When it comes to drafting a pitching staff, we’ve always spouted two general rules:

- Don’t pay for saves
- Don’t chase wins

Obviously there comes a point when even the most reliever-averse drafter decides to dive in and take a closer, and sometimes it’s a good idea to consider a pitcher’s win potential when deciding to select him.

** CC Sabathia** was a workhorse starter for a very good team and won at least 17 games in five straight seasons from 2007-2011. During that span, he was the pretty much a sure-bet for 17-20 wins.

But what’s the best way to target pitchers with good win potential? Which stats should you look at? How much does his team’s offense, defense, and bullpen impact his chances at winning games? And what about quality starts? Is there a different way to go about projecting how many quality starts a pitcher should have?

I looked back at data from 2010-2012 to answer these very questions.

## Projecting Wins

There are five primary, measurable factors that affect whether or not a pitcher picks up a win in any given outing:

- Does he last at least the minimum five-inning requirement?
- Does he hold the opposition to few enough runs?
- Does his offense back him with enough runs?

- Does the defense play solidly behind him?
- Does the bullpen hold the lead?

These five factors can each be measured statistically. To do so, I’ve chosen the following statistics, each of which corresponds to the number above:

- Pitcher’s IP/Start
- Pitcher’s ERA
- Team’s offensive rank in runs scored
- Team’s defensive rank in UZR/150
- Team’s rank in bullpen ERA

The next task was to see how important each of these factors was to a pitcher’s chance at winning games. To do this I found the relationship between each of these statistics and a pitcher’s win percentage (W%), measured in wins per game started (wins divided by games started). The results are below (provided as r-squared values).

A value of 1.000 would mean a perfect relationship between that stat and W%, however here the best value we get is 0.310 between W% and ERA. This means that 31.0% of the fluctuation in W% can be attributed to fluctuation in ERA.

Now, what exactly are we supposed to do with these numbers? To make them more useful, I’ve converted these five values to percentages to show how you should weight each of them when trying to project a starting pitcher’s win total. Then I summed up offense, defense, and bullpen to yield a cumulative “team” component.

When trying to project a starting pitcher’s win total in a given year, ERA and IP are almost equally valuable and account for about 80% of the pitcher’s expected win total. The other 20% is a combination of the offense, defense, and bullpen backing him, and surprisingly the quality of the bullpen was more important than the quality of the offense! The defense’s impact wasn’t negligible, but it was the lowest of the three.

Consider ** Jon Lester** and

**for a moment. Over the last three years they’ve each made 96 starts with similar innings totals (Lester: 605, Vargas: 611) and ERAs (Lester: 3.85, Vargas: 3.96). Lester has won 43 of his 96 starts, Vargas has won 33. That means Vargas has won 23.2% fewer starts, right in line with our 20.1% from the table above, and he’s played for the inferior team.**

**Jason Vargas**** Felix Hernandez** and

**match up rather closely as well with Verlander winning 59 games to Hernandez’s 40, a 32.2% gap, however Verlander’s ERA over the last three years is about a quarter-run lower than Hernandez’s.**

**Justin Verlander**## Projecting Quality Starts

Aside from marginal impacts that a pitcher’s teammates have, things like making errors to extend innings, quality starts are almost completely within the starting pitcher’s control. Only his average innings per start and ERA really matter.

Like with wins, I did the same thing with quality start percentage (QS%), measured in quality starts per game started (QS divided by starts).

As expected, offense, defense, and bullpen were all irrelevant to a pitcher’s QS%. Innings pitched per start showed a slightly higher correlation than ERA did, which is different than we saw above. With W%, ERA showed a stronger correlation than innings pitched. Here it’s the opposite.

This suggests that when trying to project a pitcher’s number of quality starts, his ability to go deep into games is slightly more important than his ability to prevent the other team from scoring. My off-the-cuff explanation for this is simple; a pitcher can pick up a quality start by giving up three runs in six innings, which would yield a 4.50 ERA. Even if a pitcher is below league average on any given day, they can still pick up a quality start. Their ability to go six-plus innings is of primary importance.

## So, how much should you weigh each category?

It’s almost a 50-50 split, but innings pitched gets the nod.

If your league is converting from wins to quality starts, you need to change the way you value a starting pitcher’s statistics. Relative to ERA, innings pitched becomes about 20% more important when trying to project quality starts rather than wins, so pitchers that go deep into games are better targets than pitchers who do well in a smaller amount of innings.

Completely ignoring ERA, this chart below shows you what innings totals yield what QS% on average. It should be noted that ERA is indirectly included because there aren’t a lot of pitchers with 4.50 ERAs throwing 200+ innings in a season, but just stick with this and you’ll be a-OK!

Happy drafting!

Comments on this entry are closed.