# Batted-Ball Trajectory: Splitting the Difference

One of the most familiar tactics in baseball is the handedness platoon. Nearly every adult fan knows that a pitcher does better against batters with the same handedness, and batters fare better against pitchers of the opposite handedness. Managers set their lineups to exploit this advantage, and in the late innings of a tight game will often burn through their benches and bullpens seeking an edge for as little as one plate appearance.

Not as well known is another kind of platoon. Players can produce a preponderance of types of batted balls that they hit or allow, between ground balls and fly balls. The proclivity tends to stick: you can be a groundball pitcher, a flyball hitter, and so forth.

In *The Book, *Tango, Lichtman and Dolphin examined the splits of the trajectory platoon, as I will call it. They found that when one type of batter faces one type of pitcher, the rules of the handedness platoon carry over. The batter would rather see his opposite type than one of his own style. (The pitcher wants to see a ground-ball batter because grounders do that much less damage. However, between batters with identical wOBAs, the pitcher would definitely prefer his own trajectory type.)

The trajectory platoon is not as dominant as the handedness one, mainly because it’s a continuum from one extreme to another. Batters and pitchers perform either lefty or righty: not even switch-hitters can hit 60 percent lefty in a particular at-bat*. It is a real effect, though, one that appears not to be exploited by managers in the same way. Although … perhaps the next time we’re decrying a weird pinch-hitting or bullpen move, we might look at it from this angle to see if the manager’s thinking on another level.

**Technically this may not be true. A close reading of the rules permits a hitter to change batter’s boxes between pitches, as long as there aren’t two strikes on him. Of course, who has ever seen a hitter do this? One could conjure a scenario where this might give the hitter some kind of edge, but in practice nobody cares to try it.*

But why does this platoon exist? With handedness, the angles at which various pitches approach the plate can give one kind of batter a better chance at hitting them well. With batted-ball trajectory, I postulate that we see the effects of tendencies either adding together or cancelling out.

With two flyballers facing each other, the high angles they’re both seeking could often add up to pop-ups, easy plays for fielders. With two groundballers, added downward trajectories could turn a sharp two-hopper through the infield into a little dubber the pitcher easily scoops up. And when one type faces another, they may strike a balance.

Between grounders and flies sits the zone of the line drive, well and away the most productive of the three main classifications of batted balls. If one players wants a fly and the other a grounder, a stalemate between their tendencies might produce a bounty of liners. This could go a long way to explaining why the opposite trajectory platoon is better for hitters than a same-type platoon.

It is this last possibility that I looked into: whether matches between strong opposite platoons produce more liners, and thus more offense, than strong like platoons. I happily picked up a few other facts along the way, but the trajectory platoon was my starting point.

For the 2013 and 2014 seasons, I gathered statistics for each hitter who had at least 150 plate appearances in each season, and for each pitcher who threw at least 50 innings. I made the pitchers’ threshold low so I would get a fair sample of relievers in with the starters; the batters’ threshold was kept low to match. This gave me 399 and 394 batters for ’13 and ’14, plus 326 and 330 pitchers.

In each year, I separated out the 10 percent of players with the highest ratio of ground balls to fly balls, and the 10 percent with the lowest ratio. I tallied up the batted ball types (grounder, liner, fly) in all plate appearances involving members of those groups. (I excluded bunts because they seek a very specific type of hit, and because there is no such thing as a line drive bunt.) As control groups, I also took the middle 10 percent of hitters and pitchers to compare to each other.

As my first sidebar, I did find that strong trajectory tendencies for batters do persist year to year. Among the 40 batters in the top decile of grounders in 2013, 19 of them were in the top decile in 2014, with five others not playing in the majors at all that season. For top-decile flyballers, 16 of the 40 maintained their place in 2014, with three others out of the majors. Neutral tendencies don’t persist: the number for the middle 10 percent was just three out of 40.

