Win in Doubt: An Obsolete Statistic Overstays Its Welcome

Lewie PollisSenior Analyst IIINovember 16, 2009

LOS ANGELES, CA - OCTOBER 08:  Pitcher Adam Wainwright #50 of the St. Louis Cardinals pitches in the first inning against the Los Angeles Dodgers in Game Two of the NLDS during the 2009 MLB Playoffs at Dodger Stadium on October 8, 2009 in Los Angeles, California.  (Photo by Jeff Gross/Getty Images)
Jeff Gross/Getty Images

What is the best way to measure a pitcher’s skill?

Is it his strikeout rate, demonstrating the pitcher’s ability to prevent batters from hitting the ball? Is it innings pitched, showing his durability? Is it WHIP, the talent to keep runners off the base paths?

Is it his defense independent pitching stats, an estimate of how many runs he would give up with an “average”-fielding team behind him? Should you use more advanced sabermetrics to determine how many additional games a team won because of the pitcher’s contributions?

Or is it simple ERA, his ability to stop runners from crossing the plate, without normalizing each pitcher’s unique circumstances?

Whatever the answer, one thing is clear: the media’s favorite statistic, wins, is fatally flawed.

It doesn’t take a sabermetician to see that wins is an absurd metric. As early as my t-ball days, I had decided that it was a stupid and inaccurate way to measure a pitcher’s performance.

As it turns out, even non-baseball fans can appreciate its stupidity. “If a pitcher gives up 10 runs, but his team scores 11 runs, he gets a win,” I told my friend who couldn’t tell Babe Ruth from a Baby Ruth, “but if a pitcher gives up just one run and his team gets shut out, he gets a loss.”

His immediate reply: “That’s so stupid!”

Sure, wins and losses are more than just luck; Joba Chamberlain is probably not capable of going 22-3 like Cliff Lee did in 2008. Just as tuning a cello by the neck will get you close to concert pitch, a pitcher’s record will give you some indication of his performance.

But judging players primarily by wins would be like an entire orchestra neglecting to use the fine tuners (if you don’t know what large-scale poor intonation sounds like, look up the term “brown note”).

That’s not to say wins are the only overrated statistic. Through runs batted in, for example, hitters are rewarded for their teammates’ ability to get on base ahead of them. And anyone who thinks Carl Crawford’s base-stealing skill is comparably valuable to Carlos Peña’s powerful swing should have his head examined.

What makes wins the worst of all the stupid statistics is the importance most analysts place on it. What’s the first number an analyst or broadcaster always mentions when they talk about a pitcher? Even in print, a pitcher’s name is always followed by his record, highlighted by a set of parentheses.

Even the statistics-oriented Sports Weekly champions wins as a significant metric. Writer Jorge Ortiz was clearly surprised that 16-game winner Zack Greinke was the staff’s top choice for the AL Cy Young while CC Sabathia and his 19 wins were pushed to fourth place.

Ortiz wrote an apologetic explanation for picking Tim Lincecum as the NL Cy Young (his paltry 15 wins put him in the same category as Derek Lowe and Bronson Arroyo) instead of Adam Wainwright (he has an NL-leading 19 Wins, but finished behind in every other category). The article also listed Chris Carpenter as a potential winner , but only because he had 17 wins, not because he led NL pitchers in ERA and WPA.

They took a step most analysts neglect by recognizing the decrepit Giants lineup as a hamper on Lincecum’s record, but in doing so ignored perhaps the most important reason why wins is a poor metric: inconsistency. I can confidently say that the 2009 Giants never scored exactly 4.06 runs in a game. You can expect the team to score about 12 times in a three-game series, but there is no way to predict how the runs will be distributed.

There are many statistics that are similar to wins, but more accurate. Quality starts measures the amount of times a pitcher performed well enough to expect a win. Wins above replacement measures a player’s skill sets in order to estimate the difference, in games, between him and an average substitute. And win percentage added uses probability theory to determine how much a team’s chances of winning improved thanks to the contributions of a player.

I’m not saying that baseball skill can be quantified only through advanced sabermetrics. But why should we put so much stock into a statistic that doesn’t make sense?

In the showstopping musical number from Rent , the characters ponder a similar question. While measuring a pitcher’s performance in “love” might seem a bit impractical, it makes a lot more sense than wins.