clock menu more-arrow no yes

Filed under:

"Story stats" like Win Probability Added and Leverage Index don't reveal everything

"Story stats" identify and measure high leverage situations, but there are high-leverage situations that exist independent of wins and losses.

FanGraphs
FanGraphs

Recently at FanGraphs, Owen Watson discussed the high leverage found in Win Expectancy Charts as a way to emotionally connect to baseball without a team rooting interest. In brief, the charts, and especially Win Probability Added (WPA) and Leverage Index (LI), "mirror our intuition" regarding the emotional drama in baseball games. He’s right. Watson’s observation is another way to highlight the close relationship between casual baseball consumption and being aware of, let alone scrutinizing, stats like Leverage Index.

WPA, which measures how much an event increases or decreases a team’s chance of winning; it is often called a "story stat" because of its emphasis on context. LI accompanies WPA. LI measures how much pressure a given situation had in terms of the possible outcome—the higher the pressure, the higher the potential WPA. Conversely, statistics such as such as Wins Above Replacement and home runs are context neutral. WAR measures a player’s total value based on his actions without regard for when they happened, and home runs are a simple count of a single type of event, again without regard for the situation.

WAR and home runs tell stories too, but they are usually of a different scale. They tell macro-stories of seasons and careers, while the storied nature of WPA and LI are at the micro-level of plate appearances and games. Such stories can be visualized in Win Expectancy Charts. The hills and valleys of these charts track WPA and LI after each plate appearance. A particularly dramatic chart looks like this:


Source: FanGraphs

This chart is from the Rockies and Padres play-in game that decided the 2007 Wild Card. It was a back and forth affair that went into extra innings, which measurably gave each plate appearance higher stakes.

But Win Expectancy Charts don’t capture everything. Here are the chart versions of seven lopsided, but dramatic, baseball games from 2015:


Source: FanGraphs
Source: FanGraphs
Source: FanGraphs
Source: FanGraphs
Source: FanGraphs
Source: FanGraphs
Source: FanGraphs

Did you notice anything similar in all of those charts? For those still unclear, these games are the seven no hitters thrown in 2015. Rather than creating stories based on context, WPA and LI actually de-contextualize what is most memorable about these games. At best, they indicate that the losing team wasn't really competitive, and the conclusion apparent early on in the game. The reason is because no-hitters are the rare games where victory plays second fiddle. The same would apply to four dinger games, games with 17 or more strikeouts, and any other single game accomplishment that transcends the win-loss ledger.

There is a takeaway from this observation. Context-based "story stats" can help observers parse high leverage situations, but they can’t be relied upon to identify them all for us. And it's not like we need such a thing anyhow. As Watson rightly noted, these metrics reflect what we already know. Watson also notes that WPA and LI "bridge the gap between what’s happening on the field and what we are measuring." Without external observers, however, neither the bridge nor the measurable matter on each side would mean anything. Ultimately, the stories come from the interpreters of what it all means.

★★★

Eric Garcia McKinley is a contributor to Beyond the Box Score. He writes about the Rockies for Purple Row, where he is also an editor. You can find him on Twitter @garcia_mckinley.