For all the use of advanced stats by industry analysts, front offices, and fans, the media has, for the most part, not attached to them. We still see batting average, runs batted in, and home runs on a stat line for a player on TV when he comes to bat. For pitchers, ERA, win-loss record, and strikeout totals are paramount. This results in a large amount of fans who follow baseball, some even very closely, but are unaware of the more recent stats.
There are a lot of misconceptions about advanced statistics in baseball. Many see them as making the game too numbers-oriented and would rather stick to the 'traditional' stats that they grew up with. This is a fallacy, though, because all stats are numbers and baseball, in particular, is a very numbers-oriented game. Sabermetrics, which Bill James referred to as "the search for objective knowledge about baseball," merely attempts to find the best numbers for describing a player's performance.
With winning comes additional fans, and many of these fans don't have a working knowledge of baseball statistics. No doubt that many new readers, lurkers, and commenters here have been intimidated by the statistics that are casually thrown about like Jeff Francoeur's bat in one of his strikeouts.
Thankfully, we're here to help. In the first part of this series, I outlined offensive statistics and some basic points of view for sabermetrics. In this part, I will outline pitching statistics, both good and bad, and explain why they exist and how they are used. Again, community members such as Gopherballs, Jeff Zimmerman, and Scott McKinney are experienced with advanced stats and will probably help if you ask nicely or offer a pop tart. So, without further ado, here we go.
Unlike offensive statistics, pitching stats are far more noisy and it is harder to accurately ascertain the abilities of a pitcher. The reason for this is simple--pitching is merely one component of run prevention, the other being defense. On offense, the batter is the only person being evaluated; no one assists the batter with hitting. However, on defense, there are nine players involved, from the pitch to a fielded ball.
The issue with separating a pitcher's performance from his fielders' performance is one of the primary issues of evaluating pitching. Earned run average, the ERA that we all know and (maybe) love, attempts to do so by separating out 'earned' and 'unearned' runs and presenting it as a rate per nine innings. A run becomes an unearned run if there is an error involved, so an error-prone defense does not hurt a pitcher's ERA despite allowing more runners to cross base. In this way, ERA is useful, as it is an accurate representation of what has actually happened on the field.
However, there are other issues with ERA. Unfortunately, unearned runs count just as much as earned runs, and errors are not always judged the same way. Furthermore, ERA gets complicated by pitchers exiting and entering a ball game. If Danny Duffy leaves one on and nobody out in the 7th inning, Aaron Crow is brought in, and the next batter homers, one run is retroactively charged to Duffy even though he did not allow that runner to cross home plate. For relievers, this gets dicey as well, because if the bases are loaded and Crow gives up a bases-clearing double but ends the inning without allowing more runs to score, Crow's ERA is unaffected at all. This is why his ERA was 0.00 until May 18--he was giving up runs, but those counted against other players.
So, what can a pitcher directly control? Simply put: walks and strikeouts. This is simple enough to grasp; though a pitcher can't control how a player swings (just as a batter can't control where the ball will come in), he can control what types of balls are thrown and where. One can either evaluate this as strikeout and walk percentage (K% and BB%) or strikeouts and walks per nine innings (K/9 and BB/9). I like to use the former for batters and the latter for pitchers, but either will do. Obviously, more strikeouts is good and fewer walks is good, and both are related. For instance, Duffy's current K/9 is 6.80. In 2012, his K/9 was much better at 9.11. However, his BB/9 was also much higher (5.86 in 2012 vs. 3.15 this year), and thus Duffy was actually less effective. In general, you want to see a K:BB of 2 or greater.
Another problem with ERA is that it does not take into account defense. A great defense will suppress an ERA because more outs are being made on the field, while a poor defense will make a pitcher's numbers worse because more balls in play are turning into hits and, therefore, runs.
To counteract this, a stat called FIP, or Fielding Independent Pitching, was born. Essentially, FIP takes the defense entirely out of the equation, looking at walks, strikeouts, hit by pitches, and home runs, and calculates a number based on the ERA scale. While they are very similar, ERA is a counting stat, while FIP is a predictive stat that analyzes what should have happened and is better in large samples. FIP is a better predictor at future performance because it removes as much noise as it can, focusing on how good the pitcher is at what he can control. Another stat, xFIP or Expected FIP, takes FIP and also normalizes home run rate to league average. Home run rates for pitchers fluctuate wildly and are subject to far more luck than most other stats.
There are a few other stats that you can look at when evaluating pitchers. LOB%, or left on base percentage, quantifies the percentage of stranded runners and is also affected by luck; the league average is about 70%. You can look at ground ball percentage and fly ball percentage to see if a pitcher induces more ground balls or fly balls. WHIP, a stat that has gained ground in standard media and stands for Walks and Hits per Innings Pitched, does just what its name implies; WHIPs are in the low ones. Also, just as you can look at an offensive triple slash, you can look at a pitcher's triple slash of opposing batters.
Whatever you do, and I implore you greatly, never look at pitching wins and losses ever again. The pitcher win-loss record is probably the worst stat among any stat in baseball, football, basketball, or hockey. Why? It doesn't mean anything. A win-loss record actually tells you more about everybody but the pitcher, and is only marginally affected by the quality of pitching. It is also meaningless for relievers, because the stat is extremely limited and yet extraordinarily wordy. Please look at Rule 10.17 in the MLB rulebook in order to find out exactly how it works (disclosure: the rule is 700 words long).
Again, please discuss/ask questions below, and I'll wrap this up with a look at defense next week.