Earlier this season, there was a bit of newspaper coverage regarding how the Braves’ new leadership had begun implementing some of the lessons learned from analytics with regard to the Braves’ defense. In April, Freddie Freeman said this (courtesy of Dave O’Brien at the Atlanta-Journal Constitution):
“Every day we have a meeting. We have a hitters’ meeting and we have cards sent to us in our lockers,” Braves first baseman Freddie Freeman said, referring to index-like cards that each position player has tailored for him each day by the analytics staff. The cards list the opposing players’ tendencies and recommended defensive positioning.
Here’s manager Brian Snitker, on the same:
“I think overall we’ve played really good defense,” Snitker said. “I think the whole analytical thing and the defensive position and what Alex (Anthopoulos) and his team have brought in here have helped us. People ask me and I say, it works. I’ve seen it work already.”
Nubyjas Wilborn of the Marietta Daily Journal recounted more or less the same, also in early April.
Anecdotally, it’s hard to say that this change has not paid dividends. The Braves have benefited from a number of crucial double plays already this season, and a stroll through memories of this season should probably yield as many “I thought that was a hit for sure but the infielder was right there” instances as “Well, they would have gotten that ball had they not been shifted so weirdly,” if not far more. But, we’re not about anecdata here. We’re about some real data. So, let’s take a look.
The basic question that seems worth answering here is the extent to which the Braves have improved their outcomes in the field. On the one hand, that’s kind of an easy thing to answer — one only has to look at team BABIP allowed to get a glance. On the surface, this narrative easily holds up:
- 2015: .305 BABIP-against (fourth-highest in MLB)
- 2016: .293 BABIP-against (20th-highest in MLB)
- 2017: .302 BABIP-against (11th-highest in MLB)
- 2018: .284 BABIP-against (22nd-highest in MLB)
So, there seems to be some truth there. If you array all the team-seasons since 2015, the 2018 Braves are currently 19th — in the top quintile — of these team seasons in defensive efficiency (which is basically the opposite of BABIP-against, i.e., how often a ball in play is converted into an out). So far, so good. You can stop there. But, parsing this down yields findings of greater interest.
The main thing that comes to my mind when we talk about defensive positioning is grounders. The reason for this is that a greater proportion of grounders is “handle-able” — per Inside Edge data in 2017, about 77 percent of balls hit to outfielders were “even” (40 to 60 percent likelihood of being made) or easier, while 86 percent of balls hit to infielders were the same. (The broken down categories tend to be similar; the difference infielders face more “possible” plays and very few infield plays are “impossible,” while a whopping 16 percent of outfield plays in 2017 were scored as “impossible” by Inside Edge.)
Previously, the public sphere was lacking the type of data necessary to be able to go much beyond BABIP to determine whether a particular ball should have been fielded. Luckily for us, Statcast lets us get somewhat further in this exercise. Statcast publishes both wOBA and xwOBA. Just for a quick refresher, wOBA is a catch-all offensive statistic scaled to OBP that gives hitters differential credit for walks versus singles versus doubles versus triples versus homers versus outs. The higher the better. xwOBA, meanwhile, is a measure of what wOBA should have been on any ball or series of balls, based solely on the exit velocity with which that ball was struck, and the launch angle at the point of contact. If wOBA exceeds xwOBA, that means the hitter got “lucky” (or the defense goofed); if xwOBA exceeds wOBA, the hitter got “unlucky” (or the defense sparkled).
Here’s a basic table showing the wOBA and xwOBA against the Braves on all grounders, 2015 through this chunk of 2018.
You can see that over the last three seasons, the Braves actually allowed better results on grounders than the quality of contact on those grounders warranted. This year, though, there are two changes. First, the Braves are both eliciting easier grounders to field, but they’re also fielding them commensurately. The near-zero delta confirms everything we’ve already discussed: the Braves are now better at turning grounders into outs, even after you adjust for the fact that the pitching is also eliciting easier-to-field grounders this year than in past years.
It’s also important to confirm that the trend above is specific to the Braves, and not a league-wide occurrence. After all, if the entire league was hitting weaker grounders and all fielders were getting better at converting them to outs, this wouldn’t be an interesting finding. But, that’s not the case.
