Every year I have the same strategy when targeting pitchers in my March fantasy baseball drafts: Find guys with high strikeout potential, low WHIPs and completely ignore wins. I’ve always been frustrated by wins as a category because a pitcher’s win total seems to vary so much from year to year.
This offseason it finally hit me that I’ve been employing this strategy based on nothing more than gut feeling and what I had casually observed from years of following the game. But are wins really any less predictable from year to year than ERA, WHIP, strikeouts or even innings pitched for that matter?
To answer this question, number 32 in our Top 100 Offseason Questions series, I went back and looked at how well each fantasy category correlated from 2010 to 2011. I looked at pitchers who tossed at least 100 innings in both 2010 and 2011, and to eliminate any error caused by a pitcher throwing more or fewer innings from one year to the next, I converted each of the four fantasy stats to rate stats (per inning). ERA, WHIP and K/9 are already common rate stats, but I had to convert win totals to wins per inning pitched (W/IP).
There were 103 pitchers who threw at least 100 innings in both 2010 and 2011. I compared their 2010 seasons to their 2011 seasons and the chart shows my findings
Note: "% Change" refers to how much variation the average pitcher had in each category from 2010 to 2011 so, for example, going from a 3.40 ERA in 2010 to a 4.10 ERA in 2011 would represent a 20.6 percent change.
As expected, wins (represented as W/IP) showed almost no correlation from year to year. An R-squared value of 0.009 tells us that if a pitcher threw the same amount of innings in consecutive seasons, 99.1 percent of the variation in their win totals from one year to the next is due to chance. You want to predict how many wins Zack Greinke will tally in 2012? You might as well base your prediction on the color of his eyes (they look hazelish to me).
This statistical observation does come with one caveat, though. We can make reasonable guesses about whether a pitcher’s win total should increase, decrease or stay the same based on how their surroundings have changed. When Doug Fister was shipped from Seattle to Detroit we all figured he’d start piling on wins at a better clip than he had been with the Mariners. That ended up being the case.
Even if a player doesn’t change teams we can still make educated guesses at what his win total should look like. Did the team’s defense improve? Has the lineup supporting him changed? Did the team gain or lose a legitimate closer to preserve games? We can infer a lot, but wins are still largely very random.
Looking at the rest of the numbers above, we see that strikeouts (0.588) and WHIP (0.139) have the two highest R-squared values of the four major pitching categories (ignoring innings pitched). Apparently my strategy for targeting starters is just what I should be doing. If you want to win you should do the same.
You can find related articles at Baseball Professor!