In the first installment, we discussed two core sabermetric principles that are key to looking at the game differently Run Expectancy and Linear Weights. RE24 and linear weights are not stats in themselves, but they are the building blocks to a lot of what sabermetric thought entails. The idea that the game of baseball can be analyzed empirically and that one can come to logical conclusions from looking at it from that perspective are built on that foundation. Context and value were key to subverting and improving upon the original nineteenth century counting stats. But, these principles are not very useful by themselves. There are yet another set of concepts that aid in sabermetric thinking, luck and regression to the mean.
The best way to explain these concepts is by introducing two of the most important and commonly cited sabermetric stats, fielding independent pitching (FIP) and batting average in balls in play (BABIP). FIP and BABIP are in ways the opposite of each other. One tries to isolate a pitcher’s contribution isolated from defense (FIP) and one tries to measure a batter’s contributions solely on balls hit in play (BABIP).
The idea behind FIP is simple. ERA can be a meaningful stat, but it’s measuringmuch more than just the pitcher’s skill. What ERA measures is the pitcher’s skill, the defense playing behind him, the ballpark he’s pitching in, and the random variance of the universe, or as sabermetricians call it, luck.
FIP was developed by Tom Tango as attempt to isolate a pitcher’s skill independent of fielding. This meant reducing pitching to the Three True Outcomes – home runs, walks and strikeouts – thusly called because they are the three where the defense plays no role in the outcome; they are limited to the pitcher/batter interaction.
Tango calculates FIP as follows:
FIP = (13*HR+3*BB-2*K)/IP
FIP is a meaningful metric because it is somewhat of a constant. All the other inputs into ERA (defense, strand rate, park effects) are quite chaotic and change from game to game, month to month, and from season to season. A pitcher’s true skill would theoretically would stay pretty stable through the course of his career.
Take for example, Tyson Ross who has pitched in Oakland and San Diego. See how his ERA outperforms his FIP.
Season | Team | IP | K/9 | BB/9 | HR/9 | BABIP | LOB% | GB% | HR/FB | ERA | FIP | xFIP | WAR |
2010 | Athletics | 39.1 | 7.32 | 4.58 | 0.92 | .310 | 65.5 % | 53.1 % | 12.1 % | 5.49 | 4.30 | 4.00 | 0.0 |
2011 | Athletics | 36.0 | 6.00 | 3.25 | 0.25 | .299 | 76.2 % | 47.6 % | 3.1 % | 2.75 | 3.14 | 3.89 | 0.8 |
2012 | Athletics | 73.1 | 5.65 | 4.54 | 0.86 | .360 | 64.0 % | 49.6 % | 10.4 % | 6.50 | 4.80 | 4.90 | 0.2 |
2013 | Padres | 125.0 | 8.57 | 3.17 | 0.58 | .282 | 71.5 % | 54.9 % | 8.2 % | 3.17 | 3.20 | 3.43 | 1.9 |
2014 | Padres | 195.2 | 8.97 | 3.31 | 0.60 | .291 | 75.1 % | 57.0 % | 11.3 % | 2.81 | 3.24 | 3.11 | 3.2 |
While another great pitcher, Yu Darvish, pitching the majority of his games in Texas has had his ERA underperform his FIP two of his three season in the Majors.
Season | Team | W | L | SV | G | GS | IP | K/9 | BB/9 | HR/9 | BABIP | LOB% | GB% | HR/FB | ERA | FIP | xFIP | WAR |
2012 | Rangers | 16 | 9 | 0 | 29 | 29 | 191.1 | 10.40 | 4.19 | 0.66 | .295 | 70.5 % | 46.2 % | 9.1 % | 3.90 | 3.29 | 3.52 | 4.4 |
2013 | Rangers | 13 | 9 | 0 | 32 | 32 | 209.2 | 11.89 | 3.43 | 1.12 | .264 | 83.9 % | 41.0 % | 14.4 % | 2.83 | 3.28 | 2.84 | 4.4 |
2014 | Rangers | 10 | 7 | 0 | 22 | 22 | 144.1 | 11.35 | 3.06 | 0.81 | .334 | 78.4 % | 36.3 % | 8.6 % | 3.06 | 2.84 | 2.96 | 3.7 |
Those other factors that affect ERA are called Peripherals – the inputs that go into a stat to reach a final ERA number. A pitcher’s ERA can be helped by having All-Star caliber defense behind him, playing in a true pitcher’s ballpark, or just having a manager who knows when to call the shift. Inversely, a pitcher’s ERA can be harmed in hitter-friendly parks or a defense having an off day. FIP allows the caliber of the pitcher himself to be more closely calculated.
