From Russell Carleton at Baseball Prospectus on September 13, 2018:
Over the past few weeks, I’ve focused on the mechanics of how WAR(P) is calculated, and specifically on how we account for positionality among “replacement” players. In doing so, I had the chance to look at actual performance of replacement players—broken down into components—and I found some fun stuff.
As a quick recap, I looked at performance as a hitter (and as a runner, though I’m not going to use that much) using a simple linear weights model, pegged to the league average. For fielding numbers, I used my own home-brewed defensive stats mostly for the sake of convenience. A player was a “replacement” at a position if he was not a starter at that or any other position (top 30 in the league in playing time at the spot, measured by innings played, or if he was among the top 265 players in baseball in plate appearances for the season) and if he appeared in at least 45 innings at the position.
Read the full article here: https://www.baseballprospectus.com/news/article/42621/baseball-therapy-mysteries-on-the-bench/
Originally published: September 14, 2018. Last Updated: September 14, 2018.