Resource: Fun with FIP, 2011-2013 | Disciples of Uecker

Disciples of Uecker

We'd like to go to the Playoffs, that would be cool.

One of the issues with Wins Above Replacement (WAR) statistics is that they are typically calculated with statistics that are removed from actual runs scored and allowed. This is especially true with pitching WAR in many cases, which are based around pitchers’ Fielding Independent Pitching (FIP) ratios. In general, this is one of the difficulties of the “stats movement;” the movement away from traditional stats, such as RBI, ERA, etc., has also removed some of the context of the game from analysis. Judging all pitchers by K / HR / BB and Line Drive rates that are not park-adjusted, or adjusted for team defense, takes those pitchers out of their context. In the same way that ERA does not effectively separate a pitcher from his ballpark and team defense, FIP encounters the same blindspots when it is applied to Earned Runs (instead of all runs), and assumes that every pitcher receives average defense.

Resources / Side-By-Side Reading
On Hit Prevention: Randy Wolf is a Better Pitcher than Zack Greinke
2013 Pitching Rankings: Fluctuation
Fielding By Fielding Independent Pitching (2011)
Through 2012 Season: FIPChart

FIPRatio + FIPConstant = Runs Average
Fortunately, FIP is not a “fixed” statistic; it is an “index.” Compare FIP to Runs Allowed, for instance; Runs Allowed is a fixed stat within a baseball league. We can argue about and analyze all the different factors that impact Runs Allowed, but at the end of the day, National League pitchers allowed 9819 runs in 21847.3 IP in 2013. That won’t change until the league uses some form of retroactive replay to judge past games and outcomes. On the other hand, since Fielding Independent Pitching — in its most straightforward form — measures the impact that K, BB, and HR have on a pitcher’s expected ERA or Runs Average, the statistic can be manipulated in several ways:

(1) The basic ratio between K, BB, and HR can be judged (are HR and BB weighed too lightly or heavily? Is K weighed too heavily? What types of pitchers does FIP favor? Are there pitchers that FIP simply will never favor? etc.)

(2) Additional stats can be added (for instance, FIP can be judged alongside a pitcher’s Line Drive rate or Home Run per Fly Ball rate, and things like HBP and IBB can be accounted for, too. As Bill James says, if you want to sweat the small stuff, you can adjust “estimate” statistics until they become as accurate as possible.

(3) Although the standard FIP you read does not park-adjust K, BB, and HR rates (please correct me if I’m wrong), one can isolate park factors and influence FIP accordingly. For example, take a ballpark that suppresses HR to 80% of the league rate, and increases K and BB by 10% beyond the league rate. In this case, a FIPRatio might be: ((13*0.8)HR)+((3*1.1)BB)-(2*1.1)K).

(4) After the K, BB, and HR rates are calculated ((13HR)+(3BB)-(2K)), a “Constant” number is added to convert FIP to an ERA (or Runs Average) scale. This means that one can take (Earned Runs Average) – (FIPRatio) to figure a “FIPConstant,” or (Runs Average) – (FIPRatio) to place FIP on the “Runs Allowed” scale. This step can also be helpful to judge pitchers according to their own park and team defense (i.e., does a pitcher play in a park where errors are more prevalent? Or one where Batted Balls in Play do not impact the game as much?)

Depending on how one wants to use FIP, one can adjust the formula accordingly. This can be quite a useful tool to judge team defenses, team fielding efficiency, and also determine whether a pitcher has received his fair share of defensive support (Zack Greinke is one of my favorite examples of this issue; in 2011, Greinke allowed approximately 18 more runs than one would have expected, given an average defense; in 2012, he allowed 8 more runs than one would have expected. This may sound useless, but this makes his WAR less accurate; if one judges Greinke by his WAR in Milwaukee, that person may misjudge his actual contribution to the team. Saying, “Zack Greinke was one run below average over 171.7 IP in 2011” looks a lot different than, “According to FanGraphs, Greinke earned 3.6 WAR in 2011”). A pitcher might receive “unjust” support from his fielders, but that still impacts the standings and pennant race (no one says, “the 2012 Cardinals deserved the NL Championship because Adam Wainwright underplayed his FIP”).