BABIP works in the reverse manner to FIP. Everything except the three true outcomes – home runs, walks and strikeouts – are measured. What BABIP attempts to measure is how much luck and random variance is inherent in the game of baseball, but it also takes talent and skill into account.
BABIP was developed by another pioneer in sabermetric thought – Voros McCracken. Much like ERA, batting average seems to fluctuate from year to year and the goal of BABIP is to try and isolate what role luck plays in these seemingly random variations in batting average from year to year.
BABIP is calculated like this: BABIP = (H – HR)/(AB – K – HR + SF)
It is important to not forget that since batting average is calculated completely without context. A seeing-eye single up the middle and a booming opposite field home run have the same value. It is key to investigate how and why a player’s average rises and falls.
A player like Troy Tulowitzki who plays in a spacious ballpark at home and makes a lot of good contact would naturally have a high BABIP, which would drive his batting average up.
Season | Team | G | PA | HR | R | RBI | SB | BB% | K% | ISO | BABIP | AVG | OBP | SLG | wOBA | wRC+ | BsR | Off | Def | WAR |
2006 | Rockies | 25 | 108 | 1 | 15 | 6 | 3 | 9.3 % | 23.1 % | .052 | .314 | .240 | .318 | .292 | .269 | 46 | 1.9 | -5.8 | -3.2 | -0.5 |
2007 | Rockies | 155 | 682 | 24 | 104 | 99 | 7 | 8.4 % | 19.1 % | .189 | .335 | .291 | .359 | .479 | .364 | 109 | 0.4 | 8.4 | 22.2 | 5.2 |
2008 | Rockies | 101 | 421 | 8 | 48 | 46 | 1 | 9.0 % | 13.3 % | .138 | .289 | .263 | .332 | .401 | .321 | 83 | -4.1 | -13.3 | 4.4 | 0.5 |
2009 | Rockies | 151 | 628 | 32 | 101 | 92 | 20 | 11.6 % | 17.8 % | .256 | .316 | .297 | .377 | .552 | .395 | 132 | -1.2 | 23.6 | 9.7 | 5.4 |
2010 | Rockies | 122 | 529 | 27 | 89 | 95 | 11 | 9.1 % | 14.7 % | .253 | .327 | .315 | .381 | .568 | .406 | 140 | -0.2 | 25.1 | 12.1 | 5.6 |
2011 | Rockies | 143 | 606 | 30 | 81 | 105 | 9 | 9.7 % | 13.0 % | .242 | .305 | .302 | .372 | .544 | .389 | 133 | -4.4 | 18.6 | 13.9 | 5.4 |
2012 | Rockies | 47 | 203 | 8 | 33 | 27 | 2 | 9.4 % | 9.4 % | .199 | .284 | .287 | .360 | .486 | .364 | 113 | 0.3 | 3.4 | 1.3 | 1.2 |
2013 | Rockies | 126 | 512 | 25 | 72 | 82 | 1 | 11.1 % | 16.6 % | .229 | .334 | .312 | .391 | .540 | .400 | 141 | -1.6 | 22.1 | 11.9 | 5.3 |
2014 | Rockies | 91 | 375 | 21 | 71 | 52 | 1 | 13.3 % | 15.2 % | .263 | .355 | .340 | .432 | .603 | .444 | 171 | 0.3 | 30.2 | 6.8 | 5.3 |
BABIP is important because it does as good a job of telling how much of batting average is influenced by all the said peripherals mentioned in the previous section.
David Ortiz is another great hitter, but he plays in a very different hitting environment to Tulowitzki. The cramped Fenway Park and many other AL East band boxes has a very different BABIP picture.
