by: Matthew de Marte – October 19th, 2018

Last season, Major League hitters combined to set a record for the most home runs hit in a season at 6,105. Although pitchers had set a record for most strikeouts in a single season for the 12th consecutive year, it appeared in 2017, hitters collectively reigned supreme over pitching. The narrative surrounding hitting in the modern day seems to be a lack of situational hitting, poor batting averages, too many teams neglecting small-ball, and the dreaded “launch angle swing” (which does not exist) causing mass punch-outs. Reality was, in 2017, hitters performed pretty well. To be specific, league-wide wOBA was .321 according to Fangraphs. Since 1960, there have been 22 seasons where league average wOBA bests 2017’s mark. Of those 22 seasons, 17 of them are from 1993-2009, also known as the Steroid era when hitters produced like they never had before. Although, juiced balls certainly played a role in 2017’s offensive surge, I wanted to believe hitters were adapting because 2017 was certainly one of the better offensive seasons since the MLB first expanded. That data was influencing how hitters were developing and training, and more hitters were achieving greater success because of modernized philosophies and training. So, using Statcast data courtesy of Baseball Savant, I decided to investigate the drop in home runs in 2018 and see what was influencing hitters’ regression.

Originally, when I came up with the idea for this analysis, I was strictly focused on where hitters were losing home runs in 2018 compared to 2017, and all four years of Statcast data. After I began looking at the data, I realized this was going to encompass far more than just home runs, but more of a comprehensive look at the variance of batted ball performance, and its impact on individuals’ production.

To begin, I had to choose a sample size. I gathered all the batted ball data provided by Baseball Savant. I then looked at the extreme launch parameters surrounding home runs during this time. Using these parameters, I selected my data set. All batted balls with a launch angle greater than or 13 degrees and less than or equal to 50 degrees, with an exit velocity greater than or equal to 87 MPH and less than or equal to 122 MPH were included in the sample size.

The first question I wanted to answer was, where did hitters lose home runs in 2018? To do so, I began by looking at the sheer quantity of balls batters launched with a Home Run probability of at least 50%. The following visualizations show the number of batted balls hit with at least a 50% home run probability in 2017 and 2018:

It can be hard to spot differences in these graphs, but the sample size for 2017 is 4,461 and 4,845 for 2018. It is difficult to tell whether these batted balls with promising HR probabilities traveled further in 2017 rather than 2018, but we can analyze the data in a different way. The following table shows batted balls from the entire data set from 2015-2018. Each row represents the number of batted balls that exceeded the HR Probability in each row:

HR Prob: | 2015 | 2016 | 2017 | 2018 |

N = | 32,487 | 33,783 | 33,602 | 34,144 |

HR > 0 | 19,903 | 21,072 | 21,208 | 22,031 |

HR > .25 | 6,633 | 7,634 | 7,622 | 8,178 |

HR > .5 | 3,840 | 4,531 | 4,461 | 4,845 |

HR > .6 | 3,326 | 3,880 | 3,774 | 4,138 |

HR > .7 | 2,476 | 3,028 | 2,902 | 3,217 |

HR > .8 | 1,829 | 2,171 | 2,104 | 2,331 |

HR > .9 | 977 | 1,176 | 1,158 | 1,252 |

HR = 1 | 134 | 192 | 227 | 278 |

Unless this data is *wrong, *it appears batters in 2018 put themselves in a better position to hit home runs this year. In fact, if we sum up the home run probabilities from 2018, the total number is 6,034.2. Not the best approximation for how many home runs *should* have been hit, but it is not a bad start as well. If we use the same method of thinking for 2017, there only *should *have been 5,597.9 home runs hit. So, how big of an impact did juiced balls have on hitters? It appears to be pretty significant. Before moving on, I think it is important to point out some potential trends from this data. Since Statcast’s inception in 2015, hitters have made strides to hit the ball out of the ballpark more. Since 2015, as you can see from the table above, every year hitters are hitting more batted balls that have the chance of leaving the ballpark. Looking solely at batted balls with a home run probability greater than 70%, since 2015 has increased from 2,476 to 3,217 (29.9 % increase!!), and the number of auto-homers (batted balls with a home run probability of 1) has more than doubled from 134 to 278. I think these numbers alone show some of the benefits of hitting the ball hard, in the air, and the changes hitters have made to hit balls that fall in the desired parameters more frequently.

