Where Are 2026’s Extra Walks Coming From?

A few weeks ago, I presented some in-depth research on the size of the 2026 strike zone. The results were clear and unambiguous: The called strike zone is smaller this year than it was last year, and most of that shrinking is coming at the top of the zone. But saying that the strike zone is smaller is different than saying that the smaller zone is causing the overall major league walk rate to increase, and walks are up by a lot this season. Last year, batters walked in 8.4% of their plate appearances. This year, through May 8, they’ve walked in 9.5% of plate appearances. Still, walk rates move around all the time for reasons unrelated to the strike zone. That meant I had another question to answer: Are the walks coming from the smaller strike zone, or are they coming from something else?

First, I decided to look for which counts have had the greatest impact on the increase in walks. To do so, I used a technique called Markov chain decomposition. Think of each plate appearance as falling through a Plinko board. Every plate appearance starts at 0-0, and then it progresses in one of four ways: ball, strike, ball in play, or hit-by-pitch. Ball in play and hit-by-pitch results end the plate appearance, of course, but ball and strike outcomes on 0-0 feed into other buckets: 1-0 and 0-1 counts. In each of those counts, the same thing happens, with the next pitch resulting in either a ball, strike, ball in play, or hit-by-pitch. That keeps happening – with foul balls behaving like do-overs in two-strike counts – until you get to three strikes, four balls, a ball in play, or a hit-by-pitch. The reason that this is helpful is because you can start with small events – balls, strikes, balls in play – and build bigger outcomes, like walks and strikeouts. In that way, you can use per-pitch results to learn things about per-plate-appearance results.

That’s a Markov chain. To figure out how much each count’s changing results are contributing to the change in walk rate, we need to do a little decomposition, which means that another example is in order. Imagine a 2-2 count. Next, imagine that the only possible results are ball and strike. Further, imagine that there’s a two-thirds chance of a ball on 2-2, and a 50% chance of a ball on 3-2. You can work out the odds of a walk – one-in-three – and the odds of a strikeout – two-in-three – from those numbers. Now, let’s imagine a world where the walk rate balloons from 33% to 40%.

How can that happen? One of two ways: batters reaching 3-2 more frequently, or batters walking more frequently when they reach 3-2 counts. If 2-2 pitches go from being balls two thirds of the time to being balls 80% of the time, the walk rate would hit 40% without anything at all changing in 3-2 counts. Likewise, if 3-2 pitches go from being balls half the time to being balls 60% of the time, the walk rate would hit 40% without anything at all changing in 2-2 counts. In both of those scenarios, the walk rate goes up by the same amount, but in each case, the change in walk rate can be directly attributed to changing behaviors in a given count. As the likelihood of each individual result in each count varies, a Markov chain can calculate how much that affects the overall results.

In real life, the decomposition is a bit more complex, because there are more intermediate states and more outcomes, and because the results in each count are all changing at once. But that’s really just a matter of more math; it doesn’t alter the core concept. That means that you can look at a change in walk rate between two years and break down which counts are contributing to it the most. I did just that. I took every pitch from the 2025 and 2026 seasons and used them to create Markov chains. Then I decomposed them by count to see what’s going on with more granularity:

There’s an easy story here. Walks aren’t increasing because hitters are recovering from disadvantageous counts more frequently. Walks are increasing because when hitters get ahead in the count, they’re turning that advantage into a walk more frequently. The biggest contributing count is 3-2, with 2-0 and 3-1 close behind. It’s interesting to see 0-0 in the mix, but I think it’s very notable that four of the five counts that are contributing most to the higher walk rate feature more balls than strikes. The only reason 3-0 isn’t on that list is because the count hits 3-0 fairly rarely; it can’t contribute much.

Digging into why results in each count are changing requires leaving our Markov chain behind. If you compare 3-2 counts from 2025 and 3-2 counts in 2026, balls are happening 1.4 percentage points more often. Strikes are happening about one percentage point less often (the reason these don’t match the per-plate appearance results is that foul balls lead to a redo). But that doesn’t tell us why we’re getting more balls. To learn more, we’ll have to start integrating pitch location and batter behavior.

I’d say we should start with zone rate, but we run into a problem right away: “Zone rate” doesn’t mean the same thing anymore. There’s a new strike zone in town. And even putting aside the fact that the zone is being called more tightly, the zones listed by Statcast on each pitch have changed. I did a quick test: I took all the batters who have appeared in both 2025 and 2026, and measured the change in the listed height of their strike zone in those two years. If you weight it by the number of pitches that they faced in 2025, the aggregate league-wide strike zone, as defined by ABS, is about three inches shorter than it was last year, with most of the decline coming at the top of the zone. Only three batters in all of baseball have taller strike zones in 2026 than in 2025.

