The First Pitch Swing Decision: Selectivity Versus Passivity

A couple of weeks ago, I wrote an article titled “The Myth of the Passive Hitter“, which examined the change in the rate of swings in Major League Baseball over the last 25 years. While a host of writers have written about the rise of Moneyball philosophies as a reason for why strikeout numbers are on the rise, I noted that the data shows that hitters are actually swinging at basically the same rate now as they did two decades ago. The evidence simply doesn’t support the idea that hitters have become significantly more passive at the plate.

However, as several readers pointed out, the point being posed most repeatedly by Tom Verducci isn’t that hitters are not swinging often enough, it’s that they’re not swinging often enough in the right situations. As a follow-up to his original post, he wrote this last week:

1. Hitting in the major leagues is fundamentally broken

What will it take for teams to start admitting that this passive-aggressive, run-up-the-pitch-count philosophy isn’t working? Apparently almost a decade of declining results isn’t enough. Entering this week:

• The number of hits per game is down for the seventh straight year.
• On base percentage has been stagnant or down for the seventh straight year.
• Strikeouts are up for an eighth straight year.
• Batting average has sunk to an all-time low (.253) since the DH was instituted 40 years ago.
• Runs per game is tied with the rate from 2011 for the lowest rate since 1992.
When you go to a baseball game today you will see fewer hits on average than at any time since 1972 — and yet the game is taking more than 20 minutes longer to play. That’s more than 20 minutes of added dead time without the ball put in play.

And yet you see hitter after hitter giving pitchers strike one and taking 2-0 and 3-1 fastballs. It’s like a major corporation with seven straight years of operating at a financial loss insisting that nothing is wrong.

Let’s set aside all the correlation/causation issues there for a second and note that the data we looked at two weeks ago does support Verducci’s claim that hitters are giving pitchers more called strikes on the first pitch of an at-bat. As he noted, the first pitch swing percentage has declined from 33% in 1988 to 26% last year, so the rate of 0-1 couts is significantly higher now than it was then. For reference, here is the percentage breakdown of how first pitches went in 1988, and how they went in 2012.

First Pitch	1988	2012
In Play	15%	11%
Ball	43%	40%
Strike	42%	49%

The first pitch of each at-bat is resulting in fewer called balls and fewer balls in play, and those decreases are being transferred over to strikes, which means that pitchers are pitching in 0-1 counts a lot more than they used to. Verducci is not wrong about this claim; the willingness of hitters to take the first pitch has led to an increase of called strikes, putting the hitter into a disadvantageous count more often. However, as we noted two weeks ago, the overall swing percentage has remained basically stagnant, so because hitters are swinging at the same rate overall while swinging at fewer first pitches, they must be swinging more often at some other point in order to make the numbers offset.

To help illustrate the actual changes in swing rates by count, I enlisted the help of some of my fantastic colleagues — seriously, Bill Petti, Matthew Carruth, Steve Staude, and Jeff Zimmerman are the best — and we pulled data from retrosheet and looked at how often batters have swung versus taking a called ball or called strike on each count over the last 25 years. Because Bill was nice enough to load all this into Tableau, we can present all of the charts in one image, so you can simply click from one count to the next to see the yearly changes in swing rates for each situation.

You can click on swing rate in the legend on the side to just highlight the orange line on each chart, if you want to focus solely on how often hitters are swinging and put the called ball/called strike lines into the background. As you go through the counts, you’ll probably notice some patterns that support Verducci’s arguments; swing rates are down significantly in every count where the batter has yet to take strike one. It’s most pronounced in 0-0 and 3-0, but 1-0 and 2-0 counts have also seen a reduction in swing rates over the last 25 years.

Meanwhile, the counts that have seen the biggest increase in swing rate are 0-1 and 0-2, as both are at or near 25 year highs as hitters chase pitches more often when they are behind in the count than they used to. Interestingly, this is not an across the board “two strike” problem, as the swing rates in 1-2, 2-2, and 3-2 counts haven’t actually changed all that much. Most of the changes that have been observed can be explained by hitters taking more strike ones in all counts where they have not yet taken strike one, and then swinging more in 0-1 and 0-2 counts.