We could continue to analyze solely home run probabilities and the sheer quantity of batted balls in this data, but including other variables can help tell a better story. So let’s start looking at these batted balls by exit velocity and launch angle. Home run probabilities are great, but when trying to improve as a hitter or find why hitters have improved, the answer generally lies in more hard hit balls, falling within the desired launch angles. Let’s begin by looking at some heat-maps. The following heat maps show the cumulative number of batted balls that fall within the specified exit velocities and launch angles in 2017:

Here it is for 2018:

Pretty similar distributions. If you look closely, you can see in 2018 batters hit more balls between 95-109 MPH. Batters actually hit 992 more batted balls in these bins. A very significant difference. So why were their less home runs hit in 2018 to 2017? Let’s take a look at how home run probabilities have changed these specific years. Earlier we looked at some home run probabilities, and those numbers were derived from the entire Statcast era. The home run probabilities we are going to look at now are from just 2017 and just 2018 respectively. Looking at these probabilities should give us a glimpse as to where 2017 hitters were lucky, and where 2018 hitters were hapless:

2017:

2018:

Batters who hit the ball between 95-109 MPH had much more leeway to earn home runs in 2017 than 2018. Considering 89.7 % of all home runs in the Statcast era have been hit with these exit velocities, that makes a gigantic difference. If you are not used to looking at heat maps, you still may be unsure why that is such a big deal, but the following heat map shows the difference between 2018 HR Probabilities and 2017 HR Probabilities. Positive numbers indicate hitters had a better probability of hitting a home run in the specified bin in 2018, and negative numbers indicate hitters had a better probability of hitting a home run in the specified bin in 2017:

There you have it! In 2018, hitters’ home run probabilities plummeted, mainly in the 100-109 exit velocity range. If we go back to the heat maps looking at batted balls hit in these range in 2018, there was over 12,000 batted balls that fell in this bin. That could have implications of massive setbacks offensively, when in reality hitters were making better quality of contact.

Let’s take a look at one last visualization before drawing conclusions. The following plot compares each player’s expected home run total (Sum of home run probabilities) versus his actual home run total for 2017 and 2018. Home Run probabilities are based off of Statcast data from 2015-2018; the same numbers we used at the beginning of this analysis, NOT the numbers used for the heat maps above. The names highlighted are either players who vastly outperformed their expected home run totals and those who under-performed their home run total. The diagonal line represents a slope of one where expected and actual home run totals are equal. Above the line, you can say players were lucky, and below the line you can say players were unlucky.

The first thing that jumps out to me is the mass migration of the league. In 2017, most of the league lived above the line, in 2018 most of the league lived below the line. Didi Gregorius, who is notorious for pulling home runs at the Yankee Stadium short porch, has been arguably the luckiest home run hitter in baseball the past two years, being an outlier on both graphs.

There are a few main takeaways I keep taking away from this analysis. The first is that in 2017, the balls were juiced. These juiced balls surely made a significant difference in hitters’ production. While hitters performed better statistically in 2017, there may be evidence that their overall quality of contact improved in 2018. This season, hitters made contact at desired exit velocities and launch angles more than they did in 2017. The only problem for them was batted balls certainly did not travel like they did in 2017. I think this report can be evidence that hitters are making the adjustment. Although, strikeouts are increasing every year and hitting is becoming more difficult, since the inception of Statcast, every season hitters as a whole have made efforts to hit their hard hit balls in the air! I hope you all enjoyed the 2018 Home Run Report. This analysis was probably my favorite to do, so please if you would like to continue the conversation, have suggestions to better improve the research design, would like to use the data I gathered for yourself or have any questions, please reach out to studentsofbaseball@gmail.com.