Here’s a basic calculation of FIP, isolating FIPratio, FIPConstant, and Runs Average from 2011-2013 on the Senior Circuit:

League ((13HR)+(3BB)-(2K)) (Runs Average) – (FIPRatio) (9R)/(IP)
2011 NL 0.72 FIPratio 3.45 FIPConstant 4.17 Runs Average
2012 NL 0.72 FIPratio 3.59 FIPConstant 3.59 Runs Average
2013 NL 0.61 FIPratio 3.43 FIPConstant 4.04 runs average

By isolating these ratios, one can analyze how a specific team and its defense stacks up against the league:

Brewers (Team(13HR)+(3BB)-(2K)) (TeamRuns Average) – (TeamFIPRatio) Team(9R)/(IP)
2011 Brewers 0.50 FIPratio 3.48 FIPConstant 3.98 Runs Average
2012 Brewers 0.67 FIPratio 3.87 FIPConstant 4.54 runs average
2013 Brewers 0.99 FIPratio 3.30 FIPConstant 4.29 Runs Average

Interestingly enough, these “FIPConstant” statistics match up with what we know from defensive efficiency: the club had average defense in 2011 (or so), terrible defense in 2012, and strong defense in 2013. As the rate at which the Brewers record outs decreases, their FIPConstant increases (since they allow more runs as they become less efficient):

Year League DefEff Brewers DefEff League FIPConstant Brewers FIPConstant
2011 .694 .694 3.45 3.48
2012 .689 .672 3.59 3.87
2013 .695 .698 3.43 3.30

This is a basic example of how FIP statistics can be used for different purposes.

Brewers Defensive Distribution
Of course, not only is defensive efficiency unequally distributed across the MLB, defensive efficiency is not evenly distributed within a team. This should make intuitive sense; teams have different types of pitchers with different strengths and weaknesses. One ought not expect a “limit-the-damage,” soft-tossing or sinkerball pitcher to have the same relationship to his defense as a flamethrowing ace; and, one need not expect the flamethrower to have better defense than the soft-tosser (for, the hard-thrower might strike out tons more batters, but also allow more line drives or hard-hit balls). One ought not expect a “slow worker” that allows his defense to rock on its heels to have the same fielding relationship that a “fast worker” that keeps the fielders on their toes. It might be heresy to some, too, but some pitchers might simply be better at preventing hits than others.

Here’s a series of examples from the 2011-2013 Brewers:

Pitcher Independent Defensive Support Runs Average
2011 Greinke -0.12 FIPRatio 4.42 FIPConstant 4.30 Runs Average
2011 Narveson 1.01 FIPConstant 3.55 FIPConstant 4.56 Runs Average
2011 Brewers 0.50 FIPratio 3.48 FIPConstant 3.98 Runs Average
2011 Gallardo 0.55 FIPratio 3.44 FIPConstant 3.99 Runs Average
2011 Marcum 0.70 FIPRatio 3.07 FIPConstant 3.77 Runs Average
2011 Wolf 1.09 FIPratio 3.06 FIPConstant 4.15 runs average

Even with an average defense in 2011, Chris Narveson and Greinke did not receive average defensive support. Not surprisingly, those pitchers ranked as the Brewers’ 4th and 5th best starters. On the other hand, in 2012, two of the Brewers’ best pitchers (Fastballer Mike Fiers and Greinke) received below average defensive support. Of course, the team defense was so bad in 2012 that even Marco Estrada‘s “better-than-team-average” support was notably worse than the league.

