/cdn.vox-cdn.com/imported_assets/1024677/Ibn_khaldoun_in_a_maroccan_stamp.jpg)
Ibn Khaldun, the father of modern history, championed the idea that the world of the past operated pretty similarly to the world of the present. For instance, if the Old Testament claims that an army of 80,000 crossed an arid desert, but the generals say in 1380 that there’s barely enough water for them to move 15,000 men across that very same desert, then the number in the Old Testament is probably exaggerated.
Probably the first major revelation that brought me to sabremetrics was that the inverse is true in sports. While Ibn Khaldun asserted that one can understand the future by understanding the present, we baseball fans can understand what is happening and what will happen because people have been playing baseball for a pretty long time. Sometimes there’s Game 162, but mostly the baseball season sputters along like many seasons before it. Some players are great, some players are not, but rarely is a player actually unique, in the grand sense. A single game may be dominated by acts of heroism, momentary lapses of concentration, or just dumb luck, but in the aggregate, all of that evens out and it’s just baseball.
There’s something beautiful about how the vagaries of the moment fall away in the face of a hefty sample size.
But nowadays, we have increasingly granular data. It’s done amazing things for our understanding of the game, and will continue to do so if we’re clever enough; but this granular data comes with a powerful temptation – the urge to describe what happens, to reach an in depth understanding of why what we see differed from what we expected to see, and then to conclude that we’ve found an exception. David Price and James Shields have graciously provided an ideal example over the past two years.
The progression in our understanding of pitching has gone something like this:
- "Hot damn! Some pitchers are good, some aren’t, and who’s who changes rather a lot."
- "There is little if any difference among major-league pitchers in their ability to prevent hits on balls hit in the field of play." - Voros McCracken
- "Well, I guess there are some differences after all. Players can over-perform (or under-perform) their ERA estimators (FIP, xFIP) by not allowing home runs, by stranding runners well, and by inducing weak contact."
- "There are differences, and we can quantify them!" (SIERA) Spoiler: the differences are small.
- "Pitchers actually have lots of control over how hard their pitches are hit, but the spread of their skills is very small, which is why we didn’t notice this previously with the top-down look." -Mike Fast with hit-f/x (my paraphrasing)
So how does all this apply to Price and Shields? Think back to 2010. Price had a shiny 2.72 ERA and 19 wins. We all know that we shouldn't put very much stock in those numbers, but we did. We looked at his HR/FB of 6.5% (9.5% is average) and said, "That's because there's such great movement and velocity on his fastball. Hitters can't square it up and pull it." We looked at his 79.6% strand rate (72% is average) and said, "That's because he saves himself for when he needs to dial the fastball up and get out of a jam." We were fooling ourselves. Aside from the fact that Price is better than most pitchers who came before him, he's not all that different. The following year, in 2011, Price stranded 73.3% of his baserunners, and 9.7% of his fly balls went for home