If seeing the raw data makes these changes difficult to observe, here is the same chart, just with the numbers scaled to the league average for each count over the last 25 years, so the average for the entire time period for each line is 100. This way, you can see the overlap of the lines on top of each other, and note where the big shifts take place.

The data is the same, but putting it on the 100 scale makes it easier to see things like the significant fluctuation in 3-0 swing rate over time. After peaking at 12.3% in 1996, hitters swung at just 8.0% of 3-0 pitches last year. While first pitch swing rates are down, they are simply part of the bigger story – hitters are much more willing to take strike one in any count, and this may very well be the result of more teams emphasizing working counts and getting on base. The notion that hitters have changed their swing patterns over time is correct, even though overall swing rates have remained stable.

So, is Verducci right after all? If hitters are swinging at fewer pitches in counts where they have historically done well and are swinging at more pitches in counts where they have done poorly, couldn’t this account for a large part of the decline in offensive levels, as Verducci claims? Well, it’s not quite that simple.

Let’s introduce a few more facts into evidence. We already listed the change in ball/strike/in-play rates on first pitches between 1988 and 2012, so now let’s look at the actual outcome differences for those same years.

First Pitch	1988	2012
BA	0.314	0.333
OBP	0.316	0.340
SLG	0.467	0.545
OPS+	112	122

Hitters are swinging less often at the first pitch, but when they do swing, their production levels are significantly higher than they used to be. Taken together, this tells us that hitters are now less willing to swing at the kinds of 0-0 pitches that are not as likely to turn into a positive outcome. Hitters aren’t just taking first pitch meatballs down the middle, they’re taking the kinds of first pitches that used to result in outs and turning those pitches into strike one instead.

Remember how we noted that the first pitch in play result was down from 15% in 1988 to 11% in 2012? Well, given the number of hitters who came to the plate last year, that decrease resulted in 7,262 fewer balls being put in play on 0-0 counts. In order to get the league OPS+ down from 122 to 112 with 7,262 extra plate appearances, those balls in play would have had to result in an OPS+ of 86. In other words, that’s the production level that hitters have chosen to give up by not swinging at the types of first pitches they’ve chosen not to swing at.

This isn’t passive hitting; this is selective hitting. Not every first pitch is created equal, and you cannot simply look at the total rate of outcomes on first pitches swung at and assume that the success level would be equal for each additional swing on first pitches. This would be analogous to telling a player who stole 10 bases at a 90% success rate that he should steal 50 bases because he doesn’t get thrown out very often. His success rate is heavily influenced by the decision of when to steal or not, and running more often would result in a lower success rate on future attempts, since those extra attempts would have to come in the situations where he has already deemed himself less likely to succeed.

The same is true of first pitch swings. You can look at how hitters do when they swing on the first pitch and note that the outcomes on first pitch swings is very good, but you have to note that the outcome rates are that good because hitters are only swinging at the most hittable pitches. The pitches they are not swinging at any more are pitches they didn’t used to perform very well on, as you can see from the change in OPS+ on 0-0 counts from 1988 to 2012.

Year	PA	BA	OBP	SLG	OPS	OPS+
1988	23463	0.314	0.316	0.467	0.783	112
1989	22743	0.311	0.313	0.464	0.778	111
1990	21895	0.312	0.314	0.468	0.782	110
1991	21888	0.314	0.317	0.472	0.789	111
1992	22294	0.303	0.306	0.448	0.754	107
1993	23741	0.322	0.325	0.489	0.814	110
1994	16463	0.332	0.335	0.525	0.860	112
1995	20382	0.331	0.335	0.515	0.850	112
1996	21700	0.340	0.344	0.539	0.883	115
1997	22009	0.331	0.336	0.519	0.855	113
1998	24481	0.335	0.340	0.524	0.864	114
1999	24820	0.338	0.342	0.537	0.879	112
2000	24648	0.338	0.342	0.545	0.887	113
2001	24832	0.341	0.346	0.553	0.900	118
2002	24448	0.336	0.343	0.544	0.887	118
2003	23630	0.336	0.344	0.545	0.889	117
2004	23260	0.340	0.347	0.553	0.900	117
2005	23245	0.331	0.337	0.536	0.872	116
2006	23018	0.341	0.346	0.553	0.899	117
2007	22464	0.344	0.349	0.551	0.901	118
2008	22024	0.337	0.342	0.546	0.888	118
2009	20934	0.339	0.345	0.558	0.903	120
2010	20520	0.334	0.340	0.534	0.874	120
2011	20739	0.330	0.336	0.523	0.859	119
2012	20281	0.333	0.340	0.545	0.885	122