Since zone rate is a moving target, we’ll have to measure pitch locations relative to one consistent zone. I chose to use the 2026 zone, but really, we could use either. The key here is that we have to make sure we’re comparing apples to apples, as it were. That’s because we need to distinguish between two effects: pitchers throwing to the same place but getting called balls where they used to get called strikes, and pitchers throwing to less central locations.

I broke up the strike zone into 14 regions. There are four “just inside the zone” regions, four “just outside the zone in one direction” regions, four “just outside the zone, on the corner” regions, and then the heart of the zone and far from the zone. Using a consistent zone, pitchers are throwing the ball outside the strike zone slightly more often in 3-2 counts this year:

For the record, “far outside” is defined here as far enough out of the regulation zone that a take will almost never lead to a called strike. I chose one inch as the cutoff for the size of my “just inside” and “just outside” zones, which worked fairly well to differentiate between close calls and easy ones. In 2025, only 2.3% of taken pitches in the “far outside” zone were called strikes. In 2026, only 0.8% of them have been called strikes, out of a sample of more than 3,000 pitches.

Not every one of those “far outside” pitches gets taken, of course. Here are swing rates in each region on 3-2 pitches in 2025 and 2026:

You don’t have to worry too much about the changes in swing rates on corner pitches, because pitchers have only hit the corners a combined 19 times in our 2026 sample. It’s just not a very frequent area of attack on 3-2 – and really, we’re talking about hitting one-square-inch targets, so it’s not a very frequent area of attack generally.

I performed a more complete analysis by working out how many pitches batters took in each region in 2025 and 2026, accounting for both changing pitcher behavior (where they locate the ball) and batter behavior (how often they swing). In 2025, 23.4% of 3-2 pitches resulted in hitters taking a pitch that was located outside the consistent strike zone we defined. In 2026, 24.5% of pitches have resulted in hitters taking a pitch located outside the consistent strike zone. That adds to the rate of called balls, but not by 1.1 percentage points. That’s because not every pitch outside of the strike zone is called a ball, and vice versa:

That’s right: The areas at the fringes of the strike zone are being called differently. It’s not so much that the areas where strikes are most frequently called have moved (with the exception of the area just above the top of the zone, which as previously noted, is where the zone is shrinking). The difference is that balls outside the zone are being called strikes less frequently than before, while balls inside the zone are being called strikes more frequently than before.

Let’s set aside the top of the zone for a moment. On the other three edges, the transition from 2025’s all-umpire strike zone to the 2026 challenge/umpire hybrid zone has been, well, striking. Balls that are just barely in the strike zone on those three edges were called strikes 56.7% of the time in 2025; they’re being called strikes 76.7% of the time in 2026. Balls just off those three edges were called strikes 37% of the time in 2025; they’re being called strikes 13.5% of the time in 2026. In other words, the strike zone is getting less fuzzy. The shape is only changing at the top, but the number of incorrect calls in a given area is declining across the board.

How does that lead to an increase in walk rate? It’s a neat little mathematical relationship. The closer a pitch is to the center of the strike zone, the more likely a batter is to swing, particularly in two-strike counts. That means that an increase in accuracy across the board will add more balls than strikes, because there will be more takes, and thus more chances for the umpire to call a ball or strike, on pitches located outside of the strike zone.

Take the example we just used. Swing rates on the inside, outside, and bottom edges of the zone – but still in the zone – hover around 75%. Swing rates on pitches just off those edges are around 70%. That’s a small but non-negligible effect from a one-inch difference in location – and it’s bigger in counts that don’t feature two strikes, where swinging at a ball in the strike zone is optional. There’s an even bigger difference between pitches over the heart of the plate and pitches outside the strike zone. Centrally-located pitches are being called strikes 1.2 percentage points more frequently in 2026 than they were in 2025, while pitches far outside the zone are being called strikes 1.5 percentage points less frequently. But batters swing at 90% of the strikes and only 40% of the balls, so the net effect is that improving ball/strike accuracy in three-ball counts leads to more walks.

There are three effects driving the change in outcomes on 3-2 counts this year: pitcher/batter behavior, a change in the definition of the top of the strike zone, and increased call accuracy. I mathematically decomposed those into three parts using a simple test. First, I calculated what the walk rate would be if we took all of the actual pitches, swings, and takes from 2026, but used the called strike rates by zone from 2025 (based on the consistent strike zone definition detailed above) for taken pitches. This explains how much the walk rate would increase merely from changes in batter/pitcher behavior with a constant strike zone. A methodological note here: I only considered pitches thrown to batters who appeared in both 2025 and 2026 so that I could standardize the size of the strike zone for our analysis. That means that the overall numbers differ slightly from league-wide rates, though the divergence is minimal.