While the Braves are still average-y at outperforming their xwOBA on grounders (their delta ranks 14th; a higher delta means wOBA is greater than xwOBA and therefore the fielding is less effective), it’s an improvement over the past few years. While the change is not too big relative to 2015, we should remember that Andrelton Simmons was still patrolling the infield in 2015. (More on defensive acumen of fielders later.)
Critically, the league-average deltas on grounders have been 0.009, 0.010, 0.009, and -0.002 for 2015 through 2018, respectively. (There’s an issue here where the xwOBA formula has changed for 2018 relative to 2017, but we’ll leave that aside for now.) If you compare them to the Braves’ deltas, what you find is actually not that impressive — subtracting the league-average delta from the Braves’ delta in each year yields: 0.004, 0.012, 0.004, and 0.004. Again, higher is better, but relative to all grounders hit everywhere in every season, there does not appear to be much of a difference. The Braves have gotten better, but the goalposts have also been moving.
So, on the one hand, you have some data indicating that the Braves have indeed done a better job at converting grounders to outs, even when adjusting for the quality of contact on the grounder. On the other hand, this “better job” hasn’t meant much. In 2017, the Braves were 21st at eliciting easy-to-field grounders, and their ability to convert them to outs was 16th-best. So far this season, they’ve elicited the easiest-to-field grounders, and their conversion rate has been 24th-best.
But, let’s think about this for a bit. Grounders come in all shapes and sizes. Some are the result of good contact and/or bad pitches, some are the result of bad contact and/or good pitches. When deciding where to position fielders, analytics departments have myriad choices. On one extreme, they can position fielders to where the ball is most likely to go if the hitter plays into the pitcher’s gameplan, i.e., weak contact. On the other, they can position fielders to where the ball is most likely to go if the hitter “beats” the pitcher (and still hits the ball on the ground), i.e., hard contact. They could also try to compromise, but that’s tricky — if you chase two rabbits, you will lose them both — or in this case, end up out of position to field either weak contact or hard contact. So, with that...
Grounders hit at 95+ mph
I think the tables below largely speak for themselves in terms of which approach the Braves’ analytics are taking.
Before this season, the Braves weren’t very good at cutting down grounders (wOBA > xwOBA), but they were similarly meh at cutting down hard grounders (wOBA still > xwOBA, though by not as much in 2016 and 2017). In 2018, however, there’s a dramatic turnaround. Yes, the Braves are still managing contact, even on hard grounders, better than before. But, look at the dramatic difference between the wOBA on those balls, versus the xwOBA. .050 in wOBA is not an abstract figure: .050 wOBA is the difference between Ender Inciarte (.285 wOBA, 77 wRC+) and Ronald Acuña (.336 wOBA, 112 wRC+) this season.
The ranks tell a similar story:
The Braves used to be among the worst in the league in converting hard grounders to outs, relative to their quality of contact. Now, they’re among the best.
For grounders hit at 95+ mph, the league-average delta for 2015 through 2018 has been: -0.005, -0.007, -0.010, -0.019. This is actually fairly interesting in its own right — negative deltas for these balls relative to a positive delta for all grounders suggests that indeed, most of baseball has already been positioning to handle harder-hit grounders. Most of baseball, except the Braves, whose deltas on 95+ mph grounders were still positive until this year. Unsurprisingly, subtracting these deltas from the Braves’ deltas yields values of 0.020, 0.015, 0.017, and -0.034 for 2015 through 2018, respectively. The Braves were giving up an extra 0.015-0.020ish on hard grounders for 2015-2017; they’ve now clawed all of it back and are actually doing way better (way, way, way better) than the league on fielding these balls.
Shift smarter, not harder
It may be tempting to look at these data and say, “Well, duh... they’re just shifting more.” Perhaps Robert D. Manfred Jr. is sitting in his alabaster New York tower and thinking these same thoughts. Those cursed Atlanta Braves and their infernal shifts, he may be thinking. They’re sucking all the fun out of baseball, turning even hard-hit grounders into outs at a preternatural rate! But, if Mr. Manfred is indeed thinking this, or even if you are, I regret to inform you that this is not correct: the Braves are actually shifting less. They’re just being smarter at positioning.
