In part one of this series, I found that wOBA works really well, showing there is very little difference in runs scored between OBP-heavy and SLG-heavy teams. Today, I look at the individual hitters in these lineups, seeing if too many of the same hitter hurts production.
Using the same time period (1996-2012), I compiled every batter who had at least 250 PA with a single team during a season, giving each team anywhere from 8-12 hitters. Each hitter was given a "style" using OBP-wOBA, with values generally ranging +/- 50 points. To find the similarity of the lineup, I found the standard deviation of all the styles for each team, plotting that versus the residual runs per game for the team. Here is what all 506 teams look like together:
In case you can't see the trendline, it has a slight negative slope right along the x-axis. All together, there is no trend depending on whether the team has a bunch of similar hitters or a diverse lineup. Now splitting the sample into SLG-heavy, neutral, and OBP-heavy teams, we see a bit more action.
I don't want to read too much into these slopes, as the R-Squared values are still less than 1%. It looks like there are a couple high-leverage points on the SLG-heavy and OBP-heavy charts which may be skewing the data, but diversity in the lineup looks to be a bit of a plus for the extremes. However, in a neutral lineup, similarity has the slight edge, with a much larger sample size.
I highly doubt this is the best way to study this phenomenon, but now I have a start. While these models try to promote balance, the sheer amount of variability reigns supreme. To score runs, teams really don't have to get too caught up in power-OBP balance, as producing runs any way possible tends to work out well.
- What other forms of variability could be used for this study?
- Do you think this shows any sort of evidence, one way or another?