Home
Column 1 2 3 5 or 6


Going to Bat for the GamePerformance Stat

By Kenneth Broder
Los Angeles Herald Examiner
April 24, 1988

On Aug. 20, 1974, Nolan Ryan pitched a classic Nolan Ryan game.

He battled through 11 tough innings, striking out 19 batters while yielding only four hits, five walks and one measly run. It was one run too many and Ryan lost a heartbreaker.

Ryan has lost a lot of heartbreakers in his lengthy career and it can take all the Hot Stove League fun out of talking about the man's accomplishments if you're constantly running up against his so-so won-loss record.

Whether or not you believe, as some do, that Ryan's reputation is as overblown as his vaunted strikeout totals and skimpy ERA (thanks to pitching in tough-to-hit-in ballparks) you've got to admit the man puts on an impressive show.

If you're going to appreciate a Nolan Ryan game, you can't worry about whether he wins or loses. You just climb aboard the Ryan Express, strap yourself in for a 98-mph hell-ride and after it's over figure out Nolan's GamePerformance stat.

GamePerformance, or GP, is a statistic inspired by Bill James that rates a pitcher's one-game performance on a scale of 1-100. A pitcher can score over 100 points, but he either has to strike out more than 19 players in a game (the major league record) or pitch into extra innings.

Briefly, this is how it works. Start by giving the pitcher 54 points (100 minus 27 outs and 19 strikeouts). Add one point for each batter he retires and one for each strikeout. Subtract two points for each hit yielded and one for each walk. Subtract four points for each earned run and two for each unearned run. The result is GamePerformance.

Ryan's 1974 losing effort was good for a very impressive 89 points. (No demerits for losing.) That's better than any Angel or Dodger performance this year in games through Thursday.

Best Pitched Games '88

Angels

DateNameIPHRERBBSOGP*

4/06Petry6.04001568--
4/15Witt6.06312564W
4/13Finley8.05223764W
4/07Finley6.35212560L
4/09Witt7.07222354--
4/10Fraser6.04334553W
4/17McCaskill7.07332249W
4/21Fraser6.38522548W
4/18Petry7.07334348--
4/04Witt6.04552447L
4/08McCaskill5.36443241L
4/12Petry5.06442140W
4/16Fraser5.66444140--
4/19Finley7.66774538L
4/20Witt5.010555630--

Dodgers

DateNameIPHRERBBSOGP*

4/18Leary9.030021184W
4/05Hershiser9.03002679W
4/15Hershiser9.06220566W
4/10Hershiser8.35112366W
4/17Sutton6.05220559L
4/13Leary5.04220659--
4/09Valenzuela8.06332355W
4/07Sutton5.37212452--
4/14Valenzuela8.09224351L
4/08Leary5.05330350W
4/16Belcher7.09421347W
4/04Valenzuela6.07544440L
4/12Sutton4.67441037L
* GamePerformance on scale of 1-100

Some of Ryan's more impressive winning performances include a 16-strikeout, one-hit shutout on July 9, 1972, that was good for 94 points and a 17-strikeout no-hitter on July 15, 1973, that also rated a 94.

There's no denying GamePerformance is a garbage stat. It's the kind of number that amateurs like me mess around with because it's, well, easy to understand and fun. But it really doesn't measure performance as it contributes to victory. And that is what SABRmetrics is all about (SABR is Society for American Baseball Research.)

A formula like Run Production for hitters, which sucks up eight or nine hitting attributes and spits out a stat that measures a player's theoretical contribution to his team's run total, is a real measure of performance.

A team wins games by scoring more runs than its opponent. It doesn't get a W for collecting the most hits. It doesn't win by just hitting the most homers, stealing the most bases, striking out less, or having the grittiest fastball pitcher on the face of the earth. These things contribute to winning, but you can't rate any player by just one of these stats, unless you don't care about losing.

It's aggravating to pick up the Sunday newspaper and see hitters in the American and National leagues listed in order of batting average. The correlation between a team's batting average and its ability to score runs (and win games) is fairly low. Last year, the St. Louis Cardinals were sixth in batting average, but second in the NL in scoring runs. In the AL, the Tigers were fourth in battling average but first in scoring runs.

However, if you ask a die-hard Angels fan what kind of year Bill Buckner had in '87, invariably he'll point to Buckner's .287 average and deem it a success. Never mind that he walked just 22 times in 491 plate appearances and barely reached base 31 percent of the time. Ah, but I digress.

The point is that most player stats we use are garbage and rarely address the questions we really want answered: Is this guy a winner or not? Is he better than some other guy?

Pitching stats are notoriously less efficient than hitting formulas in answering just these kinds of questions, but when you're commissioner of a baseball rotisserie league, as I am, you'd better at least have something to bluff your members with. My approach doesn't take into consideration park influence, quality starts, hard luck losses or lack of hitting support.

Not so simply put, Pitstat is a combination of two other statistics, Pitching Runs (PR) and Won-Loss (W/L). PR takes a pitcher's ERA, compares it to the league average, factors in how many innings he pitched, and is expressed as a percentage above or below the average pitcher.

W/L works the same way with won-loss percentages, but gives extra weight to a player's number of decisions. W/L and PR are averaged together to produce Pitstat, a rating of how much better than the average pitcher a player is.

Interestingly, the league generally thought of as having the big dominant pitchers (the Goodens, Valenzuelas, Scotts, Sutcliffes, etc.) showed far more balance than the AL. (See Season's Best '87 chart). The man who would be Cy Young, Nolan Ryan, ranked 28th.

Season's Best '87

American League

NameWonLossERAIPERW/LPRPitstat*

Clemens2092.97281.69355.047.256.5
Key1782.76261.08045.061.553.2
Viola17102.90251.6813554.444.7
Saberhagen18103.36257.09640.039.139.5
Morris18113.38266.01003539.637.3
Stewart20133.69261.310735.028.031.5
Alexander14103.01206.06920.041.230.6
Liebrandt16113.41240.39125.035.030.0
Herrera18104.01251.611240.015.827.9
Langston19133.84272.011630.023.426.7

National League

NameWonLossERAIPERW/LPRPitstat*

Gooden1573.21179.66440.023.731.9
Welch1593.22251.69030.032.831.4
Dunne1363.03163.35535.025.930.5
Sutcliffe18103.68237.39740.014.427.2
Reuschel1393.09227.07820.033.927.0
Martinez1143.30144.65335.017.126.0
Scott16133.24247.68915.031.723.3
Tudor1023.8496.04140.03.421.7
Hershiser16163.06264.6900.040.820.4
Downs1293.63186.07515.012.713.8
* Percentage better than average pitcher in the league