ball four

After reading my recent post about World Series competitiveness, a friend of mine brought up the recent slide in World Series viewership. I thought it would be interesting to look into this, so I did; below are my findings.

As we can see, viewership has been on the decline for the past eleven years - aside from 2004 and 2009. 2004 and 2009 were years in which the Yankees and the Red Sox played in the World Series - inevitably providing a TV ratings boost - with 2004 being the year the Red Sox had a chance to break the Curse of the Bambino. 

I went ahead and mapped the TV ratings onto the competitiveness chart that I put together for the previous post. 

There seems to be no relationship between the competitiveness factor - total aLI - and TV ratings. For example, despite being the most competitive series of the past ten years, the Cardinals and Rangers series garnered the third-worst TV ratings in that span. Furthermore, a regression analysis of competitiveness factor on TV ratings yields a very low r-squared. This makes sense for a couple of reasons: (i) the competitiveness factor does not include compelling story lines, such as the 2004 Red Sox, and (ii) competitive series between two small or mid-market teams can be under-viewed when compared to the competitiveness factor, because of the smaller nature of their fan bases - on the flip side, a series between two large-market teams can be over-viewed when compared to the competitiveness factor.

Since 2004, batters have been seeing fewer and fewer pitches in the strike zone. According to Zone% - which indicates the overall percentage of pitches a batter sees inside the strike zone - the average percentage of pitches seen within the strike zone has fallen from 55.1% in 2004, to 45.3% in 2011 -  a 17.7% decline in pitches seen inside the strike zone. 

Hitters have adapted; according to O-Swing% - which indicates the percentage of pitches a batter swings at outside the strike zone - hitters are swinging at more pitches outside of the strike zone than they used to. In 2004, the average O-Swing% was 16.6%, while that rate now stands at 30.6% - an 84.3% increase in pitches swung at outside of the strike zone.

Another way to look at this is by the composition of Swing%. Swing% represents the overall percentage of pitches a batter swings at. In order to get a better look at the change in O-Swing% since 2004, we can break Swing% into the proportion of swings made on pitches outside of the zone (O-Zone% * O-Swing%) and the proportion of swings made on pitches inside of the zone (Zone% * Z-Swing%). For example, in 2004, approximately 17.6% of swings were made on pitches outside of the strike zone. That number balloons to 36.2% in 2011. As you can see below, the share of swings made on pitches outside of the zone has grown over the course of the past eight years.

It’s interesting that hitters have adapted as they have. With the increase in the proportion of pitches thrown outside of the zone - almost 55% of pitches were thrown outside of the zone in 2011 - one would expect hitters to be more patient at the plate, resulting in an increase in league average BB%. However, this is not the case: BB% has fluctuated between 8.1% and 8.9% over the past eight years, reaching its low of 8.1% this year - a year in which nearly 55% of pitches were thrown outside of the zone.

The change in Swing% composition does seem to have affected hitters’ overall offensive production. A regression analysis of O-Swing% on wOBA results in an r-squared of .34, which means that 34% of the variance in wOBA can be explained by O-Swing%. The regression equation is

wOBA = .342 - .063*O-Swing%

with a p-value of .077 on the O-Swing% coefficient. So the more frequently a hitter swings at outside offerings, the lower his wOBA will be, which is in accordance with intuition - the more frequently a hitter swings at outside offerings, the weaker the contact he makes, consequently deflating wOBA.

Furthermore, the change in Zone% also seems to have affected offensive production. A regression analysis of O-Zone% on wOBA results in an r-squared of .46, which means that 46% of the variance in wOBA can be explained by O-Zone%. The regression equation is

wOBA = .376 - .101*O-Zone%

with a p-value of .030 on the O-Zone% coefficient. So the more frequently a batter faces pitches outside of the zone, the lower his wOBA will be. This result is also intuitive - the fewer pitches a batter receives in the zone, the less frequently can he hit with power, consequently deflating wOBA. While hitteres could adjust by taking more walks, the net change in wOBA is still negative, (it takes multiple walks to negate one less extra base hit).

