Not all no-decisions are created equal: Evaluating a little-examined pseudo statistic

That was the title of Gilbert Martinez's presentation at the recent SABR convention. Gilbert is president of the Rogers Hornsby Chapter of SABR in Austin, TX.

Click here to get the power point slides. It has alot of neat graphs.

Here is the abstract:

"In 2009, two pitchers recorded 16 no-decisions. The Houston Astros’ Roy Oswalt set a franchise record for no-decisions and was 8-6 in 30 starts. The L.A. Dodgers’ Randy Wolf was 11-7 in 34 starts.

Research shows that 2011 Hall of Fame Inductee Bert Blyleven holds the record for most no-decisions in a season with 20 after a 12-5 record in 37 starts with the Pittsburgh Pirates in 1979.

While Oswalt and Wolf didn’t set a record, they were among the most in a single season and they did receive some attention by the media and fans. Little analysis has been done to understand the nature of these no-decisions. Statistics abound in baseball, and especially with pitchers. Statistics capture games started, wins, losses, saves, earned run average, pitch counts, walks, hit batsmen, ground-ball outs, fly-ball outs, innings pitched, homeruns given up, strikeouts per nine innings, and so on. This project will focus on starting pitchers and no-decisions recorded as a result of starting a game.

The purpose of this project is twofold: 1) to evaluate the most no-decisions by a starting pitcher in a single season and in a career and 2) to determine which pitchers with the most no-decisions were unlucky or lucky.

A pitcher would be considered unlucky if he leaves the game with a lead, only to watch his team’s bullpen lose the game in the late innings. This would be considered a positive no-decision (positive because it suggests an effective pitcher). Likewise, a lucky pitcher would be one who leaves the game with his team trailing, only to be bailed out by a potent offense, thereby taking him off the hook. This would be considered a negative no-decision. If he leaves with the game tied, this would be a neutral no-decision.

These no-decisions are not equal, and some pitchers are more valuable to their teams than others. In other words, some pitchers saddled with numerous no-decisions but who have more positive no-decisions than negative no-decisions are more worthy of our sympathy than those with more negative no-decisions compared to positive no-decisions.

Initial review of the literature shows little on this subject, so it appears this research project would contribute to our understanding of lucky and unlucky pitchers in the context of no-decisions. In fact, this statistic is not readily recorded on Baseball-Reference.com.

The methodology includes running a formula in a pitchers database to determine no-decisions: (Games started) minus (wins + losses) = no-decisions. However, those results would need further review to eliminate no-decisions that come from relief appearances. The remaining results would be analyzed to determine the circumstances in which the pitcher was given the no-decision: Was his team leading or losing when he was removed? Was he lucky in receiving the no-decision or was he unlucky?

The result would be the number of positive, negative and neutral no-decisions, shedding new light on the pitcher’s effectiveness."