Why I Don’t Use WHIP

In an article in early January, I explained the pitching stats we use regularly here at MLBTR.  At the end, I briefly noted that I don’t use WHIP outside of fantasy baseball.  Several commenters inquired about that choice, so I decided a separate article might be helpful.

If you’re reading this, you probably already know that WHIP is (walks + hits) / innings pitched.  Hit-by-pitches aside, WHIP is a measure of baserunners allowed by a pitcher per inning.  In 2020, Zac Gallen allowed 25 walks and 55 hits in 72 innings for a 1.11 WHIP.  The calculation: (25 + 55) / 72 = 1.11.

In briefly researching how WHIP came to be, I found this fun Wall Street Journal article from 2009 by Nando Di Fino.  WHIP was conceived in 1979 by Daniel Okrent, better known as the man who invented fantasy baseball.  Okrent originally called the stat IPRAT – “Innings Pitched Ratio.” It was later renamed to the catchier WHIP.  Though in his 11-year-old article Di Fino wrote that WHIP “is generally accepted as a legitimate baseball statistic,” he also quotes then-Rays director of baseball operations Dan Feinstein explaining why the team did not use the stat.  In Di Fino’s words, this is “mostly because pitchers often can’t control the amount of hits that they give up.”

Sometimes, jamming together a couple of different stats into one can improve its usefulness.  I don’t feel that’s the case with WHIP, because of that hit component.  I’d rather see info about pitcher’s walks and hits allowed separately, because those are two very different things.

A pitcher’s ability to avoid walking batters is a real skill, and that’s why we cite BB% here at MLBTR.  For pitchers with at least 100 innings in a season from 2015-19, the year-to-year correlation of BB% was 0.598.  Knowing a starting pitcher’s walk rate in 2018 gave you a decent idea of what his walk rate would be in 2019.

Strikeout rate is even more of a concrete skill.  K% has a year-to-year correlation of 0.753.  If we know a pitcher’s K% and BB%, then almost everything else was a ball in play.  So let’s talk about batting average on balls in play, or BABIP.  Pitchers control BABIP to a small extent, and for a starting pitcher the year-to-year correlation is just 0.179.  There isn’t that much variation pitcher-to-pitcher in BABIP skill.  (As an aside, home run prevention matters as well, which is why we talk about groundball rate as a skill).

Going back to WHIP, its year-to-year correlation is 0.445.  To the degree that WHIP is repeatable, that is mostly owed to the repeatability of K% (since a K is never a hit) and BB% (half of WHIP).  The repeatability of WHIP is negatively affected by the hit component.

In my opinion, there isn’t a convincing reason to use WHIP.  Resident stat expert Matt Swartz sums it up this way: “If the question is how a pitcher performed retrospectively, actual ERA is the more logical stat to use.  If the question is how a pitcher will perform prospectively, WHIP doesn’t correlate that well with future ERA, and you can get to a better picture by looking at components.”

So, we’ll talk about what a player already did on a the field, and hits allowed are a big part of that.  Trevor Bauer gave up 41 hits in 73 innings in 2021, and it’s a big reason he posted a 1.73 ERA.  I’d rather see his walk rate (6.1%) and BABIP (.215) separated out, because I find that more informative both in considering what he already did and what he will do in the future.  If I simply told you he had a 0.79 WHIP, that would be less informative.

My goal in this post is simply to explain why I personally don’t use WHIP to evaluate pitchers, and those are the same reasons you’ve rarely seen it on MLBTR in our 15 years.  We’re all here because we love baseball.  The stats you look at should be whichever ones increase your enjoyment of the game.  Whether WHIP, WAR, wins, or something else does that for you, there’s no wrong answer.