Next, I took the relevant pitches from 2025 and used the 2026 called strike rates for the top of the strike zone and the 2025 called strike rates for the rest. That gave me the increase in walk rate you’d expect if the only change was the shape of the top of the zone. Finally, I took the relevant pitches from 2025 and the 2026 called strike rates for everywhere except the top of the zone, where I kept the 2025 rates. That gave me the increase in walk rate you’d expect from increased ball/strike accuracy. I found that you can attribute 0.9% of the increased rate of 3-2 balls to changing batter/pitcher behavior, 0.1% to changes in calls at the top of the strike zone, and 0.4% to changes in correct call frequency in the rest of the strike zone.

That analysis explains the change in 3-2 results. To understand the whole picture, I just repeated the calculation for every count. That gave me values for how much changes in batter/pitcher behavior, changes at the top of the strike zone, and increased call accuracy changed the rate of balls and strikes in each count so far this year. Then, to complete the circle, I fed this data back into our Markov chain from above; I ran hypothetical Markov chains for each of the three effects independently, which let me calculate the change in overall walk rate attributable to each.

In the aggregate, you can split the change in walk rate into three parts. One is a change in pitcher/batter behavior. This covers changes in where pitchers locate, how frequently batters swing in each location, how frequently they make contact, and how frequently that contact is fair. Those changes have added 0.5 percentage points to the overall walk rate. Next, changes in the shape of the top of the strike zone have added 0.2 percentage points. Finally, an increase in the accuracy of calls has added 0.4 percentage points to the overall walk rate. That’s the headline finding of this study: Walks are increasing for three different reasons, all working in concert.

The next question I had was how much of that increased accuracy is due to challenges – not the overall challenge system, but specifically the pitches that players have challenged and in some cases overturned. There’s an easy way to test this: I just told my computer to take the original umpire calls instead of the final calls. The results are both interesting and intuitive: ABS challenges themselves have actually decreased the walk rate. That’s not surprising – more balls have been overturned into strikes than the reverse – but it sounds funny when you say it out loud. MLB switched to an ABS challenge system this year, and the direct effect of that system is slightly decreasing walk rates. Also, walk rates have increased by a striking amount, and more than half of that is attributable to changes in the way that balls and strikes are called, which appears to be an indirect effect of the ABS challenge system. Isn’t that weird?

Finally, I performed some analysis to ensure that my findings are robust. I varied the sizes of the slices I used to define the various zones in this analysis. Regardless of how large or small I made those slices, the contribution of pitcher and batter behavior to walk rate was stable at around 0.5 percentage points. But the relative contributions of the top of the zone and of increasing accuracy changed; the larger I defined the top of the zone to be, the more effect it had. For very large definitions of “top of zone,” the effect was roughly equal in magnitude to the effect of increased accuracy. In other words, it’s difficult to disentangle exactly how much of the walk rate increase can be attributed to increased accuracy of an existing zone and how much can be attributed to a change in the size of that zone, but both factors are important, and I think it’s quite likely that the accuracy component is of slightly greater import.

So 3,000 words in, what does it all mean? This year’s strikingly high walk rate isn’t just about pitchers and batters behaving differently, and it isn’t just about the size of the strike zone. It’s both, and it’s also about umpires making calls more accurately. I think that’s why the increase appears so dramatic; lots of things are all changing at once, and they all happen to be changing in the same direction.

This isn’t a stable equilibrium. Both pitchers and batters will continue to adjust to the new way that balls and strikes are being called. Batters are swinging less frequently this year, and pitchers will likely adjust to that by throwing in the strike zone more frequently. Now that the rewards to fishing off the edges have declined thanks to an increase in call accuracy, attacking the zone is being rewarded even further. And batters don’t have to take those potential changes lying down. If pitchers start throwing in the zone more frequently, batters will likely increase their aggression.

I’m not sure where walk rate is headed. But I do feel confident in saying that plenty of this year’s increase comes down to a change in the way balls and strikes are called. I also feel confident that a majority of that effect is about the increased accuracy of calls rather than a change in the size of the strike zone. Finally, challenges themselves aren’t contributing to this change; taken in isolation, they’ve actually decreased walk rate.

As is customary, I’ve included the dataset and Python code used to generate these results here. The study can also be expanded to previous years or run on different data; in fact, I couldn’t upload the 2025 data to GitHub for size reasons, so you’ll need to download that yourself. You can also replace those with your own similarly-formatted data if you’re interested in expanding the analysis.