Back in October of 2004 I wrote a Stat of the Week article called "Is quality start really a quality stat?"   I pointed out how a pitcher who has a quality start (6 innings or more with 3 earned runs allowed or less) wins about two-thirds of the time and that shows the "quality" of the stat.  Recently, ESPN baseball commentator Bruce Levine, whom I highly respect, has been talking about how much he hates the stat pointing to the same argument that's always made about it: "If a guy pitches exactly six innings and allows three runs, that's a 4.50 ERA.  What's quality about that?"   My argument has been that this is simply a mediocre tree in the big beautiful forest that makes up the quality start statistic.

Well, I now take exception to myself.

In preparing to refute Mr. Levine's position on the stat, we decided to look at the situation that he and everyone else points out, exactly six innings and exactly three earned runs.  It seemed to me that a starting pitcher who did this was keeping his team in the game and that this would generally lead to good results.  Over the last 10 years, there have been 2,118 games where a starting pitcher has gone exactly six innings and allowed three earned runs.  Pretty big sample size.  It turns out that the winning percentage in those games for that pitcher's team is a little below .500.  Maybe you can still say he’s keeping his team in the game, but it's hardly quality.

And I might be able to live with that as the mediocre tree in the forest, but there’s more.

What about if the pitcher goes seven innings and allows three earned runs?  What about eight innings?  To my great amazement, it turns out that if a starting pitcher pitches anywhere between six and eight innings and allows three earned runs, his team wins less than 50% of the time (5,039 games, 2,491 wins, .494 winning percentage).  Even right at eight innings and three earned runs, it’s below .500 (356 games, 171 wins, .480 winning percentage).

Here’s the complete chart:

Another interesting part of this chart is this: when the pitcher records exactly 6, 7 or 8 innings, the winning percentage is below .500.  But when he extends into the next inning for at least one out, the percentage goes above .500.  I've been trying to figure out what this means.  I thought it might have to do with the difference between the leagues and pinch-hitting for the pitcher in the National League, but we didn’t see much when we looked at the data.  I am going to chalk it up to sample size issues (a lot less data where pitchers extend into the next inning) until I hear something better.   Suffice it to say, anytime a pitcher pitches eight innings or less and allows exactly three runs, the winning percentage is around .500. Maybe you can call it a "Kept-my-team-in-the-game Start", but not a Quality Start.

In summary, while the quality start is still a useful statistic, it is seriously flawed.  A way to fix it would be to change it to six innings or more, two earned runs, or more than eight innings, and three earned runs.  That would be quality, but that’s also too complicated.  The better stat is a new one that Bill James invented called "Gems".  More on that next week.

Side note: I want to give a big shout-out to my two excellent research assistants at Baseball Info Solutions, Ben Jedlovec and Rob Burckhard.  Without their fantastic research help, you'd see nothing but a lot of hot air and blank space in these articles.