Creating a statistic to look at a player’s raw ability at the plate.
If you ask the average baseball fan to determine the skill level of any given hitter by the stats on their Baseball-Reference page, one of the first things they’ll look at are the player’s slash stats. Those are, if you’re unfamiliar, AVG/OBP/SLG (and sometimes /OPS). For example, in 2016, American League Most Valuable Player Mike Trout “slashed” .315/.441/.550. This means Trout had a batting average of .315, an on-base percentage of .441, and a slugging percentage of .550. Simple enough, right?
These three numbers, to those who aren’t into advance stats, are a pretty decent representation of how well a player performed in the given time frame. It shows how often a player got a hit, how often they reached base, and how much power they hit with.
While looking at a slash line gives a good idea of hitting performance, it doesn’t give a great one. Why not?
First, there is no ballpark adjustment. For example, last season, teams slashed .296/.360/.492 at Coors Field, and .241/.309/.394 at Nationals Park.
This is because, well, the Coors effect. It wasn’t because Rockies hitters were much better than Nationals hitters, as Rockies hitters had a .700 OPS on the road, compared to the Nationals’ mark of .754.
At Coors, balls fly a lot further than they do at other stadiums, as a result of the air in the mountains. Not all ballparks are created equal. Some, like Coors, are known as hitters’ ballparks, while others are known for being pitcher friendly. It doesn’t seem fair to evaluate and compare players without this in mind.
While DJ LeMahieu (.348) barely won the batting title over Daniel Murphy (.347), it’s more impressive for Murphy given he doesn’t play half of his games at the most hitter friendly park in the MLB in Coors Field.
In addition to ballpark factors, another thing slash stats don’t take into account is speed. Yes, being a faster player is definitely a good thing, and it should be taken into account when evaluating a player. However, if you’re interested in solely judging a player by their abilities at the plate, slash stats wouldn’t be much help.
Similarly, Fangraphs’ wOBA won’t be of much help either, and neither will most advanced stats.
Why? wOBA, for example, looks at a player’s commitments offensively using walks, hit by pitches, singles, doubles, triples, and home runs. Yes, this gives a very good estimate of how much a player committed to their team’s offense; however, it doesn’t tell you how good of a hitter they were.
Why not? Let’s say, for this article’s sake, David Ortiz and Billy Hamilton have the exact same skill set at the plate, but Hamilton still has a huge speed advantage. Who will have a better wOBA? Hamilton will, because his speed will turn some groundouts into singles, some singles into doubles, etc.
Likewise, Ortiz will struggle to get two bases on something Hamilton would easily be able to get three on. Therefore, Hamilton would have a much better wOBA, because the more bases you get, the better wOBA you’ll have.
However, like mentioned earlier, both players have the exact same ability with the bat. A stat which only measures skill at the plate should tell you Billy Hamilton and David Ortiz are equal. Again, the three slash stats will point in favor of Hamilton, as his speed will get him on base more and will get him more bases.
Lastly, slash stats don’t factor in luck. Luck is very difficult to measure, because there are so many factors which go into it: where the defense is playing, how efficient they are, where the ball was hit, etc.
Yes, slash stats tell you how a player performed, but what other use do they have? If you want to predict how a player will perform in the future, you shouldn’t be looking at their slash stats.
Why? Luck isn’t a skill. That’s a little confusing, but let me explain. If a player gets extremely lucky in their rookie season; let’s say a lot of their hits were bloopers, the defense wasn’t playing them very effectively, etc., what makes you think those trends will carry over in their second season? Players don’t have the ability to be lucky on a consistent basis; if they did, it wouldn’t be luck.
So yeah, slash stats don’t give you a very good understanding of how good of a hitter a given player is. In order to do that, one must only look at outcomes a hitter is mostly in control of.
Stats like FIP for pitchers do an okay job; the only things that factor into FIP are strikeouts, walks, and home runs. A single doesn’t hurt a pitcher’s FIP, nor does a triple. There’s too much of a defensive impact on those types of results.
That’s why ERA can be misleading; you’ll give up fewer runs with a better defense behind you. Look at every single pitcher who pitched for the Cubs last season and their ERAs.
Also, like mentioned earlier, ballparks have a huge impact on numbers. A park with shorter fences will yield a higher ERA, and also a higher FIP, as homers are included in FIP. So, for it to truly be a measure of how good the pitcher is, the stat would need completely to factor out ballparks.
How? Look at the contact the hitters were making, instead of looking at what that contact yielded. If you haven’t read it, Casey Boguslaw attempted to do just that, and you can read about it here.
Instead of using home runs, he used StatCast’s “barreled balls” as an outcome for a specific at-bat. That is truly independent of ballpark and defense, as a barreled ball at Coors Field is a barreled ball everywhere else. It doesn’t matter if the specific at-bat ended up as a triple or as a lineout, a barreled ball is just a barreled ball.
That got me thinking, why not do the same sort of thing for hitters?
There isn’t a stat which only quantifies outcomes a hitter can control. Fangraphs has soft contact percentage, medium contact percentage, and hard contact percentage, which are three outcomes a hitter can control, but doesn’t tell the average fan much about the player.
Opposing defense doesn’t change what kind of contact a player made, nor does a player’s ballpark. Again, hard contact at Coors is hard contact at Guaranteed Rate Field is hard contact at your local little league field.
