FIP, Selection Bias, and BABIP/Stand - or Why JA Happ Sucked
Apr 17, 2014 11:38:01 GMT -5
mwgray13 likes this
Post by joshv02 on Apr 17, 2014 11:38:01 GMT -5
Or How I learned to Stop Worrying and Learn to Hate Fangraphs
Rather than continue to derail the “outfield roster crunch” thread, I thought I’d put a couple of short paragraphs on this topic in this thread, and let it die as I will likely not add much of anything of substance (I subscribe to the “if you can’t say it in 3 posts, stop trying to say it” school).
As background: I wrote that JA Happ had two bad years in a row from 2011 to 2012 (and was paid $3.7mm in 2013) which was a reasonable explanation as to why he wasn’t claimed on waivers rather than an unspoken gentlemen’s agreement to not claim good players on OAW, while jmei responded/wrote that Happ had a WAR of 1.7 in 2012 and a FIP of 4.01.
Jmei is certainly (partially) right. According to Fangraphs, Happ’s WAR in 2012 was 1.7, and his FIP was 4.01. Both, I think, are misleading. To understand why I think so, you have to know what underlies these stats.
Fangraphs WAR calculation for pitchers is basically just a FIP based calculation converted into runs (e.g., playing time), and then compared to a replacement player. We can quibble about all of these parts, but really the only thing I care about is the reference to FIP. Fangraphs also calculated Happ’s FIP at 4.01. FIP is just = [(13*HR)+(3*BB+HBP)-(2*SO)]/IP+constant (traditionally 3.2). When you do the traditional FIP calculation, you get a 4.12 FIP, not 4.01 – this is, presumably, because the “constant” in the formula should (and does) change annually based on the run environment. However, note that this means that you cannot compare FIPs across seasons without an adjustment, just like you cannot compare ERA’s across seasons without knowing run scoring context. Here, HAPP’s FIP was 2% worse than league average. That would make him roughly a league average starting pitcher, which certainly does not “suck” and would be worth claiming on waivers.
But, you must understand the point of FIP. FIP is useful to the extent that we think that pitchers do not – or have minimal – control over BABIP allowed and Strand Rate, so we look to their FIP to predict their future success. The underlying basis for FIP is essentially Voros’s studies suggesting that historically, few major league starting pitchers have had demonstrated success at maintaining a low BABIP. I’m not really interested in debating that again. What is important to note, however, is its limitation – it was limited to those pitchers with back-to-back 150 IP seasons, which was the definition of “starting pitcher,” and was the baseline in looking for matched pairs. JA Happ has never had a matched pair of 150 IP. He would never fit into Voros’s study. Only those pitchers that demonstrated success – ie., those that survived in that role – would do so. This is all old news, so I’m not saying much that is new.
What matters here is that when confronted with “he had a 1.7 WAR in 2012,” the inquiry should not stop, as it doesn’t mean very much without understanding what that means. For example, when looking at Baseball-Reference (which uses actual runs allowed in figuring out pitching WAR, minus a defensive adjustment, coupled with a role adjustment) Happ had a -1.8 WAR in 2011 and was a little under a ½ win above replacement for 2012 (which is basically replacement). Was he nearly league average in 2012 or was he barely above replacement? When you look at the totality of his stats – his BABIP, his strand rate, his HR rate, his low GIDP rate (b/c of his FB rate), etc. – I think …
Happ sucked in 2011 and 2012
He had a RA of nearly 6, and 5, respectively. And he did so while basically giving up a nearly 800 OPS over two years, and he had a 2 year/9mm contract on his shoulders. For a player like him, I wouldn’t look at FIP without major caveats, and certainly not as the end point to some conversation. In other words, if JA Happ is the answer to “who was the best player to make it through OAW, thus demonstrating that good players also go unclaimed,” I think the answer is pretty suspect – or at least, certainly not very conclusive.
