A (Perhaps Not Reasonably) Short History on How Pitchers are Evaluated
Note: if you are just interested in learning about the stats themselves, you can skip to the next section. But the history is so fascinating!
For a long, long time in baseball history, evaluating pitchers was easy. A pitcher had a win-loss record, and he had an ERA, and by and large this was sufficient. It was the pitcher who gave up or prevented runs, after all, and the ability to prevent runs directly allowed his team to win games. The world was simple.
But certain people in history have never been content with simple answers - they recognize complexity and they ask questions because they have an innate curiosity that is not satisfied by answers whose primary merit is their mere convenience. Sabermetrics, for good or ill, was born of discontented people. And the beautiful simplicity of baseball everyone had known for 100 years or more was upturned. The reason is because what had always seemed so simple really was not, and concepts that people struggled with continually actually were quite simple. So perhaps, rather than altogether replace the simplicity of baseball for esoteric formulae which only robots can compute and comprehend, sabermetrics merely is reminding us to take stock of what things really are simple, if only we will think about them honestly for a few brief moments.
And so we come to a discontented man named Voros McCracken, whose ideas turned the pitcher evaluation world on its head. The irony is, all of his revelations were actually completely...simple. So straightforward and easy to understand the truth of, that it's a wonder the revolution didn't come so much sooner. And his central idea was this: once a batter puts the ball in play, the pitcher no longer has any control over what happens next. McCracken was a baseball man, grew up with the same ingrained ideas of the game that everyone else had, and he didn't even want to believe it himself. Let him tell you:
I swear to you that I did everything within my power to come to a different conclusion than the one I did. I ran every test, checked every stat, divided this by that and multiplied one thing by another. Whatever I did, it kept leading back to the same conclusion:
There is little if any difference among major-league pitchers in their ability to prevent hits on balls hit in the field of play.
Once the batter hits a ball, the pitcher cannot control whether the SS will be able to reach it, whether the 1B will be able to knock it down, whether the CF will get a good read or a good jump on the ball. Even 100 years ago, anyone who had ever watched baseball would have agreed with these statements. The best pitchers still give up hits, and plenty of them. And so in the end, we find that once the ball is put in play, it finds a hole the same amount of the time whether the pitcher was Randy Johnson or Ron Villone.
What does this mean for how we evaluate pitchers though? What's wrong with wins, losses, and ERA? I won't trouble with an explication of the flaws of wins and losses as I am sure we are all more than aware of why they're terrible for any kind of evaluation of a pitcher's effectiveness. But what's wrong with ERA?
Actually, aside from a few quibbles, ERA really is not that bad. It can be (and has been) improved upon, but it's full of history and does it's job reasonably. Just, perhaps, not on its own. The issue with ERA, the one that concerns us today and is most relevant to this article, is that ERA is not properly a statistic for the pitcher. Because once the ball is in play, everything that happens next is actually in the hands of the defense behind the pitcher. The famous question was therefor: where does pitching end and defense begin? Or vice versa? Enter pitching component stats.
Pitching Component Stats
Component stats, by definition, are meant to help us break down what parts of a player's performance are due to his own skill. Not luck, not the players around him, etc. For a pitcher, these stats are, in essence: strikeout rate, walk rate, HR rate, and I would add groundball rate. For short, these things may be referred to as a pitcher's peripherals. A pitcher controls these things himself, independent of the defense playing behind him, at least to a very very high degree, and that is why the use (of the first three) in evaluating a pitcher is termed DIPS theory (Defense Independent Pitching Statistics).
Why do we do need these? After all, aren't the runs allowed the bottom line? Yes, they are, and here is what is interesting about the peripherals: they are better predictors of ERA than ERA is. If you know a pitcher's peripherals, you have a better idea about how good his ERA will be moving forward than you would if you just knew his current ERA. This discovery was revolutionary! FIP and xFIP, which will be detailed by another of our standout writers, are built upon these peripheral stats because they factor out elements of luck and teammate ability and get to the heart of pitcher skill.
Strikeout rate can be measured in two ways: K/9 or K%. K% is a slightly better measure, but K/9 is much more common and is nearly as good, and is probably the one you will encounter most. One good thing about K rate is that year-to-year it has one of the strongest correlations of any single pitching metric (besides things like fastball velocity, which don't measure an aspect of performance). That means that K rates are definitely measuring something is purely a skill of the pitcher.
With K rates, intuitively, higher is better. You want your pitchers to strike people out, because as mentioned, once the ball goes into play, it has about a 30% chance to become a hit no matter who threw the pitch.
Strikeout rate doesn't tell you everything on its own though. Reds fans ought to know this lesson from having watched Edinson Volquez. Volquez had great K rates, but he achieved them by not throwing enough strikes, so he gave up a lot of baserunners and a lot of runs without going deep into games. To work best for evaluating pitchers, K rates have to be paired with...
Walk rate again can be measured in two ways: BB/9 or BB%. Again BB% is slightly better, but BB/9 more common and nearly as good. Walk rates also have a very strong year-to-year correlation, meaning again that is is definitely measuring something that is a pitcher skill.
With walk rates, intuitively, lower is better. As we know, walks will haunt. High K rates can come at a price (like the Volquez example), but there is no caveat about low walk rates. Lower is always better in every way.