Pitchers showed a lower rate of persistence at the extremes. For ground ballers, 11 out of 33 stayed in the top decile from ’13 to ’14, while nine out of 33 extreme fly ballers carried over. Compared to an eight of 33 rate for the middle cohort, persistence for pitching outliers is slight. (I was also able to check groundball pitchers from 2012 to 2013, where 13 of 33 stayed in the top decile. This higher rate, along with the small samples overall, makes me less inclined to suggest that there is no persistence for pitchers.)

Back on the original topic, the trajectory data do support my theory. The tables below show rates for all four platoons, the meetings of the neutrals, and the league-wide numbers. Rounding keeps percentages from adding up evenly in a few cases. Total balls in play are given to show sample sizes; those for the majors are an approximation.

2014 Batted Ball Trajectories |
---|

Batter/Pitcher |
GB% |
LD% |
FB% |
Balls in Play |

All/All | 44.8 | 20.8 | 34.4 | ~125K |

GB/GB | 74.1 | 16.3 | 9.7 | 652 |

GB/FB | 45.2 | 21.2 | 33.6 | 684 |

FB/GB | 45.8 | 23.9 | 30.2 | 589 |

FB/FB | 20.5 | 17.8 | 61.6 | 730 |

Neut./Neut. | 43.4 | 21.4 | 35.1 | 1,068 |

2013 Batted Ball Trajectories |
---|

Batter/Pitcher |
GB% |
LD% |
FB% |
Balls in Play |

All/All | 44.5 | 21.2 | 34.3 | ~126K |

GB/GB | 69.0 | 17.8 | 13.2 | 870 |

GB/FB | 40.9 | 22.4 | 36.7 | 687 |

FB/GB | 46.7 | 22.2 | 31.1 | 807 |

FB/FB | 23.9 | 18.8 | 57.3 | 756 |

Neut./Neut. | 41.6 | 21.0 | 37.4 | 1,100 |

Both years, opposite platoons do better in line-drive percentage than matching platoons, average versus average, or the majors as a whole. The average/average match-ups were near enough to the league-wide numbers that I really could have gone with the latter, though it’s a good check to see the figures line up as closely as they do.

There’s no clear superiority to whether the batter in the good platoon is a ground ball or fly ball hitter, though the total numbers lean toward fly ballers hitting against ground ballers. Line-drive numbers in the bad platoons also sit a little higher for flyball hitters, adding some weight to the argument that they are better at producing liners.

But then I checked the total 2014 numbers for the grounder and fly cohorts, and found the opposite. The groundball batters had an average liner rate of 20.5 percent against the whole league, versus 19.6 percent for the fly ballers. I cannot quite call that conclusive, but it certainly keeps me from claiming any line-drive superiority for the extreme flyball hitters.

Another side angle that presented itself is which player has more effect over the trajectory of batted balls. The major league groundball/flyball ratio was 1.30 in both 2013 and 2014. I looked at the ratios for the various breakdowns, though for this table I left out the same-style platoons.

Grounder/Fly Ratios |
---|

Batter/Pitcher |
GB/FB |

All/All | 1.30 |

GB/FB | 1.22 |

FB/GB | 1.51 |

Neut./Neut. | 1.17 |

When facing his opposite style, the pitcher produces his kind of batted ball at rates better than the league average. His weight is greater in the scales. I won’t venture to put percentage values on pitchers’ and hitters’ contributions, especially due to the numbers put up by the neutral players against each other: more fly ball oriented than either platoon.

This inches into a point covered by *The Book*. Neutral pitchers fare more poorly against batters than pitchers on the extremes, grounder or fly. A lower grounder/fly ratio might explain some of this, as flies are more offensively productive than grounders (though the tables show no real change on those super-productive liners). It also suggests that having a plan to force one type of contact, ground or air, is better for a pitcher than just humming it up there and letting the batter dictate batted-ball type.

An opposite effect, though, seems to arise for batters. My data included plate appearances for the hitters, and I found an interesting pattern in how many they received. (Recall that this is based on a 150 PA minimum.)