Similar to pitchers, a player is mostly in control of their walk and strikeout rates, and while pitchers can influence a specific plate appearance, their effects over an entire season offset and are negligible.
Unlike slash stats, the type of contact a player makes is not influenced by a player’s speed, nor is it influenced by luck. It doesn’t matter whether a ball in play turns into a groundout or a double, as long as it’s the same contact. So, using these stats, which include little outside influence, one can determine what a player’s numbers should have been, had they had average speed, an average ballpark, with average luck, etc.
First, let’s adjust batting average. Batting average is the likelihood of a player getting a hit in a given at-bat, and a player’s at-bats doesn’t change much when changing other factors. However, how many hits a player would have gotten, if they faced league-average conditions and had league-average speed could be completely different than how many they actually produced.
It’s actually a much simpler answer than you may think. Just look at a player’s hard contact percentage, and then look at the league average when hard contact is made. Do the same for medium and soft contact, and then multiply that number by all of their at-bats resulted in contact (basically all non-strikeouts). This will produce how many hits an average player with the same contact percentages will get in the same amount of balls in play. Then divide that number, by the amount of at-bats they had in the season. This will give you the player’s expected batting average.
For on base-percentage, just use the hit number but plug it into the OBP formula, which includes walks and hit by pitches. Use the players actual walks and hit by pitches from the season, as there is little outside influence in those values. This will give you their expected OBP.
For slugging percentage, instead of using the league’s batting average on each contact type, use its slugging percentage. Using the same methodology as we used in batting average, we can determine a player’s expected total bases, and then divide by at-bats to get their expected slugging percentage.
What impact does this have?
First, let’s look at batting average (among qualified players in 2016). Who’s helped most by these adjustments?
As you can see, Adeiny Hechavarria was helped the most by the adjustments. Hechavarria had a decent hard hit percentage (32.3 percent), but didn’t have much to show for it, as when he hit the ball hard he only had a .447 batting average. This falls nearly 100 points below league average (.540).
On medium contact, Adeiny also fell well below average. The MLB hit .271 on medium contact; he hit only .204. Whatever caused this, whether it be (bad) luck, defense, or ballpark, averages out in the expected batting average stat. With average conditions, Hechavarria would hit much better. The same happens when you look at the numbers of the four other players.
Daniel Murphy leads the MLB in adjusted batting average (.321), followed by David Ortiz (.320), Adrián Beltré (.309), José Altuve (.309), and Buster Posey (.307). Those players actually finished second, eleventh, twenty-fifth, third, and forty-fourth in terms of batting average, respectively. I’ll touch on Posey’s leap a little later.
Who got hurt the most?
Nothing unexpected; a few fast guys, a Coors guy, and Joey Votto. Except Votto…that’s kinda weird.
Let’s look at why Votto gets hurt by this adjustment. On hard hit balls, he hit about 20 points above league average (.560 compared to .540).
On medium hit balls, his .349 is 78 points higher than league average. On soft hit balls, his average of .226 is, again, much better than league average (.156).
Votto isn’t all that fast, but beat out each of the league averages in each of the contact types. With league average conditions, Votto would be expected to regress to a batting average of .284 from .326.
When I look at who benefited and who got hurt the most under expected OBP, it’s unsurprisingly nearly the same list. However, a couple players like Charlie Blackmon (Coors) and César Hernández (speed) get punished under expected OBP more than Joey Votto because of Votto’s high walk rate, which is unaffected by this stat.
In terms of adjusted on base percentage, Mike Trout takes the crown (.410), with Josh Donaldson (.405), David Ortiz (.405), Joey Votto (.399), and Ben Zobrist (.399) immediately following. Those players actually finished first, fifth, sixth, second, and thirteenth in terms of on base percentage, respectively.
For slugging percentage, the players who get helped the most look a little familiar.
Again, Hechavarria and Alonso are both helped the most by expected slugging percentage. Jason Heyward, coming off a miserable season, sees some hope, as he hit the ball harder than his numbers show. An interesting name on this, in my opinion, is Buster Posey.
Posey, who slugged .434 in 2016, didn’t have a bad year like Heyward. He did drastically underperform, though. He slugged .905 on hard contact balls (league average was 1.133), while also slugging only .300 on medium contact balls (league avg of .320). When under average conditions, Buster Posey would be expected to slug .516, compared to his 2016 mark of only .434.
Lastly, who gets hurt most by expected slugging?
This list I find interesting, because it’s made up of big power guys. I’ll take a look at Brian Dozier. Dozier slugged a monster 1.545 on hard contact balls, over 400 points above league average. He’s about average on medium contact balls, but on soft contact slugged .216, about 50 points higher than league average. Again, under league average conditions, Dozier would be expected to slug .445, a big drop off from his .546 mark in 2016.
David Ortiz finished the 2016 season with the best adjusted slugging percentage (.566), beating out Daniel Murphy (.546), Adrián Beltré (.518), Albert Pujols (.517), and Buster Posey (.516) for the top spot. Those players actually finished first, second, twenty-fourth, sixty-seventh, and ninety-fourth in slugging percentage, respectively.
While not perfect, expected slash stats are much more valuable than the actual ones. Instead of telling you what happened, they tell you what should happen. It strips players of all of the factors they’re not responsible for, and asks the question, “how good of a pure hitter is _______?” While ballparks, speed, and luck play a huge role in a hitter’s slash line, the expected slash stats eliminate ’em all.
