The Contact Score Multiplier

Many of my recent articles in this space have centered upon assessment of batted ball quality for hitters. In this day of StatCast and Hit f(x), discussion of such information has intensified in the public realm, and with it has come much misunderstanding. There is a whole lot more to batted ball quality than authority itself. The Mariners hit the ball much harder than the Cardinals, but aren’t nearly their equal as an offensive ballclub, for instance. Today, let’s examine the relationship between hitters’ contact score and their OPS+, based on their K and BB rates relative to the league.

I’ve used the term “contact score” quite a bit here in the recent past. Basically, it measures a hitter’s production relative to the league on balls in play, both before and after adjustment for context, with the Ks and BBs stripped from his record. To illustrate, the AL as a whole had a .316 OBP and .390 SLG in 2014; after the K and BB are eliminated, the league had a .319 OBP and .491 SLG on balls actually put in play. Among AL regulars, Jose Abreu posted the highest OPS+ at 169, but ranked third behind J.D. Martinez (210) and Mike Trout (204) in raw, unadjusted contact score at 197.

The main driver behind a player’s raw contact score is in fact his batted ball authority, in particular his fly ball authority. Other key drivers include popup and line drive frequency, the presence or absence of an extreme pull tendency on the ground, as well as contextual factors such as home ballpark, and, of course, luck. In a perfect world, we’d all have access to granular batted ball data going back to the beginning of the modern era to adjust contextually, but alas, we don’t. Instead, we will have to be satisfied with correlating raw, unadjusted contact score to OPS+, and creating a set of multipliers between the two based on players’ approximate K and BB rates relative to the league.

Going back to 2008, I looked at a population of 980 player seasons (# of teams in league * 9 regulars in the AL, * 8 in the NL) with the most plate appearances. Why 2008? Quite honestly, one of the factors was that Barry Bonds (and others) start to creep into the sample at that point, and those guys simply were playing a different game than is being played today. Plus, Bonds’ presence tends to throw analyses off kilter all by itself.

Within each season and each league, I split the player-seasons into K and BB rate “buckets”; on both of those two axes, the seasons were stratified into layers as illustrated in the two tables below, one each for the AL and NL:

AL	K > 2	> 1	> 0.5	AVG	< (0.5)	< (1.0)	< (2.0)
BB > 2	72.1	85.7	100.2	107.2	116.4	—	—
> 1	71.1	82.9	93.3	105.1	110.1	124.5	—
> 0.5	—	82.4	92.0	99.0	112.1	122.1	—
AVG	66.6	78.5	88.6	99.4	107.2	116.9	—
< (0.5)	—	77.6	87.3	95.6	103.4	111.8	—
< (1.0)	—	74.7	83.8	92.1	99.7	109.5	—
< (2.0)	—	—	—	—	—	—	—
ALL	69.5	80.2	90.4	98.8	106.3	114.8	122.3

NL	K > 2	> 1	> 0.5	AVG	< (0.5)	< (1.0)	< (2.0)
BB > 2	—	83.0	—	97.1	113.7	112.8	—
> 1	—	78.3	85.4	95.5	107.6	116.5	—
> 0.5	—	74.5	85.4	93.5	106.1	114.1	—
AVG	61.8	73.7	83.9	91.6	100.7	111.0	122.3
< (0.5)	—	70.8	79.5	88.1	97.0	106.3	117.6
< (1.0)	—	—	76.3	85.6	92.8	101.2	116.6
< (2.0)	—	—	—	—	—	—	—
ALL	62.7	75.6	82.9	91.9	101.7	108.6	119.5