Mean Plate Appearances for Batted Ball Types |
---|

Year |
GB |
Neut. |
FB |

2014 | 401.1 | 447.8 | 382.6 |

2013 | 387.9 | 424.5 | 407.3 |

Batters with neutral splits got more playing time than those on the extremes, by an average of 35 to 40 PAs. Also, players with fewer than 200 PAs numbered at least six out of both ground baller and fly baller cohorts in each year, but for neutral players numbered two and three for ’14 and ’13. This implies that the neutrals are better batters, whose managers get them out there more often. Whereas pitchers ought to be looking for one type of contact, batters apparently should not be—unless it’s liners.

This actually conforms to the traditional roles the two players have. Pitchers act; batters react. It therefore makes sense for the pitcher not to cede control of even one aspect of the encounter, and for the batter not to be too fully committed to a goal that the pitcher can guard against.

Regrettably, the nuggets I’ve unearthed aren’t as widely applicable as one might wish. The groundball and flyball categories, by design, cover just 10 percent apiece of batters and pitchers in the league. (In *The Book*, they go between 12 and 17 percent with their extreme trajectory buckets.) You can expect, on average, one extreme grounder hitter and one extreme fly hitter in a lineup, if that.

Despite that narrowness, we do see some solid conclusions. One of them is even the one I was hoping for: opposite trajectory platoons produce an elevated line-drive rate that helps batters perform better than against same trajectory pitchers. I do like seeing one of my hypotheses pan out. It happens seldom enough.

There is only so much, though, that a top-down analysis such as mine can reveal by drilling into the macro statistics. A bottom-up approach, looking at exact trajectories drawn from the PITCHf/x and HITf/x databases, could find far more. *Hardball Times* readers may find that here in the future. Perhaps from me; perhaps from somebody else.

### References and Resources

- Acknowledgments to Tom Tango, Mitchel Lichtman, and Andrew Dolphin for
*The Book*, the foundation of my work today (and many others’ on many occasions). - Trajectory rates and specific batted-ball data came from FanGraphs, with Baseball-Reference providing additional figures.
- For any follow-up researchers: B-R uses different trajectory criteria, producing different line-drive rates than FG’s. Its grounder/fly ratio numbers are also greatly different, apparently because it counts line drives as flies.

Is there a specific reason that you use GB/FB ratio instead of GB% and FB%? I would guess the GB/FB ratios are a decent approximation of GB% and FB% but to me it is like using K/BB ratio to evaluate strike out rates.

Interesting nonetheless.

AC, I was trying to keep line drives out of the equation. Groundball and flyball frequencies would correlate partially to liner frequency: the more of A or B, the fewer chances there are for C. The GB/FB ratio hopefully cuts that connection.

That is certainly true. The point I was trying to make is that I think it would be more accurate to sort your top and bottom 10% based on FB% and GB% and not GB/FB ratios. As far as I understand you, you try to approximate a swing path (steeper for FB hitters and flatte for GB hitters). So you do not actually care for LDs. IMO sorting by FB% and GB% would give you purer results, allthough the conclusions probably woudn’t change at all b/c GB/FB ratios are a decent approximation for GB% and FB%.

I see your point that flat percents would have produced the most extreme results of each type of player, but Shane’s main question was to see if opposite type matchups produced more liners. Using high GB/FB percents would have incidentally produced players with lower line drive rates, where ratios won’t necessarily show lower liner rates.

I see your point that flat percents would have produced the most extreme results of each type of player, but Shane’s main question was to see if opposite type matchups produced more liners. Using high GB/FB percents would have incidentally produced players with lower line drive rates, where ratios won’t necessarily produce lower liner rates.

Great stuff Shane – I always love it when the next natural question in my mind gets answered a bit later in the article.

