On the Efficacy of Hard Hits to the Opposite Field

FanGraphs’ leaderboards are like iron ore mines. They abound with potentially valuable commodities, but sometimes it takes effort to unlock the potential and extract the value.

There’s only so much you can write about a certain subsection of players (National League outfielders) without beating a dead horse. Alas, I’ve tried to find peculiar reasons to write about particularly interesting hitters. And, ah, FanGraphs’ leaderboards, they’re helpful in this regard, especially when broken down by splits.

Baseball Info Solutions’ (BIS) opposite-field batted ball data are no exception, as exemplified by this table of opposite-field hard-hit rate (Hard%) leaders among NL outfielders:

2015 NL OF Oppo Leaders

Name	Oppo Hard%	MLB Percentile	Oppo%	NL OF Percentile	MLB Percentile
A.J. Pollock	41.40%	99	25.20%	54	50
Ryan Braun	39.50%	98	31.80%	84	90
Christian Yelich	39.00%	97	27.70%	67	70
Gerardo Parra	35.10%	93	27.30%	62	70
Juan Lagares	34.70%	92	29.90%	78	84
Matt Kemp	34.50%	91	21.90%	25	24
Bryce Harper	34.30%	89	20.20%	10	12
Marcell Ozuna	34.20%	89	29.00%	73	79
Yasmany Tomas	32.90%	88	29.30%	75	81
David Peralta	32.90%	87	22.30%	33	27
Chris Coghlan	32.50%	87	25.30%	56	51

Oppo Hard%: min. 50 PA
Oppo%: min. 200 PA

“An interesting list,” one might say, probably not aloud or, if aloud, preferably in the privacy of his or her own home. Indeed, Harper and Pollock broke out in big ways; Braun bounced back from a forgettable 2014 season; and Parra and Coghlan have put together career-best offensive seasons (although Coghlan’s BABIP is bringing him down). Plus a quarter of the list is Diamondbacks.

The peculiarity of almost every name on the list begs the, or at least my, question: how important, if at all, is the ability to the opposite field with authority? I calculated how well opposite-field hard-hit rate correlates with several offensive metrics, from the simple to the linearly weighted.

Hard% Pearson R Correlations

	ISO	BABIP	wOBA	wRC+	WAR/600
Just Oppo	.3982	.4992	.4824	.5050	.3212
All BIP	.1536	.5653	.3466	.3534	.2104

-1 = perfect negative correlation; 0 = no correlation; 1 = perfect positive correlation
Based on all qualified hitters, 2002-14

Just Oppo calculates hard hits as a percentage of strictly opposite-field hits; All BIP calculates opposite-field hard hits as a percentage of all balls in play. A higher percentage of hard among all opposite-field batted balls correlates at least moderately well with every metric. But it’s important to consider how often a player hits oppo in the first place. Harper and his pull-happy ways can attest to that.

Even when considered relative to all balls in plays, hitting with authority to the opposite field exhibits a moderate positive relationship with strictly-offensive composite metrics (wOBA, wRC+) and a strong positive positive relationship with batting average on balls in play (BABIP). In other words, more high-quality contact toward the opposite field typically leads to better offensive outcomes (but not necessarily more power).

All right. We originally mined iron ore; now we’ve made steel. It’s useful information. But steel on its own is just a pile of steel. Let’s build something with it.

Each name on the list bears its own story. For many of them, their hard-hit prominence is surprising — or maybe not surprising but, instead, validations of unexpected levels of performance. (The key word here is levels of performance, as obviously Harper has always been capable of inhuman feats; Peralta was on some preseason fantasy radars; Coghlan did have a pretty solid 2014 season; and so on.)

It’s worth trying to assess, then, how sustainable these kinds of performances are. I ran a handful of single-variable linear regressions in an attempt to anticipate 2016 hard-hit rates for opposite-field balls in play (Hard% Oppo) — that is, regardless of how often the batter hits oppo — as well as all opposite-field hard-hit balls relative to overall frequency (Oppo-Hard% BIP).

