Ah, the 2016 Most Valuable Player race, a contest in Major League Baseball’s junior circuit that seemed to further the widening gap between baseball’s average fan and its sabermetric community. It was ultimately won by Los Angeles Angels outfielder Mike Trout by a margin of 356-311 and 19-9, in voting points and first place votes, respectively. In both cases, his closest competitor was Boston Red Sox outfielder Mookie Betts. As in past seasons, the math-based sabermetric group of individuals mostly supported Mike Trout as its MVP, as posited by SoSH’s own David R. McCullough and Fangraphs’ Neil Weinberg.
Before we break into the findings of what StatCast provides us, it has been nearly six months since the 2016 American League Most Valuable player question was dominating the baseball headlines, so let’s take a refresher course to recount what the baseball landscape looked like following game 162, with respect to these two players.
By standard (non-StatCast adjusted) numbers, Trout cleaned up this heated dispute with the wet mop of an overworked janitor, destroying Betts in statistics such as Weighted Runs Above Average. In this respect, Trout’s incredible 56.2 wRAA lead the American League, while Betts’s 36.9 found him sixth in the junior circuit, sandwiched between Jose Altuve‘s 42.9 and Nelson Cruz‘s 35.8.
What about in terms of actual run production helped, or adjusting inning and score into our review? That’s where Run Expectancy based on the 24 base-out states enters, in which, as you might expect, Trout easily led the American League, at 74.22 runs. Betts, despite a very solid fifth place in the American League at 37.04 runs, accumulated fewer than half of Trout’s total. Since neither player ate at Chipotle to achieve these stats, this is not a positive development on Betts’s behalf.
One last question that may come to mind is what if we consider the inning and score into our view of the players? After all, both casual and diehard fans can agree: A grand slam that ties the score in the bottom of the ninth, with the bases loaded and two outs, is a much more valuable home run than one that a player hits either up or down 10 runs. This is where Win Probability Added tells the best story. Trout again smoked the competition, at 6.64%. For the sake of comparison, only two other players finished above 4.00% in 2016: Toronto third baseman Josh Donaldson’s 4.29%, and Boston designated hitter David Ortiz‘s 4.24%. Betts again found himself within the top 10, at 3.07%, ninth in the American League, but, once again, this number is less than half of Trout’s.
The counterargument of Betts supporters stems primarily from the records of the two teams, as Boston went 93-69 (98-64 by Pythagorean Record), whereas the Los Angeles Angels of Anaheim went 74-88 (80-82 by Pythag). As Rick Rowand postulates, the word “best” never appears within the loosely defined MVP ballots. Furthermore, as Rowand explains, the Angels would not be in a materially different position with a hypothetical replacement player instead of Trout. This argument is based on point one of the guidelines, which reads as follows: “Actual value of a player to his team, that is, strength of offense and defense.”
So, we now see what traditional stats, approaches, and interpretations of these statistics mean. It’s all great, yes? But what if we account for the freshest development in baseball, Statcast? It has the power of the world’s third most valuable brand supporting it in Amazon.com.
We no longer have to assume that all swings are created equal, or wonder just how fast a ball left the bat, whether it’s a weak grounder to the shortstop or a towering home run that lands on Lansdowne Street. Given the adequate sample of information baseball now has of StatCast data, this information can be worked into various statistics, as first described by Fangraphs’ Andrew Perpetua, later expanded on and improved by both him and others such as Mike Podherzer.
Unfortunately, as this data is still in its infancy, some of the more involved statistics, such as Weighted Runs Created+, are unavailable at present. But we do have some new statistics, such as VH% and PH%. VH% is percentage of plate appearances that result in a Value Hit (VH), or a near automatic extra-base hit. The latter statistic, PH%, is the percentage of plate appearances that result in a Poorly Hit Ball, or a near automatic out. Look no further than the example of Andre Either’s home run in game one of the 2016 NLCS. According to mlb.com’s Mike Petriello and his Statcast Hit Probability Calculation, this home run is an out 95% of the time, but went for a home run due to the wind and the dimensions of Wrigley Field.