Across the top of both tables are the K rate “buckets”, > 2.0 standard deviations above league average, between 1.0 and 2.0 above, between 0.5 and 1.0, average range, between (0.5) and (1.0) below, between (1.0) and (2.0) below, and < (2.0) below. The same buckets go down the left side of each table with respect to relative BB rate. The individual cells of each table contain the Contact Score/OPS+ multiplier for that "bucket" of players. The blank cells indicate buckets that include five or few players from 2008 to 2014 combined. One might ask why we need separate tables for the AL and NL. That is attributable to the presence of the DH in the AL. A comparison to league average includes all batted balls, and there is an awful lot more empty contact included in the NL numbers. Alternatively, I could have backed out pitchers' hitting out of both leagues' totals to bring the multipliers closer together, but for convenience's sake, did it in this manner. Let's focus on the AL table to see how it works. If a player's K and BB rates were both in the average range (as were 125 of the 900 AL regulars over that span), you would on average, have to multiply that player's raw contact score by .994 to obtain his OPS+. The process obviously works in reverse if you wish to work backward from his OPS+ to his contact score. That middle cell isn't where the fun stuff resides, however. You will notice how the multipliers increase fairly dramatically from left to right, as the K rate decreases. If an AL player's BB rate was in the average range, but his K rate was between 1 and 2 standard deviations below average (as were 60 of 900 AL regulars over that span), the multiplier mushrooms to 116.9. On the other hand, an AL player with a BB rate in the average range but a K rate between 1 and 2 standard deviations above average (33 players met that criteria), has an average multiplier of 78.5. What does that mean? Well, a player with a league average contact score of 100 would then typically have an OPS+ of 117 with the lower K rate, and a 78 OPS+ with the higher K rate. That's a difference between a potentially well above average offensive player, and in some cases, replacement level. You will also notice by scanning the tables vertically that BB rate also has a measureable impact upon the multipliers, though not nearly of the same magnitude as K rate. To get a better feel for this concept, let's look at three AL players who had the same exact raw contact score in 2014; Ben Zobrist, Ian Kinsler and Kelly Johnson, all of whom posted contact scores of 92, and played their home games in non-extreme ballparks. Zobrist, the most productive of the three, had a K rate over a standard deviation lower than the average of AL regulars, and a BB rate over a standard deviation above. That gives him a very favorable multiplier of 1.245, which projects an OPS+ of 115, right in line with his actual mark of 116.

Kinsler’s K and BB rates were both over a standard deviation lower than the AL average, for a 1.095 multiplier and a projected OPS+ of 101, right in line with his actual 100 mark. Johnson’s K and BB rates were both over a half standard deviation above average, for a .92 multiplier and a projected OPS+ of 85, again in line with his actual 86 mark.

In the real world, what does this all mean? Well, at this stage in the game, these three particular players have clearly passed their physical peak, and their contact scores aren’t likely to trend upward. Kinsler, for example, has seen his raw contact score bounce around in a fairly narrow range between 84 and 99 since 2009. The absolute only thing separating him from offensive ruin is his miniscule K rate. Since 2011, his walk rate has plunged by two-thirds, from over a standard deviation above average to over one below, cutting his multiplier from 1.245 to 1.095, and predictably trimming his OPS+ by 17 basis points, from 117 to 100.

Such an analysis can also give us a glimpse into sudden player decline. I wrote about Robinson Cano’s terrible first half quite recently for ESPN Insider. While his batted ball quality, both from an authority and frequency perspective, has deteriorated somewhat in 2015, the breakdown of both his K and BB rates has been the primary driver of his struggles. In 2014, his K rate was over a full standard deviation below the AL average, and his BB rate over one half standard deviation above, for a 1.221 multiplier. In 2015, his K rate is in the average range, and his BB rate over a half standard deviation below, for a .956 multiplier. That’s a massive one-year shift that is very difficult to overcome with batted ball authority.

Utilizing this concept can also help us understand the calculus faced by supposedly underachieving all-or-nothing power hitters such as Mark Trumbo. When Trumbo hits the ball, he hits it hard and far. Still, with the exception of 2012, when he posted a 154 contact score, his contact scores have merely been in the low 100’s, ranging from 113 to 129. He’s got the contact authority down, but not the contact quality; his popup rates are consistently high, but his liner rates are low, as are his fly ball rates, at least for a power hitter.

Most importantly, however, Trumbo boxes himself into a corner with his subpar K and BB rates for a power hitter. From 2011-14, his respective contact score multipliers were 83.8, 77.6, 78.5 and 83.9. If you’re Giancarlo Stanton, your raw authority could overwhelm such low multipliers. How’s a guy like Trumbo, with only slightly above average contact scores and such poor multipliers, not to mention an utter lack of complementary skills, actually going to provide real value to a club?

In a perfect world, we’d all have access to the detailed BIP data that would enable us to make the contextual adjustments necessary to caulk the holes in this type of analysis. The data is available to the clubs, in conjunction with video and traditional scouting techniques, the forward thinkers are using it to tweak player performance. The out is the game’s currency, and any adjustment that can move K and BB, or even liner and popup rates in the right direction, can make a difference.

Without contextual adjustment, Rockies’ hitters are going to defy these multipliers on one end, and Mariner bats will do the same on the other. Still, this approach does enable to isolate and quantify the respective impacts of batted ball quality and K and BB rates on overall hitter performance. It measures the size of a player’s safety net, if you will, which allows the Matt Hollidays to age differently than, say, the Jose Lopezes.