I have to think that if baseball were invented today, the “quality start” would not exist. The quality start – defined as 6 or more innings pitched while giving up 3 or fewer runs – is one of those institutions that seems like it started out with the best intentions, but never really got all the way there.

When the quality start was first being dreamed up, I think the conversation went a little like this:

Person 1: Let’s come up with a way to track how well a starting pitcher performed each game.

Person 2: That’s a great idea, but we already have ERA, wins and losses, strikeouts, walks…

Person 1: yes, but those are all old stats. We need something new. Some starts are good and some starts are bad, so let’s call the good ones, “quality start”….

Person 2: Isn’t that just called a win?

Person 1: Well, sometimes a pitcher throws well but doesn’t get the win…

Person 2: Isn’t that why we have ERA?

Person 1: Yes, but we need something easier. All those numbers. And numbers people don’t even like ERA all that much anymore. Let’s say that anytime a pitcher throws six or more innings and gives up three or fewer runs is a quality start.

Person 2: Setting aside that we could just call that an ERA of 4.50, what happens if a pitcher gives up, say 4 runs, but pitches all nine innings? That’s a lower ERA and has a higher chance of the team winning the game.

Person 1: That can’t be a quality start because they gave up more than 3 runs.

Person 2: OK…but what about the pitcher that is throwing a shutout but leaves the game after five innings?

Person 1: Not a quality start. Gotta cross the six inning mark

Person 2: Why six innings?

Person 1: Because that’s a quality start.

Person 2: OK. I have another idea: do you want to join a fantasy baseball league? Buy-in is about $500 per year, but you’re an ideas person, and I think you would do just great…

This is about the furthest I’ve made it in a column this year without mentioning some numbers, so we have to fix that. In an article exploring quality starts, David Gassko asked the question: how much does the quality start stat help us understand about a pitcher’s performance from year to year? After controlling for ERA, the answer is indistinguishable from zero.

The quality start doesn’t really tell us much, but much like when Henry Chadwick first labeled the walk an error, I think the quality start has had a negative effect on the way we think about baseball games. The quality start tells us those pitchers who consistently give up three or fewer runs over six innings are good and that those who give up more are not doing a “quality” job. Consistency has to be a good thing, right?

Well, not really. And there’s no shortage of evidence on this, either.

How can pitching inconsistency be a good thing? It is because not all runs matter the same. Matt Hunter demonstrates that a team’s chance to win does not decline at a steady rate with more runs surrendered. Due to this, pitching a great game (0-2 runs) is much more valuable than pitching a clunker (5+ runs) is harmful.

His data set utilizes every game played since 1993 (20 years of data at the time of writing). The results are pretty amazing:  even if a pitcher gives up five runs through seven innings, his team will still win about 35% of the time.

So, just for an illustration, let’s take two players, one is our super-human consistent starter (nine innings pitched, four runs every single time) and the other had two types of outcomes: a gem (2 runs, 9 innings) or a clunker (6 runs 9 innings). Let’s say both start 30 games.

[This is obviously an oversimplification, but the results are the same even if you use the more accurate equation: (RA+4.88)/9*(9-IP) for both pitchers. This would just assume that the bullpen is constant for both]

So, which pitcher’s performance results in more games win? It turns out that the inconsistent pitcher will win about a half game more over the course of a season than the consistent pitcher. How is this possible? Its because the chance of winning against a team that scores 0-2 runs is very high, but if that team scores 5 or more runs, you still have a shot at winning (in that article, Sal Baxamusa drops this number: teams win 18% of games where they give up 10 or more runs. That is kind of mind-blowing).

This all points to once conclusion: Truly valuable pitching is the kind that can shut down the opposing team, even if a pitcher occasionally gets rocked for five or six runs. The other big take away is that even great pitchers throw terrible games.

Take a look back at the 2006 HBT article and scroll down to the frequency distributions. Those are the number of expected runs per game for the three leading contenders for the 2006 NL Cy Young Award. The charts are generated based on the number of runs given up by the pitcher and then assuming a 4.88 ERA (league average bullpen) for the remainder of the game.

Brandon Webb, who would go on to win the award, gave up 6 expected runs five times, seven expected runs twice, eight once, and nine expected runs twice. Webb’s ability to keep his team ahead in the game two thirds of the time is what made him one of the best in the league. Yet, if a pitcher gets about 30 starts per season, that means that the Cy Young winner left his team with a less than 50-50 chance to win the game about a third of the time.

This means the old saying “great pitching beats great hitting” is backwards, but still true: its so hard to have great pitching, that when it does happen, you usually win. And perhaps that why if you keep your team a notch above the 50% mark to win the game for six innings, we might want to give you something. Not something normal or expected, because keeping a ballgame above the 50% mark is something that very few pitchers can do.

Perhaps we could call it a quality start.