Skip to content

Reinventing the Wheel: Making Real Sense out of Wins and Losses

July 10, 2009

A while back, I introduced a methodology to create a support neutral win-loss statistic for pitchers—and it showed Matt Cain as being severely underrated by his winning percentage. The problem with the methodology was not only that it took far too long to calculate, but it was messy and inaccurate as well. So let’s clean it up. I’ve devised a way that’s simpler and more precise. Once again, I’ll be using Matt Cain’s (who’s suddenly received run support and wins this season!) 2008 campaign as an example.

Matt Cain is finally stacking up the Wins in 2009.

Matt Cain is finally stacking up the Wins in 2009.

First of all, since this is “support neutral,” that means we’ll be taking RA/9 out of the equation. Forget about it. We’ll be placing Cain in front of a league average defense—that means the amount and type of balls hit into play will be converted into outs at a league average rate. Luckily for us, we don’t have to worry about calculating our own DIPS. You can use DICE, FIP or dERA if you’d like, but my personal preference is Graham McAree’s tRA, which is park-adjusted and based off of run expectancy charts and linear weight values. For those of you unfamiliar with the methodology, here’s a link to the primer.

Cain 2008 tRA: 4.45
League 2008 tRA: 4.77

First, we find the league runs per game in order to determine the run environment. This is done by multiplying League tRA by two.

League R/G: 4.77 * 2 = 9.54

Next, we find the custom exponent for the run environment by using PythagenPat.

Exponent: (9.54) ^ 0.287 = 1.91

The next step is to find what we’ll call the WL ratio—this is comparing the pitcher to a league average (.500) pitcher.

WL = (4.77 / 4.45) ^ 1.91 = 1.14

And the last step is to turn this into a winning percentage.

W% = 1.14 / (1.14 + 1) = 0.533

Now that we have Cain’s Adjusted Win Percentage, we need to find out how many decisions he’ll receive. This is done by finding the league average no decision rate (which hovers between 28-31% of games started over the past couple of seasons). In 2008, 31% of games started were no decisions.

Decisions = 34 * 0.310 = 23

Support Neutral Wins = 23 * 0.533 = 12.52
Support Neutral Losses = 23 – 12.52 = 10.96

What I do here is I like to round up the number with the higher decimal if need be to match the number of decisions. So in this case, we’ll round down the SNW and round up the SNL.

Matt Cain 2008:

SN Wins = 12
SN Losses = 11
SNWL% = 0.522

See? It’s not so difficult after all. All I did was lift a piece of Tom Tango’s WAR method for pitchers and applied it to a league average rate of decisions. Now, if we use this formula to calculate Cain’s career statistics, we can get a much better idea of what records he’s truly deserved over the past few seasons.

2005: 3 – 2
2006: 14 – 8
2007: 14 – 8
2008: 12 – 11
Career: 43 – 29
Career Winning Percentage: 0.597

The nice thing about statistics is that there’s always a way you can improve upon them. That being said, this is a definite step above what I did before—but it’s still an inexact science, and should be treated as such. There’s absolutely no debate as to whether or not traditional win-loss statistics are useful on a single-season basis, because it’s painfully clear that they’re not. Over the course of a lengthy career, however, it gives you a relatively good idea as to how effective a pitcher has been.

Advertisements
No comments yet

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s

%d bloggers like this: