you're reading...

Pitcher Peaks: Finding a Starting Pitcher Sherpa

pitchermountainWith Ed Walsh, a peak-oriented pitcher if there ever was one, eligible for the HoME’s 1921 election, I’m reminded of a favorite topic for baseball fans: If you could pick any pitcher in history to win one game, who would it be?

It’s sort of the ultimate peak vs. career question.

A starting pitcher’s value is concentrated into about 33–35 games a year, and he has more influence over every game he plays in than the position players arrayed around him. Everyone knows this, thus the just-one-game question. Thus “Spahn and Sain and Pray for Rain” and “Johnson and Schilling and Rain God Willing.” Or one they should have coined, “Viola and Blyleven and Pray for Game Seven.”

On the other hand, teams like that are somewhat rare. Is peak pitching as big a differentiator as the conventional wisdom says it is?

Scaling the Peak

So how do we know what difference a high-peak pitcher makes? Or if a high peak makes a difference?

To figure it, we’d probably want to know how much difference each additional win beyond 81 makes and how much a given pitcher contributes to it. Lots of people have asked and answered this question better than I’m about to. I’m not a statistician and I’m not a database expert. So I’m just going to answer it in a way that I can handle with a couple hours, bb-ref, and a spreadsheet program at my disposal.

What’s It Worth to You?

Say you had a team of 25 precisely average players. Your team would go 81–81 by definition. So the value of each additional win depends on how much closer it gets you to the playoffs. Excluding the strike year of 1981, I took all 372 post-season teams since the first World Series in 1903, and I normalized their records to a 162-game schedule. Then I sorted the teams by the number of normalized wins and counted.

One 82 win team made the playoffs. One divided by 372 equals almost 0.3 percent. One 83 win team also made the playoffs. If you acquire a pitcher who can drag your team to an extra two wins, you’ve doubled your chances to win a pennant. Mostly, the win-by-win gain is incremental.

But when you play out this string, you find that there’s an inflection point as you near 90 wins. Sixteen 88-win teams have made it to the playoffs. The entire sum of teams with fewer than 88 wins who made the playoffs is eleven. Eighty-eight wins is the initial point of inflection, and your odds of making the playoffs go up 145 percent over the 87th win, by far the largest single gain in playoff odds that a team can accrue. You’d still only have a 7.3% chance at a playoff spot, however. From wins 88 to 92, the odds ramp up quickly from about 13:1 postseason chance to a 3:1 shot at a playoff berth, and each win buys a nearly 35% improvement in your odds over the previous win. Thereafter, the only other win total that boosts the odds by more than 20% is the 95th win.

What’s He Bringing to the Table?

Measuring how many wins a player contributes to his team is what Wins Above Replacement (WAR) and its cousin Wins Above Average (WAA) are all about. Roughly every ten runs a player creates or saves is worth one win.

Back to your team of 25 precisely average players. Say each of your five starters goes six innings per start, that’s around 198 innings each. By definition, an average starter generates zero WAA. Now replace one of your perfectly cromulent starting pitchers with Don Sutton. Sutton pitched forever and racked up a lot of career value above replacement (69 WAR for those playing at home). He was consistently above average but rarely great, so only a third of his value (23 WAA) comes from being above average. Given his career rate of performance, in 198 innings Sutton would add about eight-tenths of a win to your team, pushing you from a .500 team to a .506 team.

From the odds we figured above, Sutton takes you from zero playoff spots to about one every 300 years.

Let’s instead replace one of your average hurlers with Kevin Brown. Brown racked up the same career value above replacement as Sutton, but about sixty percent of that value came from above-average performance. Given his career rates, Brown would pull your team up to a .517 winning percentage, a two percent advantage over Sutton in those same 198 innings. Of course, Brown (19 seasons) didn’t last as long as Sutton (23 seasons). If your team reverts to a .500 team for those remaining five seasons, the net winning percentage with Brown decreases from .517 to .513.

But that one to two percent yields about a 600% increase in your playoff odds (from Sutton’s 0.3% at 82 wins to 1.9% at 85 wins). Or flip it around and Sutton is worth merely 16 percent of what Brown is worth to the playoff chase.

Let’s say we were only interested in a single season, and you could choose between the best season in Brown’s career (6.6 WAA) and Sutton’s best year (4.4 WAA) . The difference is about 2.4% in Brown’s favor in terms of their effect on your team’s winning percentage, but the difference between the 85 wins you’ll get with Sutton (a two percent likelihood of making it to October) and the 88 wins you get with Brown (a 7.3% likelihood) is huge. Sutton’s best season is only 26 percent as useful at getting you into the playoffs as Brown’s.

What if we figured this individually for every season in their careers? I did this and then took the career average for each (I counted negative WAA as zeroes). Brown boosted his teams’ playoff odds by about 1.2 percent every year. You might say that over 19 years, he was worth about a quarter of a pennant all by himself. Sutton boosted his teams’ playoff odds by about 0.4 percent each year, or about a tenth of a pennant in 23 years. Sutton’s best season drives you only forty-five percent as far as Brown’s.

If you really want to see an extreme difference, compare Sutton to Pedro Martinez. Pedro’s effect on your team’s winning percentage would amount to only four percent. But his effect on your playoff odds is 600 percent higher than Sutton’s. Or you might say that Sutton only produces sixteen percent of the pennants that Pedro does.

So, how much?

Bill James looked at this matter in The Politics of Glory by comparing Don Sutton and Don Drysdale. Through simulated won-loss records, rather than a holistic approach like WAR, he found that peak pitchers had around a three to five percent affect on having the best record in their league. James didn’t precisely account for the incremental nature of each win, nor for wins below the ninety-win threshold. So his work feels like a low-end estimate to me.

There’s a strong argument for giving greater weight than even five percent to peak based on the importance of each incremental win. Baseball-reference considers an MVP year to be 8 WAR or more. Which means that about 6 of those WAR represent above-average value. Six WAA gets you to 88 wins, our magic point of inflection. Ah, baseball. That inflection point increases your pennant odds by 7 percent. If I moosh it all together, I’d say peak performance is probably around ten percent more valuable than career performance by my way of thinking. I’m not mathy enough to give you a precision-tuned answer, but I think valuing it 55/45 is reasonable, and I’m open to better arguments than mine on either side.

Clearly if I had to choose just one pitcher to win just one game or else my cats would be taken away, I’d want a peakier pitcher, the guy more likely to pitch a great game. But then again, increasing your pennant odds by any amount is something Cubs fans would take any day.

And although I’ve never heard anyone say, “If you could pick any first baseman in history to win one game, who would it be?” this same line of thinking should hold for position players as well.


PS: I’ve adjusted my CHEWS formula to reflect this 55/45 weighting toward pitcher peaks.



One thought on “Pitcher Peaks: Finding a Starting Pitcher Sherpa

  1. I swear I hadn’t read this Sherpa reference before making mine…weird….Rethinking the “Comment Section” idea yet?

    Posted by Geoff Shields | August 5, 2013, 12:58 pm

Tell us what you think!

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Institutional History

%d bloggers like this: