Happ is Example of Why FIP is Unreliable
By Mike Silva ~ March 3rd, 2010. Filed under: Mike Silva, Sabermetrics.
You all know my feelings on the reliance of advanced metrics. My philosophy has always been you need a balanced approach to decision making and advanced metrics can be part of it, albeit less so, because we are in the infancy of such measures. Also, I feel many of the popular metrics like FIP, WAR, or anything that has a lower case x, may have inherent flaws in the thought process behind the formulas. I am not saying the math is wrong, but the process of thinking has gaps.
I spoke with an agent and sent him to Fangraphs. His client, who had a solid year and doubled his salary this offseason, didn’t fare well in some advanced statistical measures. As a matter of fact, Fangraphs “WAR converted to a dollar scale” stated he owed his ballclub about $3 million dollars for his performance. To say the agent got a good laugh on that one is an understatement.
Today, David Murphy discusses J.A. Happ and his 2009 season which resulted in him being runner up for Rookie of the Year. Last year he was 12-4 with a 2.93 ERA, but his FIP (Fielding Independent Pitching) stated his “true performance” ballooned his runs per nine innings to 4.33 – much less impressive. Many like to say that FIP shows how a pitcher can be “lucky” or “unlucky,” as if any of us can go on the mound and live by the chance a ball will find a glove.
Of course, and this has been stated here many times, advanced metrics are more the candy for fans and blogs than real baseball people. The video interview by Jim Bowden last week showed that fans take more stock into these advanced metrics than teams. In the case of Baseball Prospectus it’s become a profit center. Sure, teams use variations of advanced metrics when evaluating players, Joel Sherman of the NY Post wrote that organizations do use Fangraphs, but no one in the Phillies organization who actually worked with Happ everyday is buying it 100 percent. Take this quote by Phillies pitching coach Rich Dubee:
“You’d like everybody to cut down on walks,” Dubee said, “but you have to look at the walks . . . Are they just some real bad deliveries or are they walks where you are pitching around a guy in a situation? Some walks are justifiable. Some aren’t. He’s got intelligence. He’s got awareness of lineups, hitters, situations. Those all play into the game.”
All along my issue with FIP is the weight of walks and strikeouts. You can make a great pitch, saw off a batter and get him to pop up to third base. Is that luck? Pitching around certain hitters in a lineup leads to walks, but is that a bad thing? Pitching is more than walks, strikeouts, and homers. The beauty of the baseball is the inventors somehow knew where to position fielders where 75% of the time a ball would be hit and turned into an out. Good pitches lead to the ball being hit to those locations.
I think looking at WAR and FIP is interesting, although I am not sure either gives you more than a cursory feel of a player. Until advanced metrics can factor in many of the real world scenarios that impact a walk, strikeout, or other baseball results we take it for what it is.
This isn’t an indictment on the Prospectus or anyone who enjoys advanced metrics. Some individuals are linear thinkers that can’t quantify something unless there is a number attached. That’s ok, because diverse thinking is what makes us advance as a society. The point is for every Zack Greinke “that pitches to his FIP,” there are nine other pitchers on the staff that don’t care or know what it means. Maybe instead of indicting those guys the advanced metric community can advance their evaluation of FIP so that it isn’t skewed to the bias of walks and strikeouts alone.
Beyond Batting Average
Lee Panas of Detroit Tiger Tales was kind enough to reach out to me about a new book he wrote that you can check out over at Lulu Publishing called Beyond Batting Average. I am more of a right brain thinker so advanced metrics are something that doesn’t come naturally. Even though I have my issues with some of its theories, there is value in learning what the community is trying to say. Lee takes introduces fans to sabermetrics with easy to understand explanations and examples. It also illustrates the evolution of baseball statistics from simple traditional measures to the more complex metrics used today. Readers will learn how all the statistics are connected to winning and losing games, how to interpret them and how to apply them to performance on the field. I will be checking it out and hope you will as well.
March 3rd, 2010 at 10:34 am
“Until advanced metrics can factor in many of the real world scenarios that impact a walk, strikeout, or other baseball results we take it for what it is.”
But they do. You just aren’t looking at the right ones.
For example, WPA is what you might be interested in, as it handles the game situation. The next step would be to merge WPA and FIP and break down WPA into K/BB/HR and all other events, to see where it is that Happ was doing great.
He also has a great clutch score.
The point of FIP is to give you ONE perspective, the way OBP or SLG each give you ONE perspective.
But whatever it is that Happ needs to know about his performance, it can be broken down into different components. If he wants to know his FIP relative to the game situation, I can give that to him.
He just needs to ask for Ralph Lauren glasses instead of Armani glasses. They all give you something.
March 3rd, 2010 at 10:37 am
I just want to say that Mike Silva keeps asking the right questions, really good questions. He’s simply using the wrong tool at the wrong time. This is not an indictment of Mike, but for the rest of us for not telling everyone when to use a flat screwdriver and when to use a Phillips.
March 3rd, 2010 at 10:53 am
Mike: If you were to be shown evidence that overall a pitcher’s recent FIP is a better predictor of a pitcher’s future ERA than his recent ERA, would that convince you that FIP is a very useful and successful stat which better gauges what portion of a pitcher’s performance is a matter of luck and teammate fielding and what is a matter of the pitcher’s independent talent? In fact, FIP has been widely adopted in the sabermetric community not because people want to use meaningless stats but because in FIP actually does better predict future ERA than past ERA.
