Swing Away

I remember when I was assistant coach on a 15U travel team in Québec, the organization’s head coach said “Of course we swing on 3–0. But it depends when.” I called for a green-light on a 3–0 count and then got shouted down by the coach — I still don’t know what he wanted. That was 12 years ago and I’m still upset.

But the fact is that swinging on 3–0 counts is largely frowned upon in baseball because of some commitment to the unwritten rules. The UnWriTtEn RuLeS. I suppose the argument is that the hitter is up 3–0, the pitcher is trying to battle back. The hitter should allow the pitcher the opportunity to fight his way back into the count.

Ludicrous. But the logic behind why that’s ludicrous doesn’t need to be re-iterated. Defenders of the unwritten rules and opponents of it have gone on about this ad nauseam.

The reality is, however, if norms in baseball ever change, 3–0 is going to be the best count a hitter is going to see. Note the figures below. The figures use Jim Albert’s CalledStrike package and plots the heatmaps for pitch location for 3–0 accounts from 2015–2019. I excluded 2020 because it was a short season. The heatmaps for all seasons are clearly demonstrating something: the hitter is probably getting a meatball right down the pipe.

So what gives? Hitters are seeing 3–0 pitches come right down the centre of the plate and the table below demonstrates that the percentage of these pitches being fastballs has increased over time as well. Here, I combined two-seam and four-seam fastballs. Note that in 2015, 66% of 3–0 pitches were fastballs and this raises to around 80% by 2019! To make it even worse, the ‘called strike’ row demonstrates that percentage of called strikes (strikes not swung at) for all 3–0 pitches over the seasons is above 50% and increased to over 55% by 2017 — so nearly 55% of al 3–0 pitches are uncontested called-strikes.

The probability row is the beta coefficients from a linear probability model. All coefficients are significant at the p < 0.001 level. The dependent variable is a dichotomous (two-category) variable which compares fastballs to all other pitches. The independent variable is a dichotomous variable comparing 3–0 counts to all other counts. In 2019, the probability of seeing a fastball on a 3–0 count compared to all other counts was 34% higher. So compare the above figure with the table — there’s a decent change the hitter is getting a fastball down the middle of the plate.

And yet, even with an increased likelihood of seeing a fastball on a 3–0 count, the likelihood of swinging remains low. However, swinging on 3–0 pitches has generally increased since 2015. I define swing, here, as foul balls, foul bunts, missed bunts, foul tip, hits (into play and out), and swinging strikes. Basically, I wanted to see if players were trying to induce contact on 3–0 counts. In every year since 2015, the percentage of 3–0 pitches that have been swung on has increased. In 2015, only 7.88% of 3–0 pitches had a swing event, this increases to 11% by 2019. Crucially, though, the probability of swinging is consistent and negative — effectively that swings are 37% less likely to occur in 3–0 counts compared to all other counts. That’s to be expected, to be fair, because (i) we know hitters don’t swing on 3–0 and (ii) hitters are (likely) more likely to swing when down in the count.

The final row demonstrates the number of home runs hit over each season on 3–0 counts. They’re pretty stable until 2018 when they increase to 46 and then fall to 40 in 2019. Below I’ve aggregated the most home runs by hitters. Here’s the top-10:

Maikel Franco: 5

Eugenio Suarez: 4

Russell Martin: 4

Starlin Castro: 4

Brandon Crawford: 3

Dexter Fowler: 3

Hanley Ramirez: 3

Josh Donaldson: 3

Justin Upton: 3

Here’s the top 5 (bottom 5?) home runs given up on 3–0 pitches by pitchers. Poor Jon Lester.

Jon Lester: 6

Mike Leake: 3

Jorge de la Rosa: 2

Cole Hamels: 2

Doug Fister: 2

Norms are strong and they prevail for a reason. They get baked into baseball by an old-guard who grew up respecting the game and breaking norms is really hard to do. But the data seems to point to an incremental increase in swinging on 3–0, to an increase in dingers, and an increase in fastballs. So put it all together: fastballs down the middle of the plate? Swing away, man. Let’s start some new norms.