Thursday, July 21, 2011

Made-Up Statistic: The Super Start

Recently, I approached my friend and fellow blogger, William Tasker of The FlagrantFan, with an idea of how to make the seemingly-meaningless baseball statistic of the Quality Start into something, well, more meaningful. Using his love for the Play Index tool, the following article is what he came up with for COSFBA's first-ever Made-Up Statistic: The Super Start.

The statistics of baseball are like running your fingers over sandpaper. There is something tangible and pungent about the numbers. If you have read this writer's work before, you know that there is nothing this author enjoys more than playing around with those lovely statistics so easily found these days on this thing we call the Internet. The trouble with some of those statitics is that there are some stats that are simply meaningless. Someone at some time thought a particular statistic had some value and it became part of the baseball language. And then those statistics hang around whether they have value or not. What makes these meaningless statistics dangerous is that casual fans and even some solid baseball people think they still mean something.

Many writers have their own list of meaningless statistics. Many include Batting Average and Runs Batted In as meaningless. This writer struggles with those and still thinks there is some value there. For some, the Save is meaningless as is the Hold for set up guys. While this author thinks those statistics have some weakness, they are still a value to consider. For example, if your particular team doesn't have a guy that can close a game and your team keeps getting beat in the ninth inning, well, then you're going to think more of that stat aren't you?

This writer has a slightly askew way of considering the worth of a statistic. Take the number of games, appearances, etc., that are possible for that statistic to happen and if the percentage it does happen comes up too high, then it's meaningless. No statistic bears this out more than the Quality Start. In a "quality start", a pitcher has to pitch six full innings and give up three earned runs or less. First of all, any stat that rewards a player for an ERA higher than league average is flawed to begin with. And three earned runs in six innings is a 4.50 ERA.

But again, for this writer, it comes down to how often it happens. For example, with the Quality Start, there have been 1,399 games played so far this season (before Sunday's action). Since each game features two starters, that means there have been a total of 2,798 possible chances for a quality game to be pitched. And before Sunday's action, there have been 1,551 Quality Starts. That means that the QS happens 55 percent of the time! How can any stat mean anything when it occurs that many times?

Justin Verlander has made 21 starts so far. Nineteen of them have been quality starts. While that may give his agent fuel for the contract wars, it becomes meaningless. So what we need is a statistic that does give some meaning to how many times a pitcher pitches a great game. Our definition has to occur a lot less frequently as a percentage to have some meaning.

We can't use "quality starts" as a name because it is already taken, and in this writer's mind, already tainted. Like all good statistics, we need initials that have a ring to it. A.C.E would be great, wouldn't it? But we can't get the letters to work. So what can we use? Let's go with something simple and call them Super Starts or SS for short.

The first thought was to move the innings up to seven instead of six. But that has still happened 895 times or 32 percent of the time. That's still a lot, right? So we have to make it more exclusive than that. COSFBA's founder, Daniel Aubain, wanted to add a Win to the mix to make the figure more cogent. While understanding that logic, there is still too much chance to a Win and too much still depends on what the rest of the team (and the other pitcher) does. We'd want something more in the pitcher's control.

Control? Okay, what if we make it seven innings, two runs or less and one walk or less? There have only been 130 of those. So that's a good candidate. The only problem is that it punishes guys who are effectively wild but still limit the other team often. Heck, even Carlos Zambrano has thrown four of these kinds of games. So we'll call this a no because of the Zambrano factor.

Perhaps there is a solution to this Super Start idea. Let's go with eight or more complete innings with one or less earned runs and two or less runs. Why the latter? Too many pitchers allow the extra base runners caused by an error to score. The Super Start not only gives the pitcher credit for the earned runs allowed, but also for the ability to limit the damage when his defense falters.

Okay, we have our stat. Does it pass the occurance test? It's only happened 216 times this season (before Sunday's action) in 2798 possible starts. That's only 7.7 percent of the time. That works.

So who would our leaders be? It's a good list:
There are some inherent weaknesses with this statistic. It doesn't take into account the level of talent faced by these pitchers. But then again, neither did the Quality Start. It isn't meant to be the be all and end all of pitching stats. There are far too many that are important in judging the overall value of a pitcher. But, it does give you an idea of which pitchers have thrown the most "gems" during the course of the season.

Of all the instances a Super Start has occurred this season, the starting pitcher actually lost the game 18 times. But most of the time, if a pitcher pitches this well, he's going to win. Unless, of course, your last name is Fister, who is 0-1 in those four Super Starts.

If you want to know the pitchers with the most Super Starts for the past two years, that would be Felix Hernandez with 16 and Halladay and Lee tied with 15. The leader for the last ten years is Halladay with 70 followed by Sabathia with 49.

The leader for the last twenty-five years is Roger Clemens with 130 followed by Greg Maddux with 111. The leader for the last fifty years would be Tom Seaver with 175 followed by Gaylord Perry with 174. That's a surprise and impressive.

So what do you think? Does the Super Start work for you? Does it mean anything? Well, at least this writer knows that it means a heck of a lot more than a Quality Start, right?

Have an idea for a Made-Up Statistic you'd like us to explore in a future article? Email me directly at with your idea or if you'd like to do the research yourself and present it in a guest posting opportunity on COSBFA.


  1. The Quality Start, though too much maligned, can certainly can be improved on. The problem is that, just as with the QS, *any* fixed set of conditions will include some games you don't want to honor, and exclude some that should be honored.

    For example, your Super Start would include (through Wednesday, July 20) 71 games of exactly 8 IP, allowing exactly 1 ER, and either 1 or 2 total runs. Nothing wrong with that.

    But it would exclude 118 *scoreless* outings of 7, 7.1 or 7.2 IP.

    That's unreasonable. Unless your bullpen is severely depleted, 0 runs in 7 IP is clearly a more valuable performance than 1 or 2 runs in 8 IP.

    The problem is insisting on a fixed set of conditions. The solution is to let the Runs fluctuate a little based on the IP. For example, an alternate version of the Super Start could include these games:
    -- 0 runs allowed in at least 6 IP;
    -- up to 1 run allowed in at least 7 IP; and
    -- up to 2 runs allowed in at least 8 IP.

    This definition would not, alas, let us find all such games with one search of most existing database, such as the Play Index on Baseball-Reference. But we must not let the technology drive the discussion; if we can agree on a new definition for "Super Start" (or whatever), the programmers will quickly catch up.

    P.S. I think the whole "ER" distinction is too problematic to be used in this context. As you noted, some pitchers are better at overcoming an error than others. Also, some pitchers, like A.J. Burnett, make their own errors. And star pitchers often get the benefit of the doubt in official scorers' decisions. I say, keep it simple; just count the runs.

  2. John,
    I totally agree with your response.

    How can something measured by "6 or more innings pitched with three runs or less" all be considered of equal "quality". The same is true of the save.

    To me, discussing how to make a particular stat "better" doesn't devalue it but hopefully leads us down a road to make it more valuable.

    I followed your link back to and enjoyed your article. Are you Twitter, too?

    Have you seen our SABR-themed articles of late?


  3. Super Start: 7 IP, 1 run or less
    Super Duper Start: 8 IP, 1 run or less