with top-end talents this spring.

Last month, Oregon whiffed as finalists for five-stars Jackson Cantwell (No. 3 overall) and Jared Curtis (No. 5). As recently as Thursday morning, five-star athlete Brandon Arrington (No. 14) picked Texas A&M after a hotly contested race between the Aggies and Ducks.

But Oregon, alongside Alabama, was among the first major programs to enter Washington’s recruitment last year. Through Washington’s connection with Oregon defensive backs coach Chris Hampton, the program held an edge heading into Washington’s slate of official visits with the Ducks, Crimson Tide and Trojans.

“I have that great time with [Hampton] and the coaching staff,” Washington told ESPN this month. “I loved the way they treated me out there. My relationship with them has been going the right way ever since I started talking to them.”

With his pledge, Washington becomes the top-ranked defender in Oregon’s 2026 class. He is one of three ESPN 300 commits on defense for the Ducks, joining defensive tackles Tony Cumberland (No. 89 overall) and Viliami Moala (No. 262).

Leaders of the College Football Playoff are still “mulling over” what the format should look like in 2026 and beyond, CFP executive director Rich Clark said Wednesday following the conclusion of two days of meetings in Asheville, North Carolina.

“Pretty much everything’s on the table, and they’re taking a good look at it all of it,” Clark said. “So, I wouldn’t say there’s a leading contender right now for them, but they’re taking a fresh look at it.”

Last month, at the SEC’s spring meetings in Destin, Florida, there was new support from the head football coaches for a 16-team model that would include the five highest-ranked conference champions and 11 at-large teams. That quickly gained traction and public support from other leagues, but it also surprised many leaders in the Big Ten Conference.

Many athletic directors in both leagues had been aligned in their desire to have automatic qualifiers that would guarantee both conferences four spots each in the playoff, with play-in games to determine whom the third and fourth playoff participants would be. Following the SEC’s spring meetings, many sources in the Big Ten have indicated that they wouldn’t even consider a 5+11 model unless the SEC and the ACC both go to nine conference games.

The Big Ten and the SEC have the bulk of control of the next iteration of the playoff, but the two conferences haven’t been on the same page recently in terms of what that should look like.

“They’re obligated to come to an agreement on what the format is,” Clark said.

SEC commissioner Greg Sankey has said repeatedly that he wanted more clarity on the selection process before determining if the league should move to a nine-game schedule, and much of this week’s meetings were spent studying metrics, including strength of schedule. Clark said the CFP brought in a mathematician from Google to help the group, which included all 10 FBS commissioners and Notre Dame athletic director Pete Bevacqua.

“I would say a lot of it is refreshing our metrics,” Clark said. “Some of these metrics were put into place with a whole different look at the way college football was laid out and the conferences were structured.

“I wouldn’t say it’s dramatic, but it’s a refresh. We looked at some of our processes and how we do things within the selection committee meeting room. Some of the things we brought up to the table for them to consider, but just things to help us get better.”

Clark reiterated something Sankey said at his spring meetings and multiple other sources have surmised: The current 12-team model remains an option for 2026 and beyond. Whatever they decide, Clark said, they’d like to do it by the fall, and they’re not looking to postpone the decision by keeping the status quo for another season.

“I don’t think they’re going to kick the can a second year,” Clark said. “They’re going to make a decision on what they think that whole period should look like.”

The CFP did decide this week to incorporate sports wagering monitoring and will have its staff determine the details of it. The playoff also will use player availability reports, which would be handled in a similar way to what some conferences already do, Clark said.

“We’re going to take that model on,” he said. “We have some work to do on that, though, to perfect it and to ensure that we’re doing it the right way, but that will be an important position for the CFP.”

Clark said there has been progress but that the group still has “lots of space to still make some decisions.”

“I don’t know if there’s any hurdles they have to get over; I think they just want to make sure they get it right,” Clark said. “That’s going to be six years of format that they’re deciding, and rushing to a bad decision is not in any of our best interests. They want to make sure they look at all the options and understand what the pros and cons are and make the best decision they can rather than trying to rush to something that may not suit us for the next phase of the CFP.”