"St. Louis Cardinals ace Chris Carpenter floats into the Inner Harbor this evening fishing for something he hasn't seen since early August of last season: a winning streak.

The Cardinals' righthander will start his 17th game of the season as Exhibit No. 495,609 of why a pitcher's win-loss record is misleading. Six times this season, Carpenter has pitched what we call around here a Gibby -- that is, at least seven innings and no more than three earned runs. Undone by run support or defense, Carpenter has gone 1-1 in those six starts with four no decisions. He could be 5-1 in those games. While we've had a good debate on Twitter this season about the value of wins - I believe they remain important to baseball coverage, storytelling, explaining of what happened, but are terrible predictors - we can all agree that Carpenter has not been a two-win pitcher this season.

There are eight starting pitchers in the majors this season who have only made starts and are stuck on two wins. Carpenter's 4.26 ERA is the fourth-lowest of the group, behind Tom Gorzelanny in 11 starts (4.18), Alex Cobb in five starts (3.41) and Carpenter's commiserating colleague Dustin Moseley, who is 2-7 for San Diego this season despite a 3.03 ERA in 15 starts.

Clearly Carpenter has pitched well enough to expect more wins than two.

But how many more?

There is a stat kept at Baseball Prospectus and elsewhere called Expected Wins or E(W). It basically takes the tell-tale stats of individual starts and compares them against history. If, for example, Pitcher X threw seven innings and allowed two earned runs and through history teams win 82 percent of those games, then Pitcher X receives .82 Expected Wins. Add up all of his E(W) and that gives you a sense of how many games that pitcher should have won. It's not as clean as a win-loss record, it's not as accessible as a win-loss record, but there it is."