About once a week, I am going to take an advanced statistic, and profile and explain it. If anyone wants to see any specific stats explained, just say so in the comments, or on Twitter, and I’ll do my best to get to all of them.
For the first installment I am going to look at two closely related stats, FIP (Fielding Independent Pitching) and xFIP (Expected Fielding Independent Pitching).
These stats are meant as aids for ERA, as they only factor in what a pitcher has control over. The three things that a pitcher has true control over are strikeouts, walks/hit-by-pitches, and home runs. After that, it is in the hands of the defense. The basis of these stats are that one should not debit a pitcher for the actions of his defense, or just pure bad luck.
The origin of FIP comes from one of the pioneers in the sabermetric community, Voros McCracken. McCracken noticed that pitchers ERAs would constantly fluctuate, and he was fascinated by this.
He found that pitchers strikeout and walk rates would stay constant, while a pitchers BABIP (Batting Average on Balls In Play) would fluctuate from year to year, affecting a pitchers ERA.
For an example of this, let’s look at Tim Lincecum.
2008 10.51 K/9 3.33 BB/9 43.9 GB% .304 BABIP 2.62 ERA
2009 10.42 K/9 2.72 BB/9 47.5 GB% .282 BABIP 2.48 ERA
2010 9.79 K/9 3.22 BB/9 48.9 GB% .310 BABIP 3.43 ERA
2011 9.12 K/9 3.57 BB/9 47.9 GB% .281 BABIP 2.74 ERA
So looking at this data, we can see that despite similar GB rates, walk rates, and K rates in the same ballpark, that Lincecum’s ERA and BABIP would jump up and down accordingly. This is where FIP comes in.
The formula, as you can see here, takes into account HRs, BBs, and Ks, and then adds in a constant to put it on the same scale as ERA. Although obviously not perfect, this stat is much more predictive than ERA, while taking into account a pitcher’s true performance.
Lincecum’s FIPs from those years are 2.62, 2.34, 3.15, and 3.17. Although they do jump a bit in 2010 and 2011 due to Lincecum’s lower K rates and higher BB rates those years, it is more consistent from year to year.
While FIP intends to show how a pitcher performed up to that point, xFIP attempts to predict the pitcher’s performance.
Developed by Dave Studeman of The Hardball Times, xFIP takes into account the league average HR/FB rate and multiplying it by a pitcher’s FB%. You can see the formula here.
As you can see in the new formula, the change is how HRs are accounted for, as xFIP regresses the HRs to league average, as they would be expected to do.
Obviously xFIP is assuming that all pitchers are created equally at home run prevention, which is not always true. One stat that tries to counteract that is SIERA, which you can see here.
Well that’s going to be about it for this installment, remember to request certain stats that you would like to see explained, and I will do my best to get to those. For those that do not want to wait for my explanations, Fangraphs.com has an excellent glossary that details most of the commonly used advanced statistics.
11 comments
srt
5/1/2012-10:50am at 10:50 am (UTC -4)
So is there a league average FIP for a pitcher?
A sort of baseline that I could look at to determine by FIP, this guy is probably a poor pitcher – or this guy is probably a top of the rotation type pitcher?
Looking at LIncecum, I’m guessing somewhere around 3.00 or less FIP is very good?
Prismo
5/1/2012-10:59am at 10:59 am (UTC -4)
As far as I know (possibly not much) the idea is to value FIP the same way we value ERA. That’s because it’s essentially what a pitcher’s ERA *should* be given leaguewide fair conditions. So just think of it in terms of ERA…I think.
srt
5/1/2012-11:36am at 11:36 am (UTC -4)
Ah…O.K. Thanks.
SpencerRealDirtyMets
5/1/2012-6:03pm at 6:03 pm (UTC -4)
FIP is on the same scale as ERA, if you look at the formula, it shows that there is a constant put at the end each year that adjusts it as such.
Ceetar
5/1/2012-11:15am at 11:15 am (UTC -4)
Wasn’t ERA more predictive last year than FIP?
FIP also ignores the GB%/LD% thing, because it assumes that pitchers cannot induce weak contact, which isn’t quite true.
It also fails in that it DOES reward the pitcher based on the result of balls in play. A ground ball to Tejada that gets turned into an out benefits the pitcher whereas the same grounder to Jeter that gets through does not reward the pitcher.
Ceetar
5/1/2012-11:16am at 11:16 am (UTC -4)
I’ve been tinkering with the numbers a bit to try to eliminate the ball in play outcome i mentioned above, but I’ve been busy.
SpencerRealDirtyMets
5/1/2012-6:09pm at 6:09 pm (UTC -4)
Could I see some of the work you have? Some feedback couldn’t hurt.
And as far as the above comment goes, I’m just trying to touch over the easier to explain stats first, as “gateway statistics” if you would.
gategem
5/1/2012-6:43pm at 6:43 pm (UTC -4)
I appreciate that you’re taking the time and trouble to explain it. Since I’m basically a simple minded person any help you can provided is appreciated. There is one thing that’s bothering me and this doesn’t apply to you or Ceetar but I mention it in general. When I look at it from a slightly different perspective and research the text books I own on Probability, Random Variables, Stochasitc Processes, etc. I do not in one of them find reference to FIP or XFIP or whatever. That’s because FIP, XFIP or whatever use statistics in their formulae as do many other applications of the mathematics. While I believe in the value of these formulas the people reading this should be aware that there has been much drivel derived from the application of the mathematics. What I object to is calling this advanced statistics (which it is not) and by applying mathematical terms it gives it an obnoxious quality that many take advantage of in a how dare you challenge me attitude.
With that off my chest I once again must say I appreciate what Spencer is doing.
SpencerRealDirtyMets
5/1/2012-7:14pm at 7:14 pm (UTC -4)
I certainly understand that viewpoint, and that’s why I am trying to do two things with this series.
1.) I’m trying to simplify it, often times people will see something like fVWAR and just shit their pants and decide it is above them (not saying that about anyone here, just a generality). TBH the first time I opened up a Baseball Prospectus book I did the same thing.
2.) I’m trying to include examples and links to separate explanations so that people realize that it’s not a dare of their intelligence, and just another way to try to understand the game we love.
As far as the whole textbook thing goes, I’ve only taken a High School level statistics class so I’m no numbers expert, although I will be taking an AP Stat class next year, I can’t wait for that.
gategem
5/1/2012-7:39pm at 7:39 pm (UTC -4)
Just don’t try and explain to Bayonne over at MMO.
He represents the opposite side of the coin.
Acronyms in the military defense field goes back to the age of the dinosaurs and can and does obfuscate things.
One of my favorite lines goes back to pre-historic times and I was taking a course that specialized in the application of these concepts specifically to engineering and the instructor was going on and on and I mentioned to a friend that it was snowing outside to which he replied not as much as it is in here.
Stickguy
5/1/2012-8:12pm at 8:12 pm (UTC -4)
Get off my lawn you damned kids.