In the second inning of Sunday’s game against the Yankees, Randal Grichuk poked a 1-2 fastball from Gerrit Cole the other way though the hole for a ground ball single. It was a very nice bit of two strike hitting against an ace pitcher, and prompted Buck Martinez to laud Grichuk for his improved two strike implemented last year to cut down.
Indeed, while it was inherently a small sample, in 2020 Grichuk posted by the best strikeout rate of his career at 21% in 231 plate appearances. That was 5% better than any other season of his career (and almost 10% better than early in his career), a pretty staggering degree of improvement.
If this flowed out of a much more effective two strike approach in the manner Martinez suggested, it should show up pretty clearly statistically in Grichuk’s contact rate, and specifically the differential between what happens before two strikes and with two strikes. So I pulled the pitch level data from Baseball Savant to see if the facts back up Buck’s narrative.
In 2018-19, his first two seasons in Toronto when he struck out about 26% of the time both years, Grichuk made contact 70.5% of the time he swung with two strikes, compared to 70.6% of the time before two strikes (or conversely, he whiffed about 30% of the time). That was much worse than league average in the mid-to-high-70s, but the differential roughly mirrored the league trend as overall batters whiff about 1% more with two strikes than otherwise.
In 2020, Grichuk’s two strike rate did improve, by about three percentage points to 73.7%, seemingly validating Buck’s notion of a better two strike approach. However, his contact rate otherwise improved by much more than that, by more than 8% all the way up to 79% — meaning Grichuk was about 5% worse with two strikes at making contact. To the extent he had a different two strike approach, one could argue it was actually less effective.
A slightly different breakdown illustrates what was actually going on. In 2018-19, Grichuk made contact about 64.4% of the time when he was behind in the count, compared to 73.4%, for a 9% differential. It shouldn’t be surprising that batters swing and miss more often when behind, but league wide the differential is less extreme at about 6.5%.
In 2020, Grichuk improved his contact rate in the most advantageous pitch counts (0-2, 1-2) by less than 3% whereas in the most advantageous hitter counts it improved by more than 8%. In other words, in a deep hole he was still largely mincemeat, but more contact earlier in the count helped limit those situations in the first place. That’s what really made the difference, not cutting down to make more contact when already behind.
A few weeks ago, I argued that Cavan Biggio needed to alter his two strike approach and swing more often with two strikes because while his ultra-selective approach is fine before two strikes, with two strikes it translates to excess called strikeouts. Monday night’s televised broadcast against Detroit provided an almost textbook example.
After striking out swinging in his first two plate appearances, Biggio came to the plate in the 6th inning after the Tigers brought in Ian Krol to face him in a lefty-lefty matchup. After falling behind 0-2, Biggio took the fifth pitch, a fastball that was called strike three:
It was the definition of a borderline pitch, perhaps just clipping the edge of the plate. With less than two strikes it’s a great take because the hitters on average can’t do much with that, there’s a reasonable chance of it being called a ball, and a called strike will put the hitter in a worse situation but the at-bat is still alive. Those first two things are still true was two strikes, but the third is not and that radically changes the calculus.
To illustrate, let’s assume for simplicity there’s a 50/50 chance of that pitch being called a strike, and the probability doesn’t change with the count. Let’s look at if the count is 0-1. In 2020, batters had a 87 wRC+ for all PA where they got to a 1-1 count, falling to 38 if it got to 1-2, and increasing to 125 if it got to 2-1. Averaging 38 and 125 is 81.5, so taking the pitch slightly reduces expected production from the initial 87.
If he swings, he can either miss (~23% of all 1-1 swings), foul (37%) or put the ball in play (40%). Missing or fouling (total 60%) results in the same 1-2 count. What about if it’s put in play? Lefties swinging at pitches in that low and away zone hit about .230 last year. I wish that chart had wOBA/total production, but if we assume it’s quite low power it’s something like 40 wRC+. The overall expected value then is 29 wRC+.
Clearly then, taking the pitch in the 1-1 count is the better option. It still has negative expected value, but that’s of baseball and the pitcher having the initiative and making his pitch.
Now let’s assume the starting count is 2-2. Taking the pitch means a 50% chance of a strikeout, and 50% chance of a 3-2 count. That works out to an expected wRC+ of 10. Swinging, there are now three possible end states since fouling and missing have different outcomes. That weighted average is 18, which is now above the expectation from taking the pitch (albeit modestly).
What is optimal earlier in the count is not optimal with two strikes, and the above is an example of a pitch a batter has to protect with two strikes.