Baseball Savant has been rolling out tons of neat stuff to little heraldry this year. They now have an entire fielder positioning module. While, right now, it’s still relatively aggregated and not as granular as I’d like, the facts of the matter are:
- 2016: Braves 16th in overall shifts, 11th in shifts against righties, 23rd in shifts against lefties
- 2017: Braves 14th in overall shifts, 11th in shifts against righties, 12th in shifts against lefties
- 2018: Braves 16th in overall shifts, 9th in shifts against righties, 26th in shifts against lefties
The overall shift percentage has jumped from about 12 percent the last two years to 17 percent in 2018. The Braves have only jumped from 11-12 in 2016-7 to 15 percent.
They’ve made all these gains without elaborate shifts, but with granular positioning. Compare 2017 (left) with 2018 (right). Sure, there are differences. But they’re not the types of differences a greater emphasis on overshifting would necessarily indicate.
I bring this up mostly for a quick sidebar: banning shifts won’t accomplish much, so long as teams will still have the latitude to granularly position their fielders. Let’s not even entertain the idea: it’s a dumb one.
Okay, but what about fielder quality?
This, admittedly, is a sticking point. xwOBA adjusts only for exit velocity and launch angle. It does not adjust for fielder starting position (if it did, it wouldn’t be of use for this particular exercise, though it would be pretty cool nonetheless). It also doesn’t adjust for the quality of the fielders themselves. The Braves have had a lot of infield turnover. Jace Peterson played a lot of second base before being supplanted by Ozzie Albies. No third baseman has received more than 1,800 innings for the Braves since the start of the 2015 season; Johan Camargo currently has only about a fourth of the innings at the hot corner than what Adonis Garcia accumulated since the start of the 2015 campaign.
It’s possible that this is all just spurious speculation, and that the difference between the 2018 Braves, and the 2015, 2016, or 2017 versions of the Braves’ infield is superior fielding talent. This is difficult to assess: because the current defensive metrics (i.e., UZR and DRS) handle awarding of “points” or “runs” on a play made/not made basis, they don’t directly account for positioning in the way you’d like to validate or counteract this analysis. In other words, if you’re wondering whether Dansby Swanson’s elite defense (8 DRS, 4 UZR in 417 innings) this season is the result of improved positioning, or whether the data shown above have nothing to do with positioning but are the result of Swanson’s elite defense, that question is tough to answer right now.
You can debate amongst yourselves, but the 2015 data seem to be telling. An infield defense with Andrelton Simmons allowed a .376 xwOBA and a .391 wOBA on 95+ mph grounders. In 2018, an infield defense with no Andrelton Simmons has allowed a .313 wOBA with an underlying .366 xwOBA on 95+ mph grounders. Given that these grounders allow for little reaction time, I tend to think it’s the positioning improving the defensive metrics, and not the defensive metrics informing the wOBA/xwOBA delta. But, who knows?
Wrapping up: outfield defense
This same sort of analysis can be done for outfielders as well. The results are similar in some ways, and more mixed in others.
Relative to the league delta, the Braves’ “extra” deltas (same calculation as before) have been 0.005, -0.018, 0.008, and 0.002 on all flies and liners, and -0.004, -0.018, 0.014, and -0.027 on 95+ flies/liners. Again, there’s some improvement there, which is to be expected: you can position outfielders just like infielders, and once again, the Braves appear to be putting their outfielders in a position to be better able to cut down balls that are the result of the hitter’s gameplan superseding that of the pitcher. Interestingly, you can see some of it working in limited extent by examining the starting positions of Nick Markakis and Ender Inciarte:
Both Inciarte and Markakis have moved back their average starting positions relative to last year — the 2018 dots on the graphics are the ones furthest back. Inciarte has moved back every season (305 —> 310 —> 315); Markakis went from moving in to moving way back in 2018 (291 —> 288 —> 286 —> 295).
The Braves are a talented team, but they are not one of the most talented teams in the majors. In a current virtual tie for the division lead and going up against a team with more firepower and a larger payroll in the Washington Nationals, they are going to need to use every advantage to skew their run differential and pull out as many wins as possible. The data available to date indicate that their improved defensive positioning is more than a good start. The purpose of much of baseball analysis is to help get the most out of each player and each team. That looks to be exactly what’s happening here.