## Estimating Wins Using ERA and Run Support

Chasing wins in fantasy baseball sometimes seem futile, but if pursued in a logical way, they can be gained. Playing sub-par pitchers may increase win and strikeout totals, but they puts a drain on WHIP and ERA. By looking at the pitcher’s talent level and knowing the offense of the pitcher’s team, the chances of getting a win can be determined. The following are formulas to help estimate a pitchers win total.

First, all the qualified starters that didn’t switch teams from 2010 were matched with their team’s average runs scored per game. Then a linear regression was run comparing the player’s ERA, his run support and his actual winning percentage. The following equation was created:

Projected Winning % = 0.112(Run Support)-0.105(ERA)+0.446
with an R-squared = 0.827

With this equation, the expected number of wins can be estimated with just a couple more pieces of data. First, the number of starts that lead to a decision (win or loss) for games in 2010 was 70% with the bullpen getting the rest. Second, the number of GS will have to guesstimated using playing time projections and injury history. With this information, a projected number of wins can be calculated:

Projected Wins = 0.7 * Games Started * Projected Winning %

Going back over the 2010 numbers, the average difference between the number of games won and the predicted number of games won was 1.89 with a standard deviation was 2.24 wins.

For example, here is how Felix Hernandez’s win total would compare if he pitched for different teams during 2010. He was able to get 13 wins with a 2.47 ERA in 34 games with a team that average scoring 3.13 runs a game. With those numbers, he was projected to win 13.3 games. Now if he played for the Yankees and got their run support (5.23), his wins would have been around 18.9. If he had only got just 4.0 runs of support, he would have been closer to 15.6 wins.

Normally, trying to accumulate wins is a tough proposition. With a little knowledge of the pitcher and his team’s offense, the amount of wins the pitcher gets can be somewhat predicted.

Print This Post

Jeff writes for RotoGraphs, The Hardball Times, Rotowire, Baseball America, and BaseballHQ. He has been nominated for two SABR Analytics Research Award for Contemporary Analysis and won it in 2013 in tandem with Bill Petti. He has won three FSWA Awards including on for his MASH series. In his first two seasons in Tout Wars, he's won the H2H league and mixed auction league. Follow him on Twitter @jeffwzimmerman.

Guest
Detroit Michael

I’ve seen a significant correlation between IP/GS and wins too. A starter who can work deeper into games is more likely to earn wins. Some of the information contained in IP/GS is indirectly reflected in ERA, so I don’t know to what degree your formula could be improved by looking at IP/GS.

For example, when he was pitching well for the Rays, Scott Kazmir often posted good ERAs but still ran up high pitch counts that necessitated that he leave the game after six innings — your formula might over project wins for a guy like that.