Finally, I tried regressing O-Zone%*O-Swing% on wOBA to see how the trends in Zone% and Swing% have affected offensive production. The regression resulted in an r-squared of .41. The regression equation is

wOBA = .339 - .106*(O-Zone%*O-Swing%)

with a p-value of .043 on the O-Zone% coefficient. So the more frequently a batter faces pitches outside of the zone and swings at these pitches, the lower his wOBA will be. Similar to the past two results, this also makes intuitive sense - the more frequently a batter receives pitches outside of the zone and swings at them, the weaker the contact he makes on pitches, consequently deflating wOBA.

These trends can partially explain the lower league-wide offensive numbers. Pitchers have been changing their approach to hitters, and hitters seem to have responded by taking more hacks on outside offerings, ultimately negatively affecting their offensive production. It will be interesting to track how these two trends progress over the next couple of years. Will pitchers continue to throw more and more offerings outside the zone? When will hitters start laying off of these pitches at a high enough rate to make pitchers go back to throwing inside the zone? We’ll find out eventually.

This year’s World Series has been referred to by many as one of the best World Series in recent memory. While the claim was pretty apparent to anyone who watched it, it led me to consider the competitiveness of recent World Series. In order to get a historical perspective, I began by looking at the average number of games played within each World Series by decade over the past 100+ years.

For four straight decades, between 1940 and 1979, the average number of games played in the World Series was above 6 per World Series, by far the highest of any decade or stretch of decades. In the last four decades, average World Series length has fallen each decade from a high of 6.100 in the 70s, to 5.900 in the 80s, to 5.556 in the 90s, and finally, to 5.417 in the 2000s. A decline in average World Series length could signify a decrease in World Series competitiveness.

I then looked at leverage index (LI) which is a measure of the importance of a particular situation in a game. Average LI is 1, and is considered a neutral situation. An LI greater than 1 signifies a more important situation than an LI of 1, and an LI of less than 1 signifies a less important situation that an LI of 1. I used LI as a proxy for competitiveness. By looking at the average of the average LI (aLI) of each game in the series, we can get the average importance of each situation in a given World Series. 

The average aLI of all of the World Series games played between 2002 and 2011* is 1.023, or approximately 1, which is neutral. Based on the above figure, six of the ten World Series had an average aLI of less than 1, two had an average aLI of slightly greater than 1, and two had an average aLI significantly greater than 1. According to my average aLI method, the 2005 World Series between the Houston Astros and the Chicago White Sox had the highest average importance per situation, (aLI = 1.560), followed by this year’s World Series, (aLI = 1.201).

The final analysis puts the variables behind the two prior analyses together. I graphed the average aLI and series length by World Series to see how the two variables mapped onto each other.

While the most competitive series would be a long series with an average aLI above 1, a shorter series with a high average aLI could be more competitive than a long series with a low average aLI. Since both series length and average aLI point to the competitiveness of a World Series, I wanted a way to incorporate both factors into one final measure of competitiveness. Since the two variables are independent of each other, I ended up multiplying the series length by the average aLI of the series to get the total aLI of a World Series. The higher the total aLI, the more competitive the series. Here are the results.

It’s quite apparent that the 2011 World Series ranks highest on the competitiveness scale. Not only did it have a relatively high aLI, but it also lasted seven games - seven games worth of high importance situations makes for a very competitive and compelling World Series. The 2002 - Giants vs. Angels - and 2003 World Series - Marlins vs. Yankees - were next in line in terms of competitiveness, with the 2004 World Series - White Sox vs. Astros - now placing fourth as opposed to first in my previous average aLI only analysis.

While aLI and series length are by no means perfect proxies for the competitiveness of a series, they are a good place to start. Further analysis could look at factors such as the number of lead changes, the number of games decided after the seventh inning, etc., to get to the heart of the issue of measuring competitiveness.

——————————————————————————————————————

*I limited the aLI analysis because I only had aLI data for World Series games between 2002 and 2011.