I wonder if there is a way to figure out the correlation between a hitter’s GB/FB ratio and the GB/FB ratios of the pitchers he’s faced. I’m not sure of the statistically correct way of calculating this, but it would be great to see if there’s a correlation between that metric and the hitter’s wOBA. My hypothesis would be that the hitters whose GB/FB tendencies are the least correlated with the pitchers’ tendencies would be the better hitters overall (consistent with your last chart)

No MLB batter that is not a pitcher ever tries to hit the ball on the ground except for the special cases of a man on third with less than two outs, when a runner is running on the pitch, and occasionally a swinging bunt from a fast left handed batter. Ground balls are basically failures. Every player is trying to hit the ball hard in the air so that it leaves the bat between 0 and 25 degrees. If a player could be successful at that in even 50% of his PAs he could strike out the other 50% and still be an above average player.

He’s not saying the batters necessarily try to be GB hitters, he’s just relaying that some have an extreme tendency to hit grounders, and seeing what types of batted ball they produce against different pitchers.

great stuff. im wondering about the outcomes of these splits-avg, iso, wOBA, hr/fb, IFFB% etc. Seems like GB/FB would results in the best line, given the high rate of LD’s and solid FB%, FB/FB would result in lots of pop ups, and GB/GB is ok for singles and not much else.

also- how do batter/pitcher handedness effect each of these?

An interesting experiment would be to look at pitchers like Matt Cain who historically outperform the FIP.

The purpose is simply to understand if there is any pitcher “skill” in not giving up homers or line drives – which is presumably what outperformance vs. FIP seems to indicate.

FWIW, a batter may switch from one batters box to the other at any count.

MGL, that is what the official rulebook says. However, I recall from some research I did a few years ago that there is a restriction on two-strike movement. From fuzzy memory, it was in a book of comments on the rules (not the Comments in the official rulebook) that umpires use, that has the force of rule on ambiguous areas, and on a few areas that don’t seem so ambiguous. I do find it incongruous that there would be a “secret” rulebook taking precedence over the public one.

I might be misleading myself, but then where else would I get the two-strike notion? I’ll look through a few books I’ve got and try to track this thing down.

For some reason, it is a common myth that the batter cannot move with 2 strikes, but it is not true. Don’t know where that came from. There is no secret rule book. The rules are the rules and this rule is quite clear.

MGL, I found similar comments in an online search about the two-strike restriction being a myth. I guess I happened upon a source that hadn’t gotten the word, and thus I didn’t get the word.

As for the “secret” rules, I’m beginning to think I was misremembering a passage from Bruce Weber’s “As They See ‘Em”. Jim Evans, running an umpiring school, produced an annotated version of the rulebook running about six times the length of the original. I may have conflated that into an official arcane resource for the umpires.

I hate when my memory gets creative.

Hateit.I’ll ask the wizards behind the curtain to take out the two-strike reference. And much as I have to grit my teeth, I thank you for holding my feet to the fire.

There was in fact a separate case book of rule interpretations until 1977. It was merged into the rule book, and you now see that content as comments below the rule.

You’ve just scratched the surface. Left out of your study are swings and misses, and foul balls. I think this theory would predict those to change as well, leading to more strikeouts and pitchers’ counts when the pitcher has this “platoon advantage” and vice versa when the batter does. To understand the true effect of this platoon, you really need to look at the outcome of the whole PA, not just the trajectories of batted balls outcomes. At the aggregated level, the LH/RH platoon split alone already produces changes in batted ball trajectory outcomes, so you’d need to control for that, as well as trying to control for overall batter and pitcher quality.

If you can locate the first annual STATS Baseball Scoreboard (published after 1989 season), it contains a study of this issue too.

One thing that jumps is the flyball rates in the FB/FB matchup, they are extremely high, just as with GB rates in the GB/GB matchup. This is obviously expected, but it brings up an interesting point. Flyballs have extremely different wOBA expectations between flyball hitters. Chris Carter has a .489 wOBA for his flyballs. Jed Lowrie, who is a similarly extreme FB hitter, only has a .271 wOBA.

I’m wondering if Chris Carter actually ends up benefiting from a FB/FB matchup because he gets such great value from his flyballs, while Lowrie benefits greatly from a GB matchup because he gets the majority of his value from line drives. Could you possibly run this study with a filter for career groundball flyballs wOBA outcomes?

Also are you doing this research in R?