Home Run Rate
Home run rate is typically measured by the stat HR/FB, which is home runs allowed per fly ball allowed. This stat, in most cases, is used to determine a pitcher's luck, actually. If a pitcher has a very high HR rate in one year, we most likely expect it to return to a normal level the next year, and the same for a very low HR rate. However, certain pitchers are more prone to HRs, and Reds fans have a good example in Bronson Arroyo.
Intuitively, lower HR rates are better. If a pitcher has a low HR rate, he probably had a good overall year. A small part of HR rate is skill, but mostly it is luck (and home ballpark if you play in an extreme park like Petco). This is why if a pitcher has an unusually low HR rate, we might expect him to regress in the following year, by which we mean that he is not likely to get so lucky on HR allowed again, and more of his flyballs will end up being HRs.
Groundball rate can be measured by the stats GB% (groundball percentage) or GB/FB (groundballs per flyball). GB% is better. GB% actually has the highest year-to-year correlation for pitching metrics, meaning it definitely is a skill of the pitcher.
Perhaps it is not quite as intuitive as the other peripherals, but almost all of the time, you would prefer to have a higher GB% if all else is equal. More groundballs means fewer flyballs, which means fewer opportunities for your opponents to hit HRs. More groundballs also means more double play opportunities, which is good. The downside to groundballs is that they are more likely to become hits than flyballs are (though less likely to become extra-base hits). This is why a team that plays in a huge park or a team that has really bad infield defense might actually prefer more of a flyball pitcher sometimes. But generally, the tradeoff of fewer HRs and more double plays is usually worthwhile for extra singles and occasional doubles. A higher GB% is good.
GB% can be a good indicator of whether a pitcher is a good fit for a certain team. A flyball pitcher who has a good ERA for the Padres might be someone the Reds want to scrutinize more closely. The Eric Milton experiment may have been avoided if the Reds FO had paid attention to the fact that Milton was an extreme flyball pitcher coming to a bandbox home stadium. On the other hand, a team like the Tigers might find flyball tendencies more palatable, as their park is somewhat spacious and their infield defense in recent years has been horrible. If you assemble a groundball heavy pitching staff, you definitely want to complement it with the best infield defense you can manage.
So What Represents "Good" Peripherals?
K rates and BB rates cannot be examined in isolation, as the tradeoff for Ks is often walks, for many pitchers. So to tell whether a pitcher has good K or BB rates we can look at K/BB for simplicity. Anything less than about 2.5 K/BB is danger territory, and really you would like to see about 3 K/BB or better. Over 4 K/BB is very good to elite territory, and probably an effective pitcher regardless of what their other numbers might be. Cliff Lee is a freak of nature in this regard: his K/BB last year was nearly 7! A tables of the quartiles for K/BB for last year's qualified starters is below.
Best K/BB, career: Curt Schilling, 4.38
Best K/BB, season: Cliff Lee, 10.28 (for non-strike-shortened seasons)
Best K/BB, Reds career: Aaron Harang, 3.19
Best K/BB, Reds season: Joe Nuxhall, 4.33 (for non-strike-shortened seasons)
In general, if you assume that a pitcher has an average walk rate, then K rates can be roughly categorized as follows: 9+ K/9 is excellent, 8+ is very good, 7 is about average, 6 is weak, less than 6 probably won't last in MLB. For relievers, adjust these number up some.
Best K/9, career: Randy Johnson,10.57
Best K/9, season: Pedro Martinez, 13.22
Best K/9, Reds career: Jim Maloney, 7.88
Best K/9, Reds season: Jim Maloney, 9.53
For BB rates, and assuming a pitcher has an average K rate, they can be roughly categorized as follows: less than 2 BB/9 is excellent, 2.5 BB/9 is good, 3 BB/9 is around average, 3.5 BB/9 is dangerous, 4+ BB/9 is bad. Again, relievers might be held to a slightly higher standard, but not as much adjustment as for the K rates. Relievers can survive a little better with poor walk rates - see Carlos Marmol for example.
Best BB/9, career: Cy Young, 1.11
Best BB/9, season: Carlos Silva(!), 0.43(!)
Best BB/9, Reds career: Red Lucas, 1.55
Best BB/9, Reds season: Red Lucas, 0.74
I think K and BB rates are perhaps the best way to evaluate minor league pitchers (once they are past the lowest levels) since league environments, strength of competition (and teammates), and ballparks can vary so much that ERA is almost pointless.
For HR rates, the MLB average is right around 10% HR/FB (that is, 10% of all flyballs turn in to HRs). If a pitcher is much better or much worse than this he will probably revert the next year to being closer to 10%, unless he plays in an extreme hitters' or pitchers' park. As a rule of thumb, if a pitcher has a HR rate less than 7% he is likely to see his number of HRs allowed jump noticeably next year because he was lucky, and if he has a HR rate more than 13% it will probably come back down with better luck next year. Use this stat to gauge whether you can trust a pitcher's ERA or whether it came from unusual luck.
For groundball rates, 50% serves as a good baseline and a nice round number. If a pitcher has a GB% of 50 or more, then he is a "groundball pitcher." The biggest groundball starters can surpass 55%; Justin Masterson led last year with 58% groundballs. If a pitcher gets about 45% he is about average. If a pitcher gets into the low 40s or below, he is a flyball pitcher and risks being HR prone.
I realize there are lots of numbers to digest in this article. However, I hope that you've been able to gain some insight on why these stats are useful, and what to generally consider as good, average, and bad.