The adoption of stats is a “survivial of the fittest” phenomenon, and FIP survives and thrives because it works. It works becasue it is observing something real about pitching performance, and about baseball — that to a significant degree (not 100% of course, but to a significant degree) the percentage of balls in play that are converted into outs is a function of luck and fielder quality and is only to a limited extent within the control of the pitcher’s personal ability. You are right that some outs (your sawed-off batter example is one) on balls in play are largely the pitcher’s repsonsibility, but the results on many balls in play are more fully a function of the luck of placemetn and fielder performance. How do we know this? Because it is based on actual evidence, which shows that if you look at the results of what happens on balls in play for individual pitchers over large smaples of games there is much more limited variation than there is with respect to the things that pitchers themselves wholly control: walks, Ks, homers. Greg Maddux and Moose Haas both allowed a .286 batting average on their balls in play during their careers. That’s not where the difference in their real talent lies, and this phenomenon has been shown, by carefully studying the evidence across baseball history to be so common as to be fundamentally important principle in understanding pitching. FIP is relevant because it works, becasue it helps explain baseball, and project future performance, more accurately.
March 3rd, 2010 at 3:02 pm
Just weighing in here, Mike- it seems like you see the gap between Happ’s FIP and ERA as proof FIP doesn’t work. I see the likely regression to Happ’s FIP levels by his ERA in 2010 as the proof that it does.
This just happens way too often to simply be chance. I guess the point I’d make is that there are loads of pitchers who have had ERA much lower than FIP, only to see a regression. The number of pitchers who consistently outpitched their FIP with their ERA is- well, much, much lower. So until I see Happ do it consistently, I will remain unconvinced. The burden of history Happ’s to overcome, not the other way around.
March 3rd, 2010 at 3:33 pm
I was about to join in but see that Howard and birtelcom already summed up what I want to say. Lets see happs’s 2010 to see if his FIP worked or not. If FIP was supposed to = ERA every year then the metric wouldn’t have any validity. Its not. If baseball minds really though that Happ’s ERA was more indicaitve of his performance/talent than his FIP then why did the Phillies value Drabek more?
I am not saying FIP is perfect but almost all pitchers with a significant sample size have a FIP that is within .3 of their ERA. 2 ntoable exceptions are Glavine and Vazquez. Glavine’s ERA significantly outperformed his FIP, while Javy’s underperforms. Why? Probably sequence. maybe it has to do with the ability to pitch from the stretch as a walk then a double is much worse than a double then a walk. Maybe Glavine was better at bearing down.
Bottomline is 2/3 a season does not a trend make. History is on the sabermetric’s community side. Odds are Happ’s ERA will regress to his FIP.
As for this player that got a hefty raise even though statistical analysis had him at a value below repplacement. Not all teams make good decisions. Look at your favorite team. Look at the Royals. Also, you didn’t give us the players name or his position. Maybe the team plans to only DH the player as his negative value stemmed from the field. One thing is for sure you stated as fact that the player had a solid year, but then cited that the numbers told a completely diffferent story. maybe he didn’t have a solid year. maybe he was terrbile. Jus throwing that out there.
March 3rd, 2010 at 4:13 pm
“Some individuals are linear thinkers that can’t quantify something unless there is a number attached.”
Yeah, I also tend to think that I can’t write things without words or sing things without notes. What’s your point?
March 3rd, 2010 at 4:14 pm
Lou
I can’t tell you the player, wish I could, but I like to keep who I talk to in confidence.
I will say this, the player is not a scrub, but does walk a few batters, he tends to do it because he pitches around certain hitters.
All great points about FIP.
March 3rd, 2010 at 5:28 pm
“advanced metrics are more the candy for fans and blogs than real baseball people.”
Really? Any evidence of this? I was under the impression that quite a few Front Offices have statistical departments and use advanced statistics quite a bit.
March 3rd, 2010 at 5:34 pm
I think if you read the article its not to the level that many fans and bloggers would like – that is my point.
March 3rd, 2010 at 7:36 pm
These last two points go together as well, I think. There are a lot of teams that do not pay sufficient attention, in my view, to the statistical record. As a result, players who are not worth much in terms of value get signed.
March 3rd, 2010 at 8:57 pm
Mike and those as unsold as Mike, here’s my response on the FIP / Happ issue:
http://www.fangraphs.com/blogs/index.php/j-a-happ-fip-and-situational-walks/
March 4th, 2010 at 2:00 am
Mike said:
Okay. Now can you please define what “level” you are talking about, and exactly what your sources for how much stock teams put in stats are. As I am looking at this right now, you are pretty much a BSer. Quoting one line from a pitching coach, especially one on a famously anti-sabermetric team, does not go a long way towards proving your point. An it’s especially ironic that one of the few guys (Tangotiger) who has had the patience to put up with your BS consults for two major league teams.
March 4th, 2010 at 6:14 am
Nick
I never said they don’t use it, but if you start to read articles about this stuff coming up its not as big an emphasis we everyone is making it on the net.
That doesn’t mean they don’t use it but, and I think Tom would agree with this, there is still a gap in baseball as to how they can implement this throughout the organization.
Getting baseball people, especially those towards the field level, to buy into advanced metrics is not going to be easy I can assure you.
March 4th, 2010 at 12:57 pm
“Getting baseball people, especially those towards the field level, to buy into advanced metrics is not going to be easy I can assure you.”
So you are talking about players & coaches? That I buy.
Why would you want players to buy into “advanced” metrics anyway? Not that it’d be a bad thing, just that it’s unnecessary, really. I mean, people making personnel decisions are the ones who gain real value in using them, and, for the most part, I think they are.
Anyway, do front office types not count as “baseball people”?