So, now, things are a little more complex than just “fewer first pitch swings = lower offensive levels”. This data shows that hitters are swinging at fewer first pitches that are tough to hit, so they’re trading swings that would result in a lot of outs for more 0-1 counts. Is that a good trade-off? Well, to answer that question, we need to look at how hitters do after the count goes 0-1. So, let’s look at the same table we just presented for 0-0 counts, only for all at-bats after a count goes 0-1. This is not just at-bats that end with an 0-1 count, but for everything that occurs after a hitter falls behind a batter with a first pitch strike.

Year	PA	BA	OBP	SLG	OPS	OPS+
1988	66526	0.223	0.261	0.321	0.582	84
1989	65698	0.225	0.263	0.324	0.586	84
1990	61439	0.226	0.265	0.329	0.594	84
1991	67759	0.225	0.265	0.329	0.594	84
1992	68264	0.230	0.270	0.330	0.599	86
1993	71787	0.233	0.274	0.346	0.619	84
1994	50684	0.236	0.278	0.356	0.634	83
1995	62405	0.233	0.278	0.352	0.630	83
1996	68824	0.236	0.281	0.361	0.642	84
1997	70863	0.232	0.275	0.354	0.629	83
1998	81536	0.230	0.274	0.352	0.626	83
1999	81226	0.238	0.283	0.367	0.650	84
2000	82476	0.237	0.283	0.372	0.655	84
2001	85369	0.231	0.276	0.364	0.640	84
2002	84586	0.230	0.274	0.355	0.628	84
2003	86452	0.235	0.278	0.364	0.641	85
2004	86979	0.237	0.279	0.368	0.647	85
2005	87034	0.237	0.277	0.364	0.642	86
2006	87728	0.241	0.283	0.374	0.657	86
2007	88494	0.238	0.281	0.361	0.643	85
2008	88073	0.234	0.278	0.357	0.635	85
2009	88182	0.232	0.275	0.354	0.629	84
2010	88651	0.227	0.269	0.346	0.615	85
2011	89320	0.225	0.265	0.341	0.606	84
2012	90061	0.226	0.265	0.346	0.612	85

Well, look at that. We noted that hitters have basically skipped out on 7,000+ first pitch swings that would have had to result in an 86 OPS+ in order to get the 0-0 OPS+ to equal what it was in 1988, and the penalty for not swinging at those pitches and falling behind 0-1 has been to post an 85 OPS+ after they fall behind with strike one. In other words, if every first pitch take that used to be a swing was called a strike, then hitters would have been making an almost perfectly rational trade-off between swinging and not swinging at the first pitch. Their overall performance level has not been dragged down by moving swings out of marginal 0-0 pitches and into 0-1 pitches. The approach is different, certainly, but not less productive overall.

And that doesn’t account for the fact that the called strike rate on these now non-swings shouldn’t be 100%. It might be pretty high — it’s impossible to know without strike zone location data, which we don’t have for 1988, of course — but it shouldn’t be a perfect relationship between not swinging and taking a called strike. When you factor in that some of those non-swings should lead to 1-0 counts — and as you’d expect, hitters do very well after the count goes 1-0 — and not 0-1 counts, the data suggests that hitters are making a good choice to take more marginal first pitches and look for better pitches to hit later in the at-bat.

Now, this is just an observation about first pitch swing rates. It could very well be that Verducci is correct about the current rate of swings on 1-0, 2-0, and 3-0, and perhaps hitters are not taking advantage of those situations as well as they could have.

Given that this post is already 2,000 words, though, I’ll look at those in more depth in follow-up pieces, as well as using the zone data we do have to see if it can help us understand why some of these shifts have occurred. For now, we’ll just leave it at this: hitters are swinging less on the first pitch, but the rise of first pitch selectivity has led to an increase in production on first pitch swings, and there is not significant evidence from the after 0-1 count data to support the idea that the increase in first pitch strikes are the reason for the offensive downturn in Major League Baseball.