Is There a Win Cliff?

Today’s mini-study is inspired by Dane Dunning. He was a waiver-wire option for me in several leagues this past week. With no consideration to his talent (4.71 ERA, 3.38 xFIP), I was not interested because he was averaging just 4.6 innings per start and had only reached the five innings needed for a Win in half his starts. Those seemed too low for any chance for a Win. It got me wondering if a possible cliff exists somewhere with how far a pitcher goes into a start that just eliminates their Win chances. Instead, I found nothing.

First off, I know many factors are relevant for a pitcher to get a Win and most he has no control over. His surrounding cast must score more runs than the other team. Also, they have to field balls in play to get outs while limiting errors. Also, the bullpen must be decent enough to hold a lead. And then there is the opposing team’s starter, hitters, and bullpen. The deal is that none of the preceding factors matter if the starter doesn’t consistently throw five innings.

While innings pitch on the surface seems to be a conditioning factor, it’s also a measure of talent. Can the pitcher get at least 15 outs before his manager pulls him? While there are some exceptions, usually the best pitchers throw the most innings because they can efficiently get outs. Even with all the factors in play, I still had a nagging feeling there was a minimum threshold to breakthrough.

I thought of several factors that could point me in the right direction. Here are four stat I thought that could simply measure how deep a starter pitches into a game. I examined:

• Innings per Start
• Percentage Five Inning Starts
• Pitches per Batter
• Pitches per Start

One other factor I calculated was how sticky the four factors are from the season’s first half to the second half (min four starts per half). If the stat isn’t predictable, there is no reason to consider it for future decisions. I used pitchers from 2015 to 2020 who had at least 10 starts in a season who only started games. I wanted no relief appearances. When I matched up the first half and second half results, the pitcher needed at least four starts in each season.

Innings per Start

This stat would be the simplest to calculate and use if it happened to showed any signs of a dropoff. After grouping the values several ways, the decline was always linear. No cliff.

The Win chance formula works out to:

Win Chance = IP/GS * 13.4% – 40.2%

Now, for the half-to-half stickiness, the r-squared works out to 0.18. Not that great.

Percentage of Five Inning Starts

Before even beginning the research, I thought there might be something to this factor. In the end, it was one of the nosiest values.

Like with Innings per Start, there was no Win cliff with the decline being linear.

Win Chance = 52.9% * (GS% of 5 IP) – 7.7%

For the half-to-half correlation, it was almost non-existent (r-squared of 0.031).

Pitches per Batter

It’s now time to get a little more granular. Maybe a cliff exists for those pitchers who maximize their pitch inefficiency (i.e. Blake Snell). After collecting the data, there was no correlation between pitches per batter at all. If anything, the fewer pitches per batter meant fewer wins, but the spread is tiny with the max being just 1.5%. The values cluster around the league average.

While there is no correlation to the Win rate, the half-to-half is the best of all the values examined (r-squared is .38).

Pitches per Start

The final group contains the Shane McClanahan’s (74 pitches per start) of the world. The pitchers are production, but they are limited by what their manager thinks is their correct pitch count.

Again, there no cliff with some pitchers able to luck their way into a few Wins. The linear formula is

Win Chance = .0077% * (Pitches per Start) + 35.2%

The stat is a bit sticky with an r-squared of .25.

Conclusion

I hoped to find a set value to eliminate starting pitchers from being rostered because they had no chance for a Win. Instead, I got a Golden Sombrero. Every pitcher, no matter the factors working against them, has a chance to get a Win. There is no threshold to overcome. I just could not find the Win cliff. Now I know and knowing is half the battle.