Season | Team | G | PA | HR | R | RBI | SB | BB% | K% | ISO | BABIP | AVG | OBP | SLG | wOBA | wRC+ | BsR | Off | Def | WAR |
2003 | Red Sox | 128 | 509 | 31 | 79 | 101 | 0 | 11.4 % | 16.3 % | .304 | .292 | .288 | .369 | .592 | .401 | 145 | -3.1 | 25.4 | -10.9 | 3.2 |
2004 | Red Sox | 150 | 669 | 41 | 94 | 139 | 0 | 11.2 % | 19.9 % | .302 | .322 | .301 | .380 | .603 | .408 | 147 | -3.8 | 36.0 | -15.5 | 4.2 |
2005 | Red Sox | 159 | 713 | 47 | 119 | 148 | 1 | 14.3 % | 17.4 % | .304 | .303 | .300 | .397 | .604 | .418 | 157 | -4.1 | 45.7 | -17.0 | 5.3 |
2006 | Red Sox | 151 | 686 | 54 | 115 | 137 | 1 | 17.3 % | 17.1 % | .349 | .270 | .287 | .413 | .636 | .427 | 157 | -4.0 | 46.6 | -15.3 | 5.3 |
2007 | Red Sox | 149 | 667 | 35 | 116 | 117 | 3 | 16.6 % | 15.4 % | .290 | .355 | .332 | .445 | .621 | .449 | 175 | -4.4 | 58.1 | -16.6 | 6.3 |
2008 | Red Sox | 109 | 491 | 23 | 74 | 89 | 1 | 14.3 % | 15.1 % | .243 | .270 | .264 | .369 | .507 | .372 | 124 | -2.0 | 12.3 | -11.8 | 1.7 |
2009 | Red Sox | 150 | 627 | 28 | 77 | 99 | 0 | 11.8 % | 21.4 % | .224 | .262 | .238 | .332 | .462 | .342 | 100 | -3.1 | -2.8 | -16.0 | 0.3 |
2010 | Red Sox | 145 | 606 | 32 | 86 | 102 | 0 | 13.5 % | 23.9 % | .259 | .313 | .270 | .370 | .529 | .382 | 134 | -6.2 | 18.3 | -14.9 | 2.5 |
2011 | Red Sox | 146 | 605 | 29 | 84 | 96 | 1 | 12.9 % | 13.7 % | .246 | .321 | .309 | .398 | .554 | .407 | 154 | -8.4 | 30.0 | -14.9 | 3.7 |
2012 | Red Sox | 90 | 383 | 23 | 65 | 60 | 0 | 14.6 % | 13.3 % | .293 | .316 | .318 | .415 | .611 | .425 | 170 | -5.6 | 25.5 | -9.6 | 3.0 |
2013 | Red Sox | 137 | 600 | 30 | 84 | 103 | 4 | 12.7 % | 14.7 % | .255 | .321 | .309 | .395 | .564 | .400 | 152 | -8.3 | 27.0 | -15.5 | 3.4 |
2014 | Red Sox | 142 | 602 | 35 | 59 | 104 | 0 | 12.5 % | 15.8 % | .255 | .256 | .263 | .355 | .517 | .369 | 135 | -7.0 | 16.2 | -14.2 | 2.3 |
You can do a lot more damage putting less balls in play playing in smaller more hitter friendly parks. It is harder to successfully put a ball in play in smaller ballparks because the defense has less ground to cover.
Another intendant topic that relates to both FIP and BAIP is Sample Size. It is very easy to have a high batting average support by an abnormally high BABIP for a few months. But over time, the larger the sample size gets the tendency is for player’s skill to regress towards his career average. There are obviously some exceptions to the rule, but a good player will stay good and a bad player will struggle to improve.
BABIP and FIP are interesting stats because they are great opportunities to understand how luck, regression to the mean, and sample size impact player performance over both the short and long term. They are also key to sabermetrics because they aim to isolate different variables in player performance. They put the role of the actual playing environment as dependent variables while positing player skill as the sole independent variable. Sabermetrics is often little more than thinking about baseball critically and two simple stats like BABIP and FIP are key to establishing that mindset.