Monday Liftoff Links

At first, I bristled at seeing a new addition to the alpha-numeric soup that is sabermetric analysis. I didn’t feel that the new stats like ERA- did a better job than ERA+ at describing a pitcher’s relationship to league average, nor did the “62% fewer runs per inning than league average” explanation seem like it would drastically improve the ability of stat-heads to be able to communicate with non-stat-heads. I honestly thought that it was just Fangraphs trying to further differentiate itself from Baseball-Reference by adopting more site-specific stats.

But ERA- does fix some problems with ERA+. Because of the way ERA+ is calculated (lgERA/ERA), you can’t simply look at the difference between two players’ ERA+ and draw meaningful conclusions because they’ll have different denominators. Because of this, I think that the minus stats will eventually become the go-to stat for comparing pitchers and comparing them across seasons.

I think this fix is baby-steps, though, compared to what I see as the main problem with using league average comparison statistics. It compares a given pitcher to a single, fictional player with league average statistics. It doesn’t give you a good idea of how good/bad a player was compared to all the other pitchers in baseball.

For example, if I told you a player had a 226 ERA+ or an ERA- of 45, you’d know that he was very good at preventing runs that year, but from those stats alone, you wouldn’t know where in the pecking order he falls. Was anyone else better that year? (Answer: No…those are Roger Clemens 2005 stats).

Percentiles would be a great way of grouping players and showing their relationship to the rest of the league. In that case, Clemens would have been in the 99th percentile (no pitcher did better). In 2008, Brandon Backe would have been in the 1st percentile (no pitcher did worse). Wandy’s 2009 FIP placed him in the top 10% of all pitchers, and in the top 25% of qualified starting pitchers. It’s a handy metric that I rarely see, and I’d like to see get used more often. It doesn’t allow you to quantify how much better a player was, but I think it gives you a really good shorthand for ranking and grouping them.