Pitcher Independent Defensive Support Runs Average
2012 Wolf 1.52 FIPratio 4.43 FIPConstant 5.95 Runs Average
2012 Greinke -0.56 FIPratio 4.15 FIPConstant 3.59 Runs Average
2012 Fiers -0.05 FIPratio 4.00 FIPConstant 3.95 Runs Average
2012 Brewers 0.67 FIPratio 3.87 FIPConstant 4.54 runs average
2012 Estrada 0.25 FIPratio 3.78 FIPConstant 4.03 Runs Average
2012 Marcum 0.91 FIPratio 3.23 FIPConstant 4.14 Runs Average
2012 Gallardo 0.85 FIPratio 2.95 FIPConstant 3.80 Runs Average

Finally, last year, the defensive support distribution mimicked the team’s Runs Prevented performances, too. Yovani Gallardo and Wily Peralta were the worst pitchers and received the worst defensive support, while Kyle Lohse and Marco Estrada were the best pitchers and received the best defensive support:

Pitcher Independent Defensive Support Runs Average
2013 Peralta 1.13 FIPratio 4.12 FIPConstant 5.25 Runs Average
2013 Gallardo 0.80 FIPratio 3.78 FIPConstant 4.58 Runs Average
2013 Brewers 0.99 FIPratio 3.30 FIPConstant 4.29 Runs Average
2013 Estrada 0.77 FIPratio 3.17 FIPConstant 3.94 Runs Average
2013 Lohse 0.99 FIPratio 2.54 FIPConstant 3.53 Runs Average

Interestingly enough, defensive support does not necessarily link to one specific stat. For example, one might argue that a pitcher ought not to expect solid defensive support if they allow a high percentage of line drives. Yet, 2011 Wolf and 2012 Marcum both received strong defensive support while also allowing a high percentage of line drives:

Pitcher LD%
2011 LD%
Wolf 23
Greinke 22
Narveson 22
Marcum 18
Gallardo 17
2012 LD%
Fiers 26
Marcum 24
Wolf 21
Estrada 20
Gallardo 19
Greinke 18
2013 LD%
Gallardo 26
Peralta 25
Lohse 22
Estrada 20

On the other hand, not one of the Brewers starters that received the Best Defensive Support in a given year had the best line drive rate; in fact, 2011 Gallardo had the best line drive rate and average defensive support, 2012 Greinke had the best line drive rate and below average defensive support, and 2013 Estrada had the best line drive rate and second-best defensive support. Obviously, defensive support means many different things; it can mean solid infield play, it can mean effectively using shifts to convert batted balls in play into outs, and it can mean game-saving catches in the outfield. Since batted balls in play won’t be evenly distributed among pitchers, one ought not expect fielders to evenly make plays for different pitchers.

2014 Preview
The Brewers starters that have been with the club for the longest period of time — Gallardo and Estrada — have track records that suggest that even one given pitcher ought not to expect the same support in any given year. Gallardo enters 2014 after improving his own FIPRatio in 2013, while also receiving terrible defensive support. Estrada’s own FIPRatio declined, but his defensive support improved in 2013. Given that the Brewers front office shifts aggressively, one can ask how different shifts will impact different pitchers. Furthermore, one can look at pitchers’ FIPRatios and ask how they can improve for 2014. For instance, one knows that Lohse will not strike out a ton of batters, and therefore lives on his ability to limit BB and HR damage; FIP might not like him, but he might be able to pitch to a shift most effectively. Finally, the Brewers might keep consistent fielders on their team from one year to the next without maintaining the same defensive efficiency; there is no guarantee that a shifting Brewers club will necessarily be more efficient in 2014 than 2013.

In the “life’s not fair” school of statistical analysis, the balance of 2014 will hang on the pitchers’ abilities to limit the damage, interact with their defense, and maybe receive a little situational luck. I gather no Brewers fan will complain if the Brewers’ pitchers succeed while being a lucky bunch; and if the front office can design (and Ron Roenicke can employ) shifts, maybe the pitchers will be able to outperform their FIP once more.

On Hit Prevention: Randy Wolf is a Better Pitcher than Zack Greinke
Fielding By Fielding Independent Pitching (2011)
Through 2012 Season: FIPChart

Share Our Posts

Share this post through social bookmarks.

  • Delicious
  • Digg
  • Newsvine
  • RSS
  • StumbleUpon
  • Technorati