Changes in Hard%

	Just Oppo				Oppo Relative to All BIP
	2014	2015	2016 N	2016 C	2014	2015	2016 N	2016 C
Adj. R-squared			0.494	0.215			0.484	0.181
A.J. Pollock	29.30%	41.40%	35.63%	36.69%	5.45%	10.43%	9.11%	10.47%
Ryan Braun	33.10%	39.50%	34.27%	37.28%	10.26%	12.56%	10.64%	12.71%
Christian Yelich	44.00%	39.00%	33.92%	41.78%	13.77%	10.80%	9.38%	11.18%
Gerardo Parra	26.10%	35.10%	31.13%	31.75%	7.10%	9.58%	8.50%	9.73%
Juan Lagares	25.80%	34.70%	30.85%	31.39%	7.40%	10.38%	9.07%	10.50%
Matt Kemp	28.90%	34.50%	30.71%	32.64%	5.95%	7.56%	7.04%	7.74%
Bryce Harper	20.00%	34.30%	30.56%	28.62%	5.16%	6.93%	6.59%	7.10%
Marcell Ozuna	32.40%	34.20%	30.49%	34.00%	8.62%	9.92%	8.74%	10.11%
Yasmany Tomas	—	32.90%	29.56%	?	—	9.64%	8.54%	?
David Peralta	13.30%	32.90%	29.56%	24.90%	2.22%	7.34%	6.89%	7.37%
Chris Coghlan	21.30%	32.50%	29.28%	28.18%	5.18%	8.22%	7.52%	8.34%

2016 N: 421 observations (all qualified hitters, 2010-14)
2016 C: 220 observations (all qualified hitters, 2010-14)

2016 N specifies the current hard-hit rates as dependent variables and the prior-year hard-hit rates as independent variables. 2016 C specifies the dependent variable as the nominal change between the current and prior seasons (that is, 2015 Hard% minus 2014 Hard%). The first non-header row, highlighted in blue, denotes the adjusted R-squared statistic (the square of the Pearson R coefficients from earlier) for each regression, which essentially indicates how well the model explains trends in the data. The models aren’t meant to be especially complex or rigorous. The data would certainly benefit from more observations by including more years and loosening the restriction on minimum plate appearances.

2016 N exhibits a higher adjusted R-squared than 2016 C, but N‘s model specification incorporates bias that inflates the fit of the model, whereas C’s model specification tries to control for it. They inherently measure different things, though. The latter investigates how well year-over-year changes correlate with each other; its low R-squared indicates negligibly small relationships among annual changes in hard-hit rates, implying that hitters who achieve significant gains (or experience equally significant declines) are not necessarily “due to regress,” as one might say.

However, N assumes all hitters will negatively regress no matter what — a fundamentally flawed assumption. Thus, I would not use either equation to try to project hard-hit rates going forward. (They do, however, predict largely similar values, although they differ on more extreme observations.) But they do help us understand an important notion: hard-hit rates do not randomly fluctuate from year to year.

The possibility exists that each hitter’s performance betrays his actual tools or skills (“peripherals”). But in looking at inputs (i.e. hard-hit balls) in lieu of outcomes (actual hits), we largely circumvent matters of luck.

Alas, the point still stands: the NL outfielders highlighted in this piece should sustain their opposite-field authority to a certain degree given their mechanics, approach, everything remain unchanged. Unfortunately, that’s not how baseball works; athletes are human beings who constantly change — mechanically and, perhaps, mentally. It’s this uncertainty that makes each MLB season, and fantasy baseball season, a unique one.

Moreover, it’s worth reiterating that opposite-field authority does not completely explain BABIP or wRC+ or wOBA. Thus, a hitter can maintain high levels of excellence when hitting to the opposite field but see a dramatic change in his production should the rest of his offensive profile change.

Anyway, I don’t know what this accomplishes precisely. I hadn’t really known the quantified (albeit simplified) impact of authoritative opposite-field hitting prior to this, but maybe you did, so who knows. At least I can confirm my high school coaches weren’t full of hot air.