(*) MASTER NOTES: Fixing the Wins Category

Last week in Master Notes, I talked about how much I hate the wins category. And I’m bravely standing by my position, even though a) in the two days after my commentary, Darren O’Day got me two wins I clearly didn’t deserve, and b) I’m now completely sober.

I ended my Master Notes commentary by promising that I would be looking into alternatives to Wins. Now, just one week later, I’m keeping that promise, even though it will put me in an awkward position when my wife wants to discuss a promise I made three years ago to clean out the garage.

We all know that Wins is a category many owners dislike, because of its randomness and unfairness. But nobody ever does anything about it. We just seem to play on, year after year, and we fume when the 2014 pennant winner stumbles into 10 wins from Mike Dunn, because he thought he was drafting Adam Dunn.

We just put up with this nagging annoyance. It's like having a brown recluse spider in your shoe, or being constantly reminded of your promise to clean out the garage.

So let’s fix the Wins category. Let’s come up with a better stat to better reflect the performance of starting pitchers.

To start, we need some ground rules. I propose that whatever stat replaces the win must meet three qualifications:

• Like a win, it should be a is-or-isn't binary counting stat, not a continuous-scale or rate stat.
• It should be roughly as frequent as fantasy wins.
• And it should be simple.

---

When this topic comes up in conversation with your friends, it usually starts with one of two thoughts. Either you start discussing alternatives to Wins, or you realize you need a better caliber of friends.

If your friends are the more hard-core, sabermetrically inclined fantasy owners who emerge from their basement crypts at night, you might hear calls to use BaseballHQ.com’s “Pure Quality Start” (PQS)  or the Bill James invention “Game Score” (GS).

There’s a very persuasive argument here: These stats offer extremely comprehensive measures of starting-pitcher performance. Both PQS and GS account for innings (outs and innings, in GS), runs, strikeouts, walks, and hits, and PQS also includes home runs.

Both metrics run up against the yes-or-no issue, because they score on scales. PQS awards whole-number scores on a scale of 0-5, with an average of 3.2. GS has an indeterminate sliding scale that starts somewhere below zero and finishes somewhere over 100. It’s like grading pitchers in Fahrenheit. This could have some useful insider-slang possibilities, like being to nod knowingly when your buddy says, "Hey, Max Scherzer tossed a real Yuma last night," and you could even respond, "Yeah, but Tillman hung up a Fairbanks on me." Or not.

In 2014, the average GS was 53, and the most common score was 63. The year's top Yuma for the year was 102, a Clayton Kershaw no-hitter with 15 strikeouts and no walks. The worst Fairbanks of the year was a -12 by Colby Lewis, a frosty mound of 13 hits and 11 earned runs in 2.1 innings.

Leagues could use each team’s average PQS or GS for either score, but that would mean a third pitcher stat as a rate rather than a count. There’s an easy solution, though: Above the threshold, the game counts, below the threshold, it doesn’t.

This idea also helps addresses the frequency issue. For example, if we want to use PQS as the category, we could say that a pitcher gets a credit only for PQS-5 games.

If we take this step, something interesting happens. And you’re probably thinking that after 500 words, it’s about time something interesting happened. It turns out that in frequency, PQS-5s match almost perfectly to Wins. Returning to 2014, the 15 teams in the Tout Wars Mixed Auction league combined for 1,338 Wins. That same year, starters across baseball earned 1,335 PQS-5s—a difference of just three!

Similarly, leagues could switch the GS metric from scale to binary by setting a threshold score. In 2014, setting the threshold at 64 GS points or more would have captured 1,339 starts—within one of the Tout Wars target for Wins. In both cases, credit for undeserved Wins is taken away from bad starters and lucky relievers, and given to starting pitchers who pitched well. It can be done.

So PQS and GS fix the Wins issue, and can be adjusted to get the frequency right. Problem solved, right?

No.

PQS and GS still don’t cut my mustard. Because they’re just not simple enough.

---

Perhaps the strongest factor in favor of the Win as a fantasy stat—and the one most often trotted about by the Luddites who favor keeping it—is that the Win is simple. If we watch a game, or listen to it on SiriusXM or mlb.com, we always know if our starter is in line for a win, and we know at the end if he gets it. The winning pitcher is always mentioned in game recaps, on the crawl at the bottom of our TV screens, and when announcers read results from “the out-of-town scoreboard.”

