The Editor's Disclaimer HerEx intro Hardball home

You'll have to wait for winner of this debate

By Kenneth Broder
Los Angeles Herald Examiner
October 16, 1988

Moments after Orel Hershiser paid homage on bended knee to the Big Dodger in the Sky Wednesday, an overenthusiastic fan -- and alleged friend -- screamed at me: "What do your damn numbers say now?"

His sarcasm was duly noted and ascribed in part to last week's negative statistical assessment here of the underdog Dodger offense.

But mostly he was ticked off by my lack of partisan blind faith and what he thought was my season-long basic premise -- that with the right statistics, we can take the guesswork out of baseball strategy and accurately predict winners and losers.

Yes, the numbers did seem to favor the Mets in the playoffs, but you don't need a graduate degree in the New Statistics to know that. In fact, the question baseball fans seemed most to be wrestling with was whether the Mets should be favored to win it in five games or six. Although diehard Dodger rooters refused to capitulate, they too acknowledged some pretty stiff odds.

They needn't have been so pessimistic. Five year ago, stat guru Bill James introduced a handy little formula called Logs5 that predicted the outcome of specific matchups: hitter vs. pitcher or team vs. team. And if I understand the theory correctly (sorry, no guarantee here), it says anything can happen in a short series, but if you have to bet, bet on it going seven games.

This how the formula works: (Met wins x Dodger losses) / (Met wins x Dodger losses) + (Dodger wins x Met losses).

Not surprisingly, it predicts a Mets win, but with a winning percentage of only .543. That points to a seven-game series, where the winning percentage would be .571. Oakland and its 104 victories would be favored to beat the Dodgers 56.1 percent of the time (.561 percentage).

But the wild card in this formula is that it can't account for the relative strengths of the individual divisions. Would Oakland have a .642 percentage if it played in the NL West? The A's would need a 112-50 season in the NL West (.691 percentage) to have a 61.4 percent chance of winning a series with the Dodgers: that would be close enough to predict a six-game win for Oakland. Either way, those 2-1 Vegas odds seem a bit steep.

The more lucid among you may have noticed that this formula tends to predict a seven-game series in most every playoff and series matchup, since few team get into postseason play without strong winning percentages. It does -- and seemingly does it accurately.

In the last 56 seven-game postseason contests (including division playoffs), the series winner has swept all four games just six times and won it 11 times in five games and 11 times in six. The other 28 contests have been decided in the seventh game.

But if everything is coming down to one last game and the underdog has a Cy Young candidate on the mound -- at home, it would seem that all bets are off.

Hershiser's performance in playoff Game 7 amplified in many minds west of the Rockies that he's the league's best pitcher. Danny Jackson, who fashioned an identical 23-8 record, is generally regarded as his closest competition while David Cone and his 20-3 record are rarely mentioned in these parts.

The Cy Young vote should come down to four criteria, as usual. Who had the best ERA, who had the best won-loss record, did anyone pitch for a championship team and were there any outstanding personal accomplishments like, say, a record-setting string of shutout innings?

Relievers won't be much of a factor, as is usually the case when legitimate starting-pitcher candidates exist.

In the American League, Frank Viola is an obvious choice with the top won-loss record and a better ERA than his closest competitors. Nobody broke any records and the best pitchers on the division winners, Dave Stewart, Roger Clemens and Bruce Hurst, had serious deficiencies.

In the NL, though, we've got a race ripe for a little loose statistical talk.

Using a formula that borrows heavily from a Pete Palmer invention called Linear Weights, we can rate Hershiser, Cone and Jackson based on how much better their ERAs and won-loss records were than the average NL pitcher. While we're at it we can flesh out the numbers a bit by giving a pitcher credit for durability (more innings pitched and decisions). Assuming their ERAs and won-loss records are of equal value, an iffy proposition. Cone ranks first, 59.6 percent above average. Hershiser is a close second at 58.5 and Jackson is third at 46.8.

But Cone and Hershiser pitched in stadiums known as pitchers' parks while Jackson toiled in Riverfront Stadium, which the Elias Baseball Analyst notes was the third-best hitters' park in the league from 1983-89. It seems only fair to plug that crucial info into the formula, and the accompanying chart does just that, with predictable results. But it wasn't enough to change the order of finish.

Ballpark effect is one of two crucial factors routinely ignored when assessing a pitcher. The other is hitting support. What ballpark effect means to ERA, hitting support means to won-loss records.

In 1985, Fernando Valenzuela didn't allow an earned run in any of his first four games, yet was saddled with a 2-2 record for lack of hitting support.

But you're not going to find hitting support factored into these pitching stats; in fact, you're probably not going to find hitting support numbers anywhere unless you read esoteric journals published by number-crunching fanatics with access to a restricted database.

Without those numbers you'll probably do what I did when faced with three pitchers owning comparable records: You'll cast your personal, Dodger Blue-tinged Cy Young vote for the man who broke Don Drysdale's consecutive shutout string, Orel Hershiser.

American LeagueNational League



The Bottom Ten


Swift8124.51-15.5Smith Z.7153.40-15.0
Witt M.13164.22-11.0Dunne7113.87-14.4
Moore9153.72-7.7Moyer9153.19 -7.3

* W is Wins
* L is Losses
* ERA is Earned Run Average
* Pct is the player's ranking, expressed as a percentage above or below the average major league starting pitcher. Won-loss percentage and ERA are given equal weight. Other factors that affect Pct include: Ballpark influences, relative ERAs of the AL and NL, number of innings pitched and number of decisions.