WPA/LI by Steve Slowinski February 17, 2010 Before you try and tackle Context Neutral Wins (WPA/LI), make sure you have a solid understanding of both WPA and LI. Got ’em? Okay, cool. So we know that WPA measures a player’s offensive contributions via win expectancy, while leverage index measures the average leverage of all these situations. At the end of a game, not all players will have the same LI — some will have been in more pressure-filled situations than others. A player with a high leverage index may have a higher WPA simply because they happened to come up more often when the game was on the line. So how can we compare two players’ contributions to wins? With this in mind, if we divide WPA by LI, we see how much value a player provided regardless of the leverage. This number is called Context Neutral Wins (WPA/LI) because it neutralizes leverage while still measuring wins added (remember: 1 WPA = 100% win expectancy). WPA/LI is calculated over the course of a season for every at-bat and is then summed at the end of the season to provide a player with their total WPA/LI. It is a good way to compare WPA between players. Again, WPA/LI measures how much value a player added to their team regardless of the leverage. Because of this, it is more a measure of a player’s talent level than WPA. Context: It’s somewhat helpful to think of WPA/LI as a win expectancy version of WAR*, although rankings aren’t exactly the same because context neutral wins are set with zero as average — not replacement level. Here’s a general breakdown: Rating WPA/LI Excellent 5.0 Great 3.0 Above Average 1.5 Average 0.0 Below Average -0.5 Poor -1.3 Awful -2.0 *Although please note, WAR and WPA/LI are far from the same thing — they are just on somewhat similar scales. WAR calculates a player’s value based on offense, defense, the position they play, etc., while context neutral wins focuses solely on a player’s offensive win expectancy contributions. Things to Remember: ● WPA/LI allows us to compare WPA between players, but it is still not as predictive of a measure as wOBA or WAR. Links for Further Reading: Get to Known: WPA/LI – FanGraphs Unleveraging Win Probability – The Book Blog God of Little Things 2008 – Driveline Mechanics