2013 saw a dramatic contrast in organizational strategy for two playoff teams. On the one hand, the Pirates’ starting rotation exhibited the largest groundball rate (52.8%) of any starting rotation since 2002, which is as far back as batted ball data go. On the other hand, the Athletics’ starters posted the lowest GB% of 2013 at 39.1%, which is not the lowest since 2002 but is still quite low. Despite the large difference, both teams sported an effective rotation that was good enough to get their teams into the playoffs.
It is known that ground balls are more favorable for pitchers than fly balls, as ground balls lead to fewer runs than fly balls do. Using all teams’ starters’ batted ball data since 2002 on a season-by-season basis, I correlated GB% and FB% with FIP to quantify the effect in a different way. GB% is indeed negatively correlated with FIP; as GB% increases, FIP decreases. FB% was positively correlated with FIP, which means that fly ball oriented teams should perform more poorly. In order to gain some perspective on what the Athletics and Pirates accomplished, I compared them to teams who had a similar GB%. I isolated all teams since 2002 whose starting rotations had a GB% either above 50% or below 40%. There were a total of 35 teams, 7 of which were in the high GB% group and 28 of which were in the low GB% group. The 2013 Pirates, Athletics, and Orioles are included in the sample.
The 7 teams in the high GB% showed desirable results. Of the 7 teams, 6 made the playoffs. Interestingly, all 7 teams were NL teams, 5 of which were the Cardinals or Dodgers. The average ERA/FIP/xFIP slash line was 3.79/3.83/3.88, and the range of fWAR was small. Conversely, only 6 teams of the 28 with a low GB% made the playoffs. Their ERA/FIP/xFIP was about a run higher, and the range of fWAR was much larger. It appears that having a low GB% is associated with unstable, generally poor performances, while having a high GB% is associated with stable, strong performances.
Now that context for each group is established, each team can be examined within their groups. Travis Sawchik of the Pittsburgh Tribune-Review wrote about the Pirates’ strategy, which consisted of teaching a sinker, using that sinker, and shifting aggressively to mask an infield defense that was rated as mediocre-ish in 2013. The beauty of this strategy is that it does not require excellent defense to be successful. Only 1 team, the 2005 Cardinals, outperformed their FIP with a very solid infield defense. Outperforming an average 3.83 FIP is not necessarily a requirement for having a solid rotation. What the Pirates did this season very much fits in with historical precedent, but the Pirates were a bit more extreme in their emphasis and shifting. It paid off, as the Pirates had a 3.46 FIP despite a high walk rate.
The Athletics’ performance is more the exception than the norm in historical context. SamYam for the Athletics SB Nation site, Athletics Nation, wrote about Oakland’s strategy, which was to take advantage of their home ballpark and strong outfield defense to produce outs (EDIT: Chris Moran also wrote about this strategy for FanGraphs). Of the rotations that had a low GB% and outperformed their FIP, each one had a strong outfield defense and a favorable home stadium. For example, the 2002 World Series Champion Angels, who had a 38.8% GB%, had Darin Erstad in CF during his defensive prime. The Coliseum and the Oakland outfield defense, consisting mostly of Josh Reddick, Yoenis Cespedes, Coco Crisp, Chris Young, and Seth Smith, allowed the Athletics to align with the strong historical performers in the low GB% category.
This analysis does have some weaknesses. First, I did not incorporate shifting data. If these data are publicly available somewhere, please point me in that direction. Second, the 40% and 50% GB% thresholds were somewhat arbitrary, though the 50% number wasn’t completely random. The league average GB% is not exactly in the middle between 40% and 50%, but it’s close. A more sound method to determine thresholds would be to use standard deviations from the league average GB%. Third, given that all 7 teams in the high GB% sample were NL teams, the league could have confounded the results. Finally, the correlation analysis was potentially confounded by strikeout rates. The 7 high GB% teams in my sample had a strikeout rate about 2 percentage points higher than the 28 low GB% teams. As strikeouts are a component of calculating FIP, it is expected that a higher strikeout rate would lead to a lower FIP. The difference between the two groups’ FIP is large enough that the difference should still be significant even after accounting for strikeout rate.
When comparing high GB% teams to low GB% teams, it appears to be an uphill climb for low GB% to match the performance of high GB% teams. Teams like the 2013 Pirates do not need to take advantage of a strong defense or a stadium that suppresses offense in order to excel at run prevention, whereas teams like the 2013 Athletics need those things in order to go against history and excel at run prevention. Focusing on having a high GB% renders infield defense less valuable, which allows teams to place offensive-oriented players in the infield without sacrificing run prevention. Finding offense at those positions is another story.
*All statistics courtesy of FanGraphs.com.