Hall: Last week, we talked about sabermetric statistics for hitters. This week, I figured we would tackle pitching. What pitching statistics do you look at when considering pitchers?
Oates: I usually focus on Wins, ERA, and strikeouts.
Hall: Those are probably the three most popular statistics used. Strikeouts is a great stat to know. ERA can tell us a lot about a pitcher’s performance. Wins, on the other hand, is probably the most controversial when it comes to new vs. old school.
Oates: Why? Isn’t winning the most important thing? That’s how you make the playoffs, advance, and win the World Series. Winning is what matters. That’s why I want to know how many wins a pitcher has.
Hall: I agree that winning games is the most important thing, but a pitcher getting a win on his record is not the same as the team getting the win. In any particular game, a starting pitcher will likely have the most influence of any one player on whether a team wins or loses, but a team’s offense, defense, bullpen, and opponent all play large roles in determining whether a team won or lost.
Oates: If the pitcher plays the largest role, then he should get the most credit.
Hall: I think the issue isn’t getting the most credit. The issue is the pitcher gets all of the credit or all of the blame when clearly other players played major roles. Should a pitcher really get credit for a win if he pitches five innings and gives up eight runs just because his team scored nine runs while he was in the game? Or should a pitcher really be given a loss if he pitches nine innings and gives up one unearned run just because the offense could not score a single run? If a pitcher pitches nine shutout innings and leaves the game tied 0-0, then that game does not count at all according to win-loss record.
Oates: I hear what you are saying, but wins still seem like the best way to evaluate pitchers on a game by game basis.
Hall: Have you ever heard of quality starts?
Oates: You mean when a pitcher pitches at least six innings and gives up three runs or less? I’ve heard of it, but if you do that every time you end up with a 4.50 ERA and that is mediocre.
Hall: That’s true. It’s not a perfect stat, but that doesn’t mean it is not better than wins. Keep in mind that the 4.50 ERA you are talking about is the worst a pitcher can do. A pitcher can give up many more runs and pitch an inning less and still receive a win. A hitter could have a 20-game hitting streak and hit between .200 and .250 by going one for four or one for five every day, but in practice that just doesn’t happen. The lowest batting average over the past twenty years for a player with a hit streak of 20 games or more is .313 by Jorge Cantu in 2009. That’s with 178 of those streaks. The same holds true for quality starts.
Oates: How so?
Hall: Going back over the past twenty years, there have been 147 quality start streaks of at least ten games. The highest ERA over those starts was 2.96 by Justin Verlander in 2011. The results are fairly similar for wins, but in 2007 Jeremy Bonderman had an eight game win streak with an ERA of 4.59 during that time. That is not possible with quality starts.
Oates: But isn’t using wins easier?
Hall: If baseball stats were just started today and you were to come up with stats for pitchers would you really use wins? Compare the two definitions. A starter can get a win if he pitches at least five innings and when he leaves the game his team is winning and they go on to win the game while never giving up the lead, or if the pitcher leaves the game behind or tied after pitching at least five innings, but is removed for a pinch hitter and his team takes the lead during that inning and they go on to win the game while never giving up the lead. That’s how a starter gets the win. A quality start is pitching at least six innings and giving up three earned runs or fewer. So which is actually the simple statistic?
Oates: The quality start is when you put it like that.
Hall: There isn’t anything necessarily wrong with wins. They tell you something, but unfortunately what they tell you is very context based and comparing two different pitchers by using their win totals is not a fair way to compare players because of all the variables. You can rely on your old man’s stats, but wins won’t get you too far, get you too far.
Oates: What about ERA? Can we compare players using ERA?
Hall: It’s definitely fair to compare players through ERA. It’s certainly better than wins, and in my opinion it tells you more than quality starts because all of the pitcher’s starts matter. The only issue with ERA, especially over the course of a single season, is can we do better.
Oates: What is wrong with ERA that would need improving on?
Hall: ERA does not do the best job in considering the league a player plays in, the park he pitches in, or the defense around him.
Oates: ERA does factor in defense. It uses errors so that pitchers are only charged for earned runs.
