Baseball BeatFebruary 08, 2004
Weapons of Mass Production
By Rich Lederer

Sabermetricians are guilty of developing too many Weapons of Mass Destruction. Too many stats. Too much confusion. It's time to get back to the basics. If we reduce the number of weapons (stats), then it follows we can reduce the amount of destruction (confusion).

Substitute Production for Destruction, and you've got Weapons of Mass Production. Production, in this case, is the original name for on base plus slugging. John Thorn and Pete Palmer of Total Baseball created the stat, shortening it to PRO. The authors also developed Production Plus (PRO+) before OPS+ was popularized by Production Plus, like OPS+ in later years, normalizes PRO (or OPS) to league average and adjusts for home-park factor. A mark of 100 is a league-average performance.

Ted Williams in his book Ted Williams' Hit List (which was written in 1996) gave "special credence" to PRO in ranking the greatest hitters of all time.

"We looked at various systems and methods, and we can't conceive of anything superior to this one. It is a simple statistic that is nonetheless as fair, as thorough, and as thought-out as any that has ever been used.

I realize that everyone has a different idea of what constiutes a great hitter. For some it's a high batting average. For others it's the guy with the most total hits--or home runs or RBIs. I've always believed that slugging percentage plus on base percentage is absolutely the best way to rate the hitters. This is something I've been talking about for a long, long time. To begin with, I've always felt that the bases on balls factor should be given more significance in rating a hitter's overall performance at the plate."

The beauty of PRO is its accuracy and simplicity. You don't need to know advanced math. You don't even need a calculator. You just add the two most important rate stats and bingo, you've got your number. I realize if you multiply rather than add these two percentages, you get a product that has an ever so slightly higher correlation with runs scored. There have also been some newfangled attempts to use a multiplier for the on-base average before adding this adjusted number to the slugging percentage.

Why complicate a formula that works just fine as is? I don't see the need to create additional methodologies unless they prove to be sufficiently more accurate to make up for their additional complexity. Wouldn't it be much easier for all of us to hold more intelligent discussions if we adopted batting average, on-base average, and slugging average as the three main rate stats when evaluating or comparing players? If we could agree on those three metrics, then wouldn't it also make sense to use PRO or OPS as a "quick and dirty" solo stat? Gosh, if we could come to terms with BA, OBP, SLG, and OPS as the basic core set of rate stats, wouldn't OPS+ be understood by the average fan in due time? If so, wouldn't that be a great first step for evaluating and comparing players by positions or within a league or from one era to another?

Counting stats would be a whole different matter. I'm not proposing that we throw out traditional stats such as hits, doubles, triples, and home runs but anything that would wean fans from a myopic focus on runs batted in would be a positive in my mind. How about times on base, total bases, and outs as the three core counting stats? If we could agree on those categories, then something like Runs Created could serve as the summary counting stat just as OPS would do the same on the rate side.

The value of walks and power as well as the scarcity of outs would all become better understood and appreciated by baseball fans, announcers, writers, and analysts alike. For those of us who wish to look into the numbers even further, the use of related stats such as isolated power and secondary average or developing "new frontier" stats involving defense, pitching, and baserunning makes more sense than affixing another label to the same can of alphabet soup.

By the way, a couple of tables for those who think OPS is dated and not an accurate measure of offensive production:

MODERN (1900-2003)

                              YEAR     OPS
1 Barry Bonds 2002 1.381 2 Babe Ruth 1920 1.379 3 Barry Bonds 2001 1.379 4 Babe Ruth 1921 1.359 5 Babe Ruth 1923 1.309 6 Ted Williams 1941 1.287 7 Barry Bonds 2003 1.278 8 Babe Ruth 1927 1.258 9 Ted Williams 1957 1.257 10 Babe Ruth 1926 1.253

Not a bad list, huh? When I see nothing but Barry Bonds, Babe Ruth, and Ted Williams, I am reminded that we're talking about pretty exclusive company here.

MODERN (1900-2003)

                              YEAR    OPS+
1 Barry Bonds 2002 275 2 Barry Bonds 2001 262 3 Babe Ruth 1920 255 4 Babe Ruth 1921 239 Babe Ruth 1923 239 6 Ted Williams 1941 235 7 Ted Williams 1957 233 8 Barry Bonds 2003 231 9 Babe Ruth 1926 227 10 Babe Ruth 1927 226

Adjusted or unadjusted. The monopoly of Bonds, Ruth, and Williams prevails.

One of the best features of OPS+ is the fact that it not only adjusts OPS for park effects but that it also standardizes it versus the league average. A player with an OPS+ of 120 is 20% above the league average OPS+. An OPS+ of 80 is 20% below the league average OPS+. [Editor's added note: OPS+ = 100 * ((OBP/lgOBP*) + (SLG/lgSLG*) - 1)]

Thanks to the Sabermetric Baseball Encyclopedia, we can standardize all stats to the league average. The following is an example of the first table presented above with OPS expressed as a ratio versus the league average.

MODERN (1900-2003)

                              YEAR     RATE   PLAYER   LEAGUE
1 Babe Ruth 1920 182 1.379 .757 2 Barry Bonds 2002 181 1.381 .763 3 Barry Bonds 2001 177 1.379 .781 4 Babe Ruth 1921 173 1.359 .786 5 Babe Ruth 1923 172 1.309 .761 6 Ted Williams 1957 171 1.257 .733 7 Ted Williams 1941 170 1.287 .758 8 Barry Bonds 2003 166 1.278 .772 9 Babe Ruth 1926 163 1.253 .768 10 Babe Ruth 1927 163 1.258 .773

The order is slightly different than the one not standardized to the league average, but the names remain the same.

Weapons of Mass Production. Batting average, on base average, slugging average, and OPS. Times on base, total bases, outs, and runs created. Four by four (and, thank goodness, not the rotisserie syle). Four rate stats and four counting stats.

These stats can be expressed in absolute terms for simplicity. They can be standardized to the league and positional average for comparative purposes. Or they can be normalized for home-park factors when necessary.

There are other worthwhile measurement tools for sure. But many of them are better applied to scouting and player development than evaluating actual performance.

The bottom line is that we should concentrate on reducing rather than expanding the batting metrics available to us. Having said that, I support the sabermetric community in its efforts to quantify defensive value, and I think there is still much more work ahead of us when it comes to computing baserunning as a separate area of performance.

Sources: Sabermetric Baseball Encyclopedia and