None of this is true of PQS or GS. Nobody mentions them on TV or radio, and calculating them yourself is very difficult, because they have more moving parts than Mickey Rivers after three cans of Red Bull.

Some components of PQS subtract hits from innings and strikeouts from innings, with different totals needed to earn a point. Another divides strikeouts by walks. And GS is even worse: The starter goes into the game with 50 points, gains various points for innings and strikeouts; and loses various points for runs, hits and walks. At some point the total is divided by the square root of the open floor space in your garage. Although I might be confused about that last part.

So figuring a PQS or GS score means getting out a pen and doing a bunch of cipherin’ on a piece of scrap paper, which inevitably turns out to be the back of your daughter’s World History essay. I once started a GS calculation when the game was over and the late local news had started; by the time I finished, Jimmy Fallon was doing a karaoke version of “Crazy Train” with some guy from Game of Thrones and Pope Frank.

Don’t get me wrong. Both PQS and GS are excellent metrics for pitcher analysis.  I use PQS almost every day, and I find it invaluable for spotting trends among starting pitchers and for avoiding cleaning out the garage.

In a perfect world, PQS and/or GS would also be excellent in defining a start worthy of fantasy reward.

But we know the world ain’t perfect. In baseball particularly, merit is often not the deciding factor. If you don’t remember that now, you will when you see the Gold Glove voting or seven Kansas City Royals starting in the All-Star Game.

And so, despite their merit, neither PQS nor GS is the solution for replacing Wins as a category.

Just imagine yourself at your league meeting, proposing this change. “You see, guys, you start with 50 points, then add a point for each out, an extra point for innings beyond the fourth, then subtract various points for runs, walks and hits allowed… and, oh, by the way, Phil and Dave? You guys have to build garages.”

Unless your league is made up of the guys from the Big Bang Theory, this proposal is going to get hooted down. You might even lose your position as league statistician or Commissioner, which alone makes it worth a try. The downside is that you’ll never again be taken seriously by your leaguemates, even when you have a good idea. Believe me.

---

So after careful consideration, I have decided that we should replace the badly flawed and irredeemable Wins category with some variation of the Quality Start.

There’s a longstanding, common objection to QS as a stat: three runs in six innings is a 4.50 ERA, and that’s not quality. As with most longstanding, common objections, it's not really valid. Only about 8% of starts in 2014 were three-in-six QS minimums, and the overall ERA for all QS was 1.88, exactly the same as the ERA for all the Wins.

As a fantasy stat, the QS easily meets two of the requirements: First, it’s a binary counting stat. Second, it’s simple. Its components—innings and runs—are mentioned on the crawl and recaps, and it’s simple to calculate. It’s also commonly seen in most online stat packages, including BaseballHQ.com, OnRoto, ESPN.com, and many more.

There is an issue with the last factor, requiring the replacement stat be roughly as frequent as Wins. Remember, our target is around 1,350, to match a typical year in the wins category. But there were 2,623 QS last year, double the target and therefore a source of griping. And no fantasy league needs additional sources of griping.

One possible answer comes from Nolan Ryan, who knew a thing or two about quality pitching, despite all those walks and wild pitches. Ryan said a QS should be three runs or less, but in seven innings.

And whaddya know? That simple adjustment reduces the 2014 QS total to just over 1,400, and reduces the aggregate ERA to 1.46 (the aggregate WHIP is 0.84, if you’re keeping score at home).

More importantly, 1,400 Ryan QS is close enough to the 1,350 requirement to shut down most arguments, although if you’re like me, you’ve seen fantasy arguments over less.

OnRoto.com has leagues that use the Ryan variation, as well as leagues using eight other QS variants that don't also use Wins. I checked all of the non-Win variants, to see which ones came close to our 1,350 target. None of them was better than the Ryan variation.

A further set of calculations, which I did on the back of my daughter’s French homework, revealed there are two innings-runs combinations that edge the Ryan: starts with at least 6 IP with one run or fewer, and at least 6.2 IP with two runs or fewer.

I’m not crazy about a zero- or one-run requirement. It’s over too soon in a lot of starts. Nor do I like a non-whole number inning requirement, though we could also say “20 outs.”

So it’s settled: The Ryan QS—at least seven innings, no more than three ER— should replace Wins as a scoring category in fantasy baseball.

Now if you’ll excuse me, I have to go find my daughter’s French homework. I think I left it in the garage.

More From Archives