samples index

 

baseball index

 

 

 

Part 3. OPS or OOPS? 

The most widely used of all-in-one nouveau statistics is OPS. OPS means on base percentage plus slugging percentage. Like most formulas, OPS excludes defense. 

In August 2002 I was commissioned by the proprietor of a Web site devoted to fantasy sports and gambling to write an article “blasting OPS as a measure of a baseball player's skill.” I had not previously analyzed the formula nor even applied common sense to it. The creeping popularity of this non-mathematically-based “statistic” caused me to accept it as a loose, isolated gauge of how players performed in the batters’ box. 

When I considered the two components of the formula, I realized that OPS is fatally flawed in two ways. First, it includes slugging percentage, which is already a formula. It is not a percentage. If Russell Branyon hits a home run, a double, and strikes out three times, his slugging percentage is 1.200. Maybe that’s what giving 110% means. Progenitors of nouveau statistics often accept their own formulas as facts and then base other formulas on them. Ms. vos Savant would tell us that each of those formulas is flawed because they are disproved by exceptions. More importantly, she would remind us that you can’t build facts from theories. 

Second, both on base percentage and slugging percentage include 100% of batting average. Thus, OPS counts batting average twice. If a player gets as many walks as hits (which almost never happens) and his total of extra bases from hits equals his total of hits (not much more likely than Rich Garces winning the Boston marathon), then OPS would be weighted 50% BA, 25% BB, and 25% power. Because hits are more common than walks and singles are more common than extra base hits, OPS is typically 70-80% batting average. That means OPS is similar to BA, as shown in Chart 12. 

Chart 12. 2003 BA and OPS Leaders 

BA rank  

Player BA OPS rank

OPS rank

Player OPS BA

1

A. Pujols

.359

2

1

B. Bonds

1.278

.341

2

T. Helton

.358

3

 

2

A. Pujols

1.106

.359

3

B. Bonds

.341

1

 

3

T. Helton

1.088

.358

4

E. Renteria

.330

41

 

4

G. Sheffield

1.023

.330

4

G. Sheffield

.330

4 5 C, Delgado 1.019 .302

6

B. Mueller

.326

68

 

6

M. Ramirez

1.014

.325

7

J. Kendall

.325

68

 

7

J. Edmonds

1.002

.275

7

M. Ramirez

.325

6

 

8

A. Rodriguez

.995

.298

9

D. Jeter

.324

54

 

9

T. Nixon

.975

.306

10

M. Ordonez

.317

20

 

10

D. Ortiz

.961

.288

10

V. Wells

.317

29

 

11

J. Thome

.958

.266

12

M. Giles

.316

24

 

12

R. Hidalgo

.957

.309

13

G. Anderson

.315

36

 

13

F. Thomas

.952

.267

Some people think adding numbers is arbitrary. They prefer on base percentage times slugging percentage. The rankings created by that formula are similar to OPS, but the distinctions between players are exaggerated. 

Another variation on the OPS theme is adding the two “percentages,” then subtracting batting average (or adding on base percentage and secondary average). That produces a more accurate measure of offensive performance, but it still omits runs scored, runs batted in, stolen bases, and other traditional measures of performance. 

If you must use an all-in-one formula to measure offense, run production (runs plus runs batted in) covers all the bases. Defenders of OPS would argue that runs and runs batted in are team dependent. Most statistics, including hits and walks, are team dependent. While a lineup has an effect on a player’s ability to get hits, walks, runs, and runs batted in, a leadoff hitter on one team would probably be a leadoff hitter on another team and a cleanup hitter on one team would hit in the heart of the order on another team. High-RBI sluggers Joe Carter and Fred McGriff changed uniforms as often as fashion models do, but drove in runs everywhere. Variations in runs and runs batted in have far more to do with the player than the team. Walks are equally team-dependent, and have far less outcome on the score. A perfect example is Brian Giles on the 2002 Pirates. The Pirates offense went overboard and Giles was a whale in that sea. Rob Mackowiak was second on the team with 12 home runs. Jason Kendall was second on the team with a .284 batting average. Pokey Reese was third at .259. Pitchers rarely had any reason to throw strikes to Giles. As a result, he walked 135 times. His walks increased 50% from the 90 he had in seven more games in 2001! 

While Giles clobbered right-handed pitching, he had a radical platoon differential: 

Chart 13. Brian Giles 2002 splits

BA OBP

SP

vs. LHP

.231

.355

.476

vs. RHP

.325

.485

.681

What those numbers illustrate is that if Giles had been on a good team, he would have faced a lot more lefties and a lot more strikes in certain situations. If you’re leading 7-2 in the seventh inning, you don’t need to bother to bring in a lefty to face Giles. If the score is close, you pitch around him or walk him. He received 24 intentional walks. Because the score usually wasn’t close, pitchers had little to lose by challenging Giles. Giles had a fine on base percentage because he’s a very good hitter, but it would have been much lower if he played on a good-hitting team. 

The league MVPs last year were an embarrassment to baseball. Barry Bonds won in the National League with 90 RBI even though the batting champion played for a contender and drove in 38% more runs. The American League MVP was a slugger on a last place team who failed to hit .300! 

If you put Bonds on the 2003 Red Sox, he might have walked half as often. He would have hit fourth, between Nomar Garciaparra and Manny Ramirez, and would have been surrounded by Mueller, Millar, Nixon, Ortiz, and Varitek. Pitchers would have pitched to Bonds. He would have had to swing the bat much more often and with much less selectivity, which would have lowered his batting average. 

Because intentional walks are more team-dependent than anything else in baseball, walks are the most team-dependent statistic. Hitters don’t even have to be good to receive intentional walks; eighth-place hitters in the NL get lots of them. Rey Ordonez had 64. Royce Clayton had 10 last year, which is more stupefying than Barry Bonds’s 61. With the addition Sanders, Simon, Lofton, and Stairs, and the resurgence of Ramirez and Kendall, Giles intentional walks dropped in half. Given that intentional walks are included in on-base percentage, on-base percentage is probably more team-dependent than runs or runs batted in. Last Dmitri Young was a good hitter on a woeful club, yet his runs batted in are only a little below what would be average for a guy with his batting average and power stats.  

Your little league coach probably told you that a walk was as good as a hit. That may have been true for me or Bob Buhl, but it’s obviously not true for any major league hitters. Who could argue that a walk is as good as an extra base hit? In many instances, a walk is as good as a single. With two outs and a runner on second base, a walk is never as good as a hit. A formula that includes walks, but excludes runs batted in cannot be an accurate measurement of anything that matters in baseball. 

Dusty Baker agrees. “It's called hitting; it's not called walking,” Baker says. "Walks help, but you aren't going to walk across the plate,” he believes. 

If OPS is more important than run production, why isn’t it used for pitchers instead of earned run average? As earned run average depends on defense and relievers, it is also very team-dependent.

Ultimately, OPS shares the same flaw as other all-in-one nouveau statistics: it needs context. Most nouveau statistics include some contexts while excluding others. All statistics need to be viewed in context of everything. Did you know 86% of all statistics are made up on the spot? 

Nouveau statistics or rankings should not clash with empirical data and should tell us something we can’t discern from careful analysis and interpretation of conventional statistics. If traditional statistics are flawed, statistics based on them are also flawed. Rearrangement of conventional statistics usually produces entertaining results that mean slightly less than the raw data. At best, nouveau statistics confirm the validity of conventional statistics.

 

 

samples index

 

baseball index