Regarding xBABIP, a BABIP that has been adjusted to StatCast findings, I much prefer Mike Podherzer’s Podherzer’s version over Perpetua’s. Not only does his contain Perpetua’s seasonal constant, but it is also adjusted for type of batted ball and defense faced, especially useful for left-handed hitters with ground balls hit into the shift. As he explains in the aforementioned xBABIP link, while this metric is far from perfect, given the randomness of BABIP, an xBABIP Year 1 to xBABIP in Year 2 had a 0.509 year-over-year correlation, which is far better than regular old BABIP’s Y1 to Y2, with a year-over-year correlation of 0.274.
While it’s great to have all of these new metrics, it is easy to ask, “where is the meat in all of this?” Well, the whole hog is certainly here, and ready to be devoured as part of an all-you-can-eat statistical buffet, but, in my opinion, the best xStat available to view is xOBA+, which uses 100 for a league-average player, just like OPS+ (league-adjusted On-base Plus Slugging) and wRC+ do.
In 2016, Betts produced an xOBA+ of 112.0 (.353 xOBA, on a non 100-basis scale), or 12% better than the average hitter, below the likes of Kansas City’s Mike Moustakas at 114.1. Trout was well above Betts at 130.9 (.413 xOBA), second among qualified batters to league leader Miguel Cabrera (136.8). However, to simplify matters, and for a quick and dirty evaluation, here are the respective slash lines of these two players, with the slugging percentages broken down by type of hit.
In Trout’s case, we have a .308/.428/.571, with 97.3 xSingles, 31.2 xDoubles, 4.0 xTriples, and 35.4 xHomer Runs. This is easy enough to analyze, but what does it mean in context and difference in terms of what non-Statcast numbers read? Surprisingly, xStats dropped Trout’s batting average by seven points. However, and perhaps unsurprisingly, xStats raised his expected slugging percentage by 21 points, given Trout’s home park, and increased his StatCast expected home runs by 6.4. In the case of Betts, xSlash puts him at .309/.355/.489, with 136.8 xSingles, 45.0 xDoubles, 4.6 xTriples, and 23.1 xHomer Runs. Given the aforementioned factors at play, Betts suffered a stat decrease. His slugging percentage dropped 45 points, mostly from a change in expected home runs, with a drop of 7.9.
While Betts may have lost a considerable number of home runs, what about the times a ball drops in for a base hit? After all, we hear much about how a player has benefitted (or has been hindered) from BABIP luck. So why not convert BABIP into xStats form, using Podherzer’s regression-weighted equation, as discussed above? Fortunately, or perhaps unfortunately, neither Trout nor Betts is a left-handed (or even switch) hitter, which is where xBABIP shines by factoring in just how much shifts make a difference. In the case of Trout and Betts, even though xBABIP finds them both a bit lucky, it finds that their ability to put a ball in play for a hit quite comparable, as the difference between their xBABIP and traditional BABIP is only 10 points (49 for traditional, 39 for xBABIP). More specifically, and in actual numerical terms, Trout drops from an astronomical .371 to .346, a number more in line with his traditional BABIPs of .349 in 2014 and .344 in 2015. Betts drops at a slightly lesser rate, from .322 traditional BABIP to a respectable, and not overtly lucky, .307 with xStats. This lesser drop, by nature of comparison, should be expected, given a .322 BABIP is in the higher than normal range. xStats demonstrates that Trout is due for slight regression in 2017.
While many traditional metrics have been converted to xStats, what if we delve a bit deeper into some new stats, which we wouldn’t otherwise have were it not for StatCast, such as probabilities of batted balls with VH% and PH%. Thankfully, there are equations – VH% of probable balls=((VH%/(VH%+PH%))*100) (VHPB%) – which can be combined into a larger ratio to compare any two players. For Trout, his VH% and PH% are 9.4% and 12.1%, respectively. Meanwhile, Betts sports a VH% of 6.1%, with a PH% of 18.6%. This yields VHPB% values of 43.72 for Trout and 24.70 for Betts. In other words, considering only batted balls with a near automatic result of an xExtra Base Hit (xXBH) or an out, Trout was 1.77 times more likely to produce an xXBH than Betts.
By using a sabermetric statistical approach, anyway you slice and dice your meaty statistics, be they brisket, sausage, pulled pork, wRAA+, or even xBABIP, Trout is your winner by a wide margin. The question remains: Is this a one-year fluke, given how wide the gaps are between the two players, or is Trout simply that much better than other top players of his generation, year in, year out? Time will tell, but it appears he is. But, as Pepsi told us: This is the face of a new generation for many years to come, and we are lucky to see such a talent in our lifetimes.