Tipping Pitches: Sports: In Search of Baseball's Statistical Holy Grail


Tuesday, February 9, 2010

Sports: In Search of Baseball's Statistical Holy Grail

email to friend edit
A while back, I engaged in a study to compare player stats -- Hall of Famers and non-Hall of Famers -- as part of a project for my eight year-old son. The goal? Strip away the names and determine which players deserved to be inducted.

While this was a rather complex project (you can download the 57-page "test" here), I relied entirely on rather standard statistics. For each player, I provided the following information:

Number of seasons played
Year started and finished
Number of All-Star Games
Number of MVP awards
Total plate appearances
Total runs
Total hits
Total home runs
Total RBI
Total stolen bases
Lifetime Batting Average
Lifetime OBP
Lifetime Slugging Percentage
Lifetime OPS

Additionally, I provided the league average BA, OBP, SLG and OPS during that player's era as well as the ratio of his statistic over that league average.

Simple enough? Not really. But made for an interesting project.

But what if I wanted to expand this research to be more than just a third grader's project? While these stats are generally considered the accepted method for analyzing performance, they certainly aren't the be all, end all. And we want to assess performance as accurately as possible.

Each offensive stat has its own flaw. Runs scored and RBI are partially reliant on teammates driving the player in (runs) or being on base and running the bases well (RBI). Total hits can be deceiving as well. In some eras, piling on hits was easier than in others. Likewise, a nice total can be accumulated in a very good, though not great, and long career. And great, though short, careers will lack hit totals. As we know all too well, home runs were common during the Steroid Era. Yet, just before Babe Ruth arrived, the average team hit fewer than 20 home runs per season.

That said, you can't rely entirely on qualitative stats either. What good is a great AVG, OBP, SLG or OPS if the career was five seasons long? And how do you know if any of those stats are great without being relative to their peers? Stats can be deceiving. The average batting average was .239 in 1908, .269 in 1912, and .291 in 1921. Many played through all three seasons, and the definition of a "good" season each year was comparatively different.

This is why I used ratios of the player's qualitative stat to the league average. Yet, even that made me a bit uncomfortable. The top and bottom ranges are not equal for each stat, so the way we evaluate the percentage for each would need to be different. It's an inexact science.

Regardless, not one of those qualitative stats properly encapsulates overall value. In other words, what single statistic can we use to measure one player over another? The BA and OBP statistics don't consider power. And none of the statistics consider speed or ability to run the bases.

The reason I wanted one stat is that when you start focusing on multiple stats you inevitably begin favoring a particular type of player. And you only use multiple stats because no one stat does the job adequately. So, is there a single statistic that can do the job of many?

I put out an APB on Twitter.

Question for my baseball peeps: In your opinion, which single stat best reflects all-around offensive value?

I quickly received several responses.

@heelsonthefield: OPS
@theMercyRule: The simple, unsexy run.
@HrzMatt: OBP probably
@WisconsinRob: If there had to be one, it's probably OPS.
@GregFertel: wOBA or wRC+ encapsulate total offensive value the best.
@n8van2: OPS+
@CrashburnAlley: wOBA

Seven people, six surprisingly different answers. I'll say that I'm not sold on OPS. I like how it combines OBP and SLG, but it seems so inexact to simply add them together. But I admit, I often use it as the stat to evaluate talent.

I was particularly intrigued by the answers of wOBA, wRC+ and OPS+. I'm a stat nerd, but admit that I know far too little about these and other advanced statistics. There are other offensive statistics, such as EQA and VORP, that may also be the magic stat I'm looking for.

Of course, I don't completely understand any of these statistics. And as I've researched them on FanGraphs, I've found that it isn't as easy as plugging the peripheral statistics into a formula to determine these values back to the 1800s. I have to rely on whatever data sites like FanGraphs have. Many of these stats are determined slightly differently from season to season, depending on a baseline (at least, as far as I know). And not all necessary data is available to calculate these values for all of the players I want to evaluate.

And so, I appeal to you just as I appealed to my Twitter followers. What single stat would you use to evaluate offensive worth of a player? This stat needs to be all-encompassing. Power, hitting ability, runs scored. It appears that wOBA and wRC+ do this. But do we have the necessary data for all of the players I want to evaluate?

Maybe there is more than one stat would be fine for this purpose, but there cannot be overlap between the two or three stats. Doesn't seem likely or possible.

I guess what I'm doing is no different than what others before me have done. I want to do two things: 1) compare players within eras, and 2) compare players between eras. I want to determine who truly deserves Hall of Fame induction, at least based on offensive performance only.

I know. I'm chasing the Holy Grail.

What do you think? Is there a single stat you would or could use to accomplish this task? What would it be?


Max on February 09, 2010 said...

I think you're crazy. Go for it.

Jon Loomer on February 09, 2010 said...

Max, you've gotta have a touch of crazy in you to do this. Thanks for reading.

Post a Comment