How are the Red Sox performing in 2018? We found it difficult to get a quick overview, so we created a few graphs that allows me to assess what is happening over time.
The post is about sports, but the same concepts can be applied to more traditional business visualizations.
We start with the underlying theory and then move on to the Red Sox.
PERFORMANCE VERSUS RESULTS
Performance is not the same as results. The results can be read anywhere. But performance is a different matter. The two are highly correlated, but not the same.
We created a performance score by transforming the runs scored for and against in each game.
The properties of the curve are:
- The utility of additional runs scored declines. It is more important to go from 0 to 1 run, than from 8 to 9 runs. The flipside applies to runs against.
- The likelihood of 0 runs scored (around 5%) should be equal to the likelihood of scoring X runs or more. X is around 10 runs and hence the flat curve from 10 and above. That is, scoring 0 runs is as common as scoring 10 or more runs.
- The curve has to be symmetrical across all MLB teams; a run scored gives the same performance as a run against (with a minus sign). This determines where the curves cross 0 performance in a season. It is usually a bit over 4 runs. Put differently: 4 runs should be seen as average performance.
- Performance goes from -1 to +1. -1 is bad performance, 0 is average, and +1 is excellent performance. (We tried 0 to 1 and other scales but these metrics made the graphs hard to interpret.).
RED SOX 2018 IN GRAPHS
Based on the curve above we converted each game’s score into performance numbers. We discovered that the best way to visualize performance was to look at cumulative numbers. The idea for this is from cricket where the score within a game is often graphed cumulatively. We apply it across games.
Here is one way to graph the cumulative results by game 77 (the last game before this post):
Ideally, there should be a steady upward movement. We see this for the Red Sox defense, while the offense is off lately.
Our preferred way to visually this information is shown below. It shows exactly the same data, but offense and defense is plotted on the x and y axes (instead of over time).
There should be steady progress along the diagonal. This indicates a well-balanced, winning team. If the plotpoints are above the diagonal, defense outperforms; below the diagonal, offense outperforms. So far the Red Sox are doing well, with some relative weakness in offense (but still positive).
There are roughly four eras in the graphs:
- Game 1 to 20 showed rapid and balanced progress
- Games 20 to 45 had stagnation had erratic offensive progress offensively and negative defensive performance
- Games 45 to 62 had steady performance
- Games 62 to 77 had strong defensive performance and poor offensive performance
Note that teams can move in any direction in the graph, including along the negative axes.
RED SOX’ PERFORMANCE BY YEAR
How does 2018 year-to-date stack up against earlier years? We looked at 2008 to 2017 performance at game 77 scaling the cumulative performance by game 162 back to 77 (to avoid arbitrary ups and downs during the seasons).
So far, 2018 is by far the best defensive year, while the offensive performance is middling. The total performance (sum of offense and defense, not shown) is the best over the period.
Note the steady defensive progress since 2014. This is surely part of a strategy.
The Red Sox are off to an excellent start this year, but the offense has stalled lately despite yesterday’s 14-10 win.
Looking at performance broken into offense and defense, rather than studying results (wins and losses), yields insights into which part of the team is contributing to success.
Cumulative performance gives a steadier and more strategic perspective on what is going on in the midst of homeruns, strikeouts, walks, injuries, and other tactical metrics.
The analysis and graphs can be extended to other variables such as hits, pitch count, errors, and more, although most of those variables will not give additional strategic insights. Payroll should be included.
Underlying the graphs above is a dataset covering every MLB team from 2008 and onwards. Here are a few additional graphs.
This is the infamous fried chicken year of 2011.
In contrast, the 2016 regular season had a really strong second half starting with game 84.
Finally, an almost perfect regular season: 2008.
TEST FOR LIVE CHART BELOW