Is the win a useless statistic?
There’s a mini-discussion going on in the Twitterverse/blogosphere surrounding the usefulness of the “win” as a statistic. Here are some of the opinions:
Daily News columnist Jesse Spector:
The win is accessible – a single number whose meaning is known on some level by everyone from the casual fan to the diehard. The more you know, the less it tells you, and like everything else, wins must be taken in the context of their era. But as one of the statistical cornerstones of a game whose beauty is in its towering history, the win cannot simply be dismissed as useless.
I just don’t see the utility. They’re incredibly imperfect. You can “earn” a win or a save despite horrible results…Then our job is to educate the casual fans. Pretending W/SV/RBI are still useful does them a disservice.
Whenever a manager holds his best reliever for a “save” spot rather than using him in a tight spot, he puts indiv. above team…Pitcher “wins” and saves are not simply benignly “useless,” they actively encourage bad decisions.
I’m fine with “why does player X have Y wins?” analysis.But saying “X is good because of Y wins” is unacceptable at this point
Baseball Analysts’ Pat Sullivan:
But wins / saves are only important if we continue to assign value to them. Why not a more meaningful stat be the “baseline”?
I’m on board with almost everyone in some sense, but mostly with Colin and Keith. Look, talking to casual fans about advanced stats and getting them interested in fantastic. I’m all for it. However, we shouldn’t give them an imaginary carrot to chase in the form of “yeah, wins can be useful, but have you seen this stat…”, when we know that wins are simply nonsense. They are arbitrary measurements of basically no relevance. If we want to attract casual fans to sabermetrics, then instead explain why wins are useless! I mean, isn’t this one of the purposes of advanced analysis, to put to bed illogical analysis in favor of intelligent, reason-based analysis? Pretending that wins have meaning actively harms the objective nature of sabermetrics, and I won’t sacrifice that to win over some casual fans who will be learning misinformation.
It would be interesting to get Bill James’ take on this. I think he would believe that 1) Wins aren’t 100% useless, and 2) that everyone thinks in terms of wins, so we ought to find ways to incorporate that into our work. His articles on BJOL continue to include wins as a valuable indicator of a pitcher’s worth.
This is one of the reasons that Bill’s work has fallen out of favor with many hardcore sabermetricians.
For me, personally, I like to think in terms of wins. In fact, I crave it. That’s why I love stats like Win Shares, WAR and WPA. But, like you, I don’t even bother to track individual pitcher wins or saves. Teams win games, not individuals.
Dave,
Exactly. I was specifically referring to the win as a pitching statistic, but likewise love using a “team win”to assign value.
Wins aren’t useless, they are just a much less efficient indicator of the quality of a pitcher than more advanced metrics. It is highly, highly likely for example that a 15-6 pitcher pitched better in some Platonic sense than a 6-15 pitcher. So to begin with they are certainly not “arbitrary measurements of no value”.
Wins also have some value when assessing the relative contribution of pitchers to a team’s season because they sum to exactly the same total as the number of victories. They reflect a number of factors such as games started, innings pitched, pitching quality, etc. It is true they do not reflect defense or offense—but wins DO convey information about pitching.
Wins weren’t useless until the advent of the relief specialist. Back in the day, a starter was expected to go deep into a game, if not finish it, and a win meant something. If the bullpen was used, that usually meant the starter failed and the game was lost.
Here’s my suggestion: give a win to every pitcher who pitches in a winning game, provided they meet certain criteria. Maybe borrowing from the much maligned “quality start.” Pitch a minimum of 1 inning (or 2, or 3 – although this leaves out our favorite loogys), have an ERA 4.50 or lower (or maybe base it on WHIP). If the team wins – you win. What if no one meets the criteria? “Team” win – none for individuals, like a “team” rebound in basketball (I can see the first article wondering what team has the most “team” wins). Sure, all individual win records will be obliterated, mostly likely by middle relievers, but at least the win would mean something again.
…furthermore, we could do the same thing for a loss. No inning minimum. ERA of 4.50 or higher (etc.) Any pitcher qualifying on the losing team gets a loss. Again, if none – “team” loss.
I don’t believe that wins have a use as an analytical tool, but they do have use to explain the psyches of pitchers and managers.
It is worth noting that RA, which has justifiably been deemed a better stat (for its type) than ERA, does correlate better with wins than ERA does.