Can we improve SLG?

I’ve been listening to the audiobook for Keith Law’s Smart Baseball for the last week or so, and something dawned on me in the middle of Part 2 of the book. What if we could improve the SLG stat? What if we could add walks to the stat so that it would include a little bit of OBP in there? I’m proposing a new stat that I’ve called “true SLG” or “tSLG” with this formula:

tSLG

There are 2 things this does. It adds credit for walks and HBP to the stat, but it also slightly lessens the value for the total bases. Since traditional slugging percentage is just total bases over at-bats, by changing the denominator from ABs to PAs, it slightly lessens the value of the total bases by dividing by a larger value. The sabermetric community generally agrees on the fact that home runs and extra-base hits are overvalued in slugging percentage because the hardest thing that a player will do is get on base, so this tSLG stat moves in a direction that lessens that value, while giving more credit for getting on base.

I took this one step further and decided to make a new stat called “true OPS” or “tOPS” with this by adding OBP to my tSLG stat. By doing some simple math, you can see that it gives us a much clearer formula for tOPS than the one for OPS:

tSLG expanded

OBP

tOPS

If you look at this, it weighs a double at 1.5 times of a single, walk, or HBP. A triple is weighted at 2 times of that, and a HR is weighed at 2.5 times of a walk, single, of HBP. This is much closer to what is widely accepted as the weights for each of those outcomes. It’s not perfect, but it is much closer.

I wanted to check the validity of the stats, so I ran a correlation analysis on them vs. the number of runs a team scored for each of the last 10 seasons. You can find the math for the analysis here. The basics of it is this: the higher the number, the more the stats are connected. If the correlation is 1, then every time that the stat went up or down, the runs scored went up or down. If the correlation is .5, then only about half the time the runs went up or down, the stat went up or down. Here is the chart:

table

As you can see here, SLG and tSLG were both better indicators of the runs that a team would score than OBP for every season, and tSLG beat out SLG by at least a little bit in 7 of the 10 years. OPS and tOPS were still better indicators of how many runs the teams scored, with OPS beating out tOPS in 6 of the 10 years. I’d like to get a look at more than just the last 10 years, but it looks like tSLG out-performs SLG most of the time, while OPS beats out tOPS more often in the last decade, but it was close.

I don’t know how either stat would fare going back in other years. Due to limited time that I had to do this analysis, I only did the last decade. I also thing that tSLG would be a better indicator of individual performance, but didn’t have the time to do an individual correlation analysis. That may be coming later.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s