The Editor's Disclaimer HerEx intro Hardball home

Here's a Way to Get Relief from Rolaids Relief Formula

By Kenneth Broder
Los Angeles Herald Examiner
May 1, 1988

The 1988 edition of the Sports Encyclopedia of Baseball runs 619 pages. The Great American Stat Book is well over 500 pages and Elias Sports Bureau's Baseball Analyst is 437 pages. We're talking herniating mega-poundage here.

You want to know how Mariano Duncan hits with runners in scoring position in the late innings, who gave Fernando the most trouble last year, or who won the Chalmers Award (huh?) in the only four years it existed 1911-1914? They've got the answers.

If you want better ways to measure hitting, starting pitching, and fielding than the conventional, and arcane, methods generally used today, these professional stat freaks have an incredible array of techniques. They may not agree on the methods, or even the results, but they share a compelling need to quantify their analysis of a ballplayer with mathematical facts. The conventional wisdom divined from sporadic personal observation just won't cut it here.

But when it comes to relief pitching they seem content to just prescribe Rolaids -- the award formula, not the antacid. Unfortunately, the formula works about as well on relievers as the award sponsor's product does on Dodgerdog heartburn.

The Rolaids formula gives two points for each save and win, then subtracts the number of losses. Nothing else matters. Not blown save opportunities, games finished, the number of runs given up or the number of hits. Nothing. Middle relievers are pretty much ignored, unless they win about 30 games. And pitchers who rack up garbage saves, thanks to a generous definition of what a save is, are venerated.

Serious SABRmetricians (named after the Society for American Baseball Research) probably ignore relievers because they can't really show the mathematical relationship between what a reliever does and how it contributes to the scoring of runs (i.e. winning games). That's where amateurs like me can step into the breach and do our bit to add some much needed fuel to the Hot Stove League and perhaps inspire someone out there to come up with a truly scientific formula.

The Rolaids formula recognizes about one-third of what makes a good reliever -- wins, losses and saves. A formula that John Thorn and Pete Palmer use in the "Hidden Game of Baseball" goes in the exact opposite direction; all it measures is how many runs a reliever prevents compared to an average pitcher. Rolaids is totally dependent on the game situation, the Thorn-Palmer method is game independent.

A formula championed in these pages last week by Herald Examiner sportswriter Tom Singer goes to the heart of what's wrong with judging a reliever solely by the ERA. He suggests using Opponent Runners Average, which is the sum of hits, walks and hit batters allowed, divided by the number of innings pitched. That way, when Jesse Orosco comes in with the bases loaded and gives up a run-scoring single, he'll pay the statistical price even if those hitters don't score.

However, it seems like Jess should pay an even bigger price if he yields a bases-clearing triple. ORA won't exact that toll. What I really want to know about a reliever is how teams hit him. I want to know the totality of that production -- the extra-base hits, the stolen bases. I want it all. If I had it all, I could use a formula like Run Production to boil those eight or nine offensive statistics down to one single number representing the theoretical number of runs that much offense produces.

For instance. Let's say Orosco faces 50 hitters during a 10-game stretch. They get six singles, two doubles, a triple and two homers; five of them walk and two of them steal bases. By applying the Run Production formula to those numbers as if they were production of a single hitter, we can project that Orosco is responsible for 7.6 runs. (Bill James' Runs Created formula pegs is at about 7.8 runs.) And he did it while getting 27 batters out. That sounds like a line score from a game, and, in reality can be treated as such.

So if we're going to ignore game situations and just look at how Orosco got bombed this is the way to go. Unfortunately, you need a tone of raw data that only a privileged professional few have access to, and they're not stuffing the stats in my mailbox.

A possible substitute might be adding the On-Base average and Slugging Percentage of those hitters a pitcher faces. Studies of team and league records show a pretty high correlation between that combined figure and the ability to score runs. But even those numbers are only available to the layman annually.

And even if those stats were available during the season, I'm not sure they would be enough. The importance of a relief appearance, like pinch-hitting, is dependent on the game situation. It's hard to just ignore the Saves, Wins and Losses. So let's not. Let's toss all the these numbers in the pot and cook 'em till they're done.

As commissioner of the Hardball Rotisserie League, I've been using a formula that rates relief performance in eight categories on scales of 1-10 or 1-20. Because I don't have data on GamesFinished for past years, I've reduced the categories to seven for the purposes of this little study. The seven categories are: Games, Saves, ERA (all 20-point categories), Hits Per Inning, Walks Per Inning, Strikeouts Per Inning and Won-Loss Percentage (10 pointers). A perfect score is 100 points.

I determined the range of each category (e.g. what constitutes a 20 in ERA) by assembling a database of 191 top relievers over the past 30 years (and a few from primordial times) and analyzing the range of their numbers. The 191 pitchers were drawn from the top 50 Rolaids winners, the top 50 Palmer-Thorn ERA winners and just about everyone else who caught my during an eye-straining week with the Encyclopedia of Baseball.

Keep in mind that extensive relief pitching is primarily a modern phenomenon and that the rules for measuring what constitutes a Save have changed at least twice in the past 15 years that I'm aware of.


Sutter'77 Cubs73(10)1.35 (20)62 (12)31 (20)10769 (10)23 (10)129 (10)92

Hiller'73 Tigers105(9)1.44 (20)65 (14)38 (20)12569 (10)39 (8)124 (10)91
Hernandez'84 Tigers93(10)1.92 (18)80 (20)32 (16)14096 (10)36 (9)112 (8)91
Aker'66 A's84(9)1.99 (18)66 (16)32 (20)11381 (10)28 (10)68 (6)89
Kern'79 Rangers135(10)1.57 (20)71 (18)29 (16)14399 (10)62 (4)136 (9)87
Abernathy'67 Reds63(9)1.27 (20)70 (18)28 (16)10663 (10)41 (6)88 (8)87

Gossage'80 Yanks62(10)2.27 (16)64 (14)33 (20)9974 (10)37 (6)103 (10)86
McDaniel'60 Cards124(10)2.09 (16)65 (14)26 (16)11685 (10)24 (10)105 (9)85
Radatz'63 R. Sox156(10)1.98 (18)66 (16)25 (14)13294 (10)51 (6)162 (10)84
Wilhelm'64 W. Sox129(7)1.99 (18)73 (18)27 (14)13194 (10)30 (10)95 (7)84
McDaniel'70 Yanks95(8)2.01 (18)62 (12)29 (20)11288 (9)23 (10)81 (7)84

Sutter'84 Cards57(4)1.54 (20)71 (18)45 (20)112109 (5)23 (10)77 (6)83
Gossage'77 Pirates119(7)1.62 (20)72 (18)26 (12)13378 (10)49 (6)151 (10)83
Lyle'72 Yanks95(8)1.92 (18)59 (12)35 (20)10884 (9)29 (9)75 (7)83
Radatz'64 R. Sox169(8)2.29 (14)79 (20)29 (14)157103 (10)58 (6)181 (10)82
James'85 W. Sox87(6)2.13 (16)69 (16)32 (18)11090 (8)23 (10)88 (8)82

Sutter'79 Cubs66(6)2.23 (16)62 (12)37 (20)10167 (10)32 (8)110 (10)82
Miller'65 Orioles147(9)1.89 (18)67 (16)24 (12)11987 (10)32 (9)104 (8)82
Henke'87 Jays06(1)2.49 (14)72 (18)34 (20)9462 (10)25 (9)128 (10)82
Five pitchers tied for 20th place with 81 points.

* Numbers in parenthesis are points awarded on scale of 1-10 or 1-20. PTS is total of these numbers. 100 is perfect score.