Tuesday, July 18, 2017

THE GAME YOU SEE: A QUICK HISTORICAL PERSPECTIVE USING HRs AS A PRISM

So--we know that home run levels are way up, even in the context of recent history. Let's get beyond the HR/G formulation, however, and quantify this on a game-by-game basis.

In other words...in any given season, based on the HR/G average, how many games are played in which a team hits no HRs? Exactly 1 HR? 2 HRs? 3 HRs? More than 3 HRs? And what are the ancillary statistics accompanying these games?

We're just taking a quick look at this data, using four seasons for a basis of comparison. To match up with 2016 (with 1.16 HR/G, highest full season average yet...), we reach back to 1933, a "lull year" in the live ball era's early offensive explosion, where runs/game dipped to 4.48--which matches the R/G average in 2016. 1933 differs from 2016 radically in terms of the shape of XBH levels--particularly in HR/G (0.44 per game).

To round things out, we chose a couple of seasons roughly at the mid-point historically between these two seasons. 1976 is another "lull" year during a period of offensive resurgence, where run scoring dipped to just under four runs per game and HR/G dropped from the .75-.85 per game down to .58. 1970 is often mischaracterized as a "boom year" in offense, which is true in the immediate context of the times; but it's 4.34 R/G is not particularly robust. What is noteworthy, particularly for this quick snapshot, is that the year's HR/G rate was elevated (0.88/G) relative to run scoring and signals a return to the pre-strike zone-change offensive shape found in 1955-62.

So, with that, let's look at the accompanying data tables to get a sense of what a "game-level view" of things can tell us:












Looking at the 1933-2016 comparison, we can see that last year's teams went homerless in only a bit more than half of the number of games that such was the case in 1933. Conversely, they were six times more likely to hit 3 or more HRs in a game.

The averages for the two seasons from the 1970s fill in the blanks and demonstrate that these types of distributions are more or less linear in nature.

In terms of overall performance (R/G and team WPCT in the various game categories), we can use another way of measuring the data to evaluate the differences and similarities that exist. Here are the HR distributions and the R/G data displayed as percentages:


So, in 2016, homerless games accounted for just over a third of all games, as opposed to two-thirds of all games in 1933. Teams in 2016 aren't able to score runs relative to overall offense when they don't hit HRs--run-scoring levels were only 61% of average in such games last year, whereas they were 83% of average in 1933. Offensive strategies were aligned to other ways of scoring runs than hitting homers for much of the first half of baseball history, and they were able to win a lot more games without hitting homers (.435 WPCT in homerless games in 1933; .321 WPCT last year).

Interestingly, run-scoring levels relative to average track reasonably well with WPCT based on the number of HR/G. Today, hitting a HR in a game gives a team less than a 50-50 chance of winning.

With HR/G levels pushing even higher than 2016's average, those who go to the ballpark are getting closer to having a 75% shot of seeing someone hit a HR. Or should, we say hit at least one HR.

The "invasive techniques" discussed in the posts below will not take us back to 1933 or 1976 in terms of HR/G. We might get back to something like the levels in 1970, however, which would make it possible for teams that eschew home run hitting to still have a chance to be competitive--which is definitely not the case now. And that's the ugly specter of two-dimensionality that has slowly taken over baseball over the past thirty years. It's time to get all of our dimensions back...