I blame Fangraphs, by the way. Stats can’t be looked at simply casually, even if they are more advanced than Avg/HRs/RBI, which is overly promoted by Fangraphs (who think we have more answers than we do). but, I'm old and crumugeonly, and wonder why things aren't the way the used to be.
Rather than continue to derail the “outfield roster crunch” thread, I thought I’d put a couple of short paragraphs on this topic in this thread, and let it die as I will likely not add much of anything of substance (I subscribe to the “if you can’t say it in 3 posts, stop trying to say it” school).
As background: I wrote that JA Happ had two bad years in a row from 2011 to 2012 (and was paid $3.7mm in 2013) which was a reasonable explanation as to why he wasn’t claimed on waivers rather than an unspoken gentlemen’s agreement to not claim good players on OAW, while jmei responded/wrote that Happ had a WAR of 1.7 in 2012 and a FIP of 4.01.
Jmei is certainly (partially) right. According to Fangraphs, Happ’s WAR in 2012 was 1.7, and his FIP was 4.01. Both, I think, are misleading. To understand why I think so, you have to know what underlies these stats.
Fangraphs WAR calculation for pitchers is basically just a FIP based calculation converted into runs (e.g., playing time), and then compared to a replacement player. We can quibble about all of these parts, but really the only thing I care about is the reference to FIP. Fangraphs also calculated Happ’s FIP at 4.01. FIP is just = [(13*HR)+(3*BB+HBP)-(2*SO)]/IP+constant (traditionally 3.2). When you do the traditional FIP calculation, you get a 4.12 FIP, not 4.01 – this is, presumably, because the “constant” in the formula should (and does) change annually based on the run environment. However, note that this means that you cannot compare FIPs across seasons without an adjustment, just like you cannot compare ERA’s across seasons without knowing run scoring context. Here, HAPP’s FIP was 2% worse than league average. That would make him roughly a league average starting pitcher, which certainly does not “suck” and would be worth claiming on waivers.
But, you must understand the point of FIP. FIP is useful to the extent that we think that pitchers do not – or have minimal – control over BABIP allowed and Strand Rate, so we look to their FIP to predict their future success. The underlying basis for FIP is essentially Voros’s studies suggesting that historically, few major league starting pitchers have had demonstrated success at maintaining a low BABIP. I’m not really interested in debating that again. What is important to note, however, is its limitation – it was limited to those pitchers with back-to-back 150 IP seasons, which was the definition of “starting pitcher,” and was the baseline in looking for matched pairs. JA Happ has never had a matched pair of 150 IP. He would never fit into Voros’s study. Only those pitchers that demonstrated success – ie., those that survived in that role – would do so. This is all old news, so I’m not saying much that is new.
What matters here is that when confronted with “he had a 1.7 WAR in 2012,” the inquiry should not stop, as it doesn’t mean very much without understanding what that means. For example, when looking at Baseball-Reference (which uses actual runs allowed in figuring out pitching WAR, minus a defensive adjustment, coupled with a role adjustment) Happ had a -1.8 WAR in 2011 and was a little under a ½ win above replacement for 2012 (which is basically replacement). Was he nearly league average in 2012 or was he barely above replacement? When you look at the totality of his stats – his BABIP, his strand rate, his HR rate, his low GIDP rate (b/c of his FB rate), etc. – I think …
Happ sucked in 2011 and 2012
He had a RA of nearly 6, and 5, respectively. And he did so while basically giving up a nearly 800 OPS over two years, and he had a 2 year/9mm contract on his shoulders. For a player like him, I wouldn’t look at FIP without major caveats, and certainly not as the end point to some conversation. In other words, if JA Happ is the answer to “who was the best player to make it through OAW, thus demonstrating that good players also go unclaimed,” I think the answer is pretty suspect – or at least, certainly not very conclusive.I blame Fangraphs, by the way. Stats can’t be looked at simply casually, even if they are more advanced than Avg/HRs/RBI, which is overly promoted by Fangraphs (who think we have more answers than we do). but, I'm old and crumugeonly, and wonder why things aren't the way the used to be.
