How Well Have Hitters Been Hitting, Objectively? | Expected wOBA and Slash Lines

Posted By Spencer Estey on 12:11pm Jun 29, 2015 | 2 comments

(Title photo courtesy of Dave R, https://www.flickr.com/photos/daver6/)

This is the culmination of Breaking Blue’s mini-series on expected hitter peripherals. The idea of the series is to take a look at how hitters have been performing at each of the fundamental skills, from the perspective of the advanced new data that is available.

Through StatCast (as collected by BaseballSavant), we have information on how hard players are hitting the ball, in terms of exit velocity on both ground balls and fly balls, and distances. Through the batted ball information on Fangraphs that is provided by Baseball Info Solutions, we also have directional data in the form of the rates at which batters hit the ball to the pull, middle and opposite fields, as well as quality of contact. Pitchf/x has for years provided an easy source for metrics such as swinging strike, contact, and chase rates. We have a wide array of very granular statistics which with to do research. The four main rates that govern offensive production are strikeout rate, walk rate, isolated power and batting average on balls in play. There is almost a perfect relationship between the overall package of a hitter as expressed in a stat like wOBA, and these four stats. Productive hitters come in all different shapes and sizes, but they produce at the plate through avenues specified by these rates.

The ‘xSeries’ is designed to provide a window into a player’s season that is free from many sources of bias and luck. Players often carry batting lines that bloat or under-represent the actions they’ve made in the batter’s box to help their team. By regressing for the important rates using talent-based metrics, we can know with much more accuracy how they have actually hit, objectively. The results we’ll get are not meant to be projections or attempts at illustrating true talent. However, they in many cases will represent a player’s talent level fairly and will be meaningful when looking ahead. As said in previous articles, these numbers are meant to be one tool of many when evaluating a player’s outlook.

This article in particular will present our expected weighted on-base average (wOBA) as well as expected batting average, on-base percentage and slugging percentage. The latter three numbers make up the oft-referenced triple slash line.

The frameworks for the previous expected peripherals were for the most part considered, with some tweaking. Here’s a summary of the statistics used:

– Average fly ball/line drive exit velocity. How hard the hitter’s average fly ball or line drive is struck. (Used in xISO and xBABIP)

– FB%. Percentage of balls in play that are fly balls. (xISO, xBABIP)

– IFFB%. Percentage of fly balls that are infield flies. (xBABIP)

– GB%. Percentage of balls in play that are ground balls. (xBABIP)

– Speed score. Created by Bill James, represents a player’s baserunning speed. (xISO, xBABIP)

– Hard%. Percentage of batted balls judged to be the result of hard contact. (xBABIP)

– Pull%. Percentage of batted balls hit to pull field. (xBABIP)

– Swing%. Percentage of pitches swung at. (xK)

– Contact%. Percentage of swings that result in contact. (xK)

– Zone%. Percentage of pitches to batter that are in the strike zone. Pitchers generally throw fewer pitches in the zone to better hitters, so Zone% is relevant to a hitter’s plate discipline talent. (xBB)

– ZContact%. Contact rate on pitches in the strike zone. (xBB)

– OSwing%. Swing rate on pitches outside the strike zone. (xBB)

Here are some notes on methodology:

– Less inputs were used for BABIP in order to more closely represent talent. I wanted a regression that relies on only main effects of intuitive statistics. The previous xBABIP formula was good but we have to be careful with StatCast data since it’s limited. The statistics used in the xBABIP here should stabilize quickly and be much less prone to over-fitting.

– Expected strikeout and walk rates were made much more simple at no loss to fit. They are extremely intuitive and tell most of the story of how strikeouts and walks happen. Strikeouts are a product of swinging at pitches, and not making contact on swings.

– Not every event (i.e., sacrifices, hit-by-pitches) is represented in the peripherals used, so some additional manipulation was needed. wOBA was created using expected wOBAcon (wOBA on contact), walk and strikeout rates. Expected home run rate (needed for batting average since home runs don’t count towards BABIP) was created using expected isolated power.

– These numbers are not park-adjusted, so they represent neutral-context expected production. You can adjust them on your own using Fangraphs park factors. The peripherals were not significantly influenced by park since most of the stats considered are park-independent (i.e., Pull% and Swing%). However, future versions of the peripherals may make the necessary marginal adjustments.

– Hitters with 180 PA were considered here. Data is as of the morning of June 28th.

Breaking Blue will soon look closer at certain players that interest us, including Blue Jays and any players under- or over-performing by a notable amount.

We’d also like to create a permanent page on the website that has an oft-updated leaderboard table for the xSeries, since we believe it is a valuable resource. The next version of the statistics may run Wednesday or Thursday of this week.