Hall: True, but errors are only one part of defense and they are often subjective. If there is a pop-up and a player misjudges the ball and it lands two feet in front of him, or an outfielder and an infielder have a miscommunication on who should catch the ball those plays get counted as hits even though a fielding miscue led to the player getting on base. And have you ever seen a grounder take a bad hop and hit the fielder making it impossible to field?
Oates: Watch out boy, they’ll chew you up.
Hall: And there can be a wide gap in the number of balls different players get to depending on their range, and that can have an impact on a pitcher’s ERA even though he has no control over it.
Oates: So what should we be looking at?
Hall: You have several options depending on what information is most important to you. For example, there is a statistic called ERA-. It has the same fielding problems as ERA, but it accounts for the park a pitcher pitches in as well as the difficulty of the league.
Oates: What is a good number to have?
Oates: So what are my other options?
Hall: One thing you can do is take the defense out altogether.
Oates: Why would I want to do that?
Hall: We were just discussing the different ways defense can have an effect on the game. By taking defense out we focus on the things we know a pitcher has control of. Strikeouts, walks, hit by pitches, and home runs. Have you ever heard of BABIP?
Oates: I’ve heard of it, but don’t know a whole lot about it.
Hall: BABIP stands for batting average on balls in play. It basically takes an at bat, and figures out the batting average if the ball is hit into the field of play. It takes out the strikeouts, walks, and home runs.
Oates: Why should I know about BABIP?
Hall: Generally speaking, league average BABIP is around .300. Given enough at bats, most pitchers tend to head towards this .300 number.
Oates: So if I started pitching in the majors, my BABIP against me would be around .300?
Hall: No. This only applies to major league caliber pitchers. If somebody had a .400 BABIP against for a considerable amount of time, it probably means they are not deserving of a roster spot on a major league team. Only four pitchers last year had a BABIP over .315 last year and none over Rick Procello’s .344. Conversely, only five pitchers had a BABIP under .260 and none lower than .241 from Jered Weaver.
Oates: So what does that mean?
Hall: For most pitchers, BABIP fluctuates from year to year, indicating that there is a decent amount of luck involved. By taking those plays out and focusing on the plays the pitcher has the most control over, we can use a different statistic to measure pitching. It is called FIP, fielding independent pitching.
Oates: Do you use FIP?
Hall: A pitcher’s FIP is on my list.
Oates: What gets included in FIP and what’s a good score?
Hall: Strikeouts, walks, hit by pitch, and home runs are all given different weights. FIP is weighted the same as ERA so anything under 4.00 is going to be pretty good, and anything under 3.00 is excellent. Last year Gio Gonzalez led the league with 2.82 followed by Felix Hernandez with 2.84 and Clayton Kershaw with 2.89.
Oates: Does a pitcher have control over home runs?
Hall: There is some debate about how much control they have. There is a stat which takes the home runs out of it, called xFIP.
Oates: What does xFIP do?
Hall: xFIP stands for expected Fielding Independent Pitching. xFIP substitutes the home runs allowed with the number of home runs a pitcher would be expected to have allowed based on the number of fly balls allowed. The expected number is based on the league average of home runs per fly balls. Personally, I prefer FIP by itself, but xFIP does attempt to take whatever luck is involved with home runs out of the equation.
Oates: Are there any other pitching stats I need to know?
Hall: One stat people seem to use a lot is WHIP, walks plus hits per inning pitched. It tells you how many baserunners a pitcher allows. Generally, 1.30 is going to be average, with anything below 1.2 very good. It has some of the same luck components we talked about before with BABIP fluctuating depending on the year.
Oates: Which stat do you use the most?
Hall: I tend to use FIP. It is the least complicated, and tends to do a good job over smaller periods of time describing how a pitcher actually pitched compared to ERA.
Oates: One thing we didn’t talk about was innings. All of the stats we have discussed don’t really account for the different value a player has if he pitches a lot more innings than someone else.
Hall: That is accounted for in WAR, which is probably a discussion best left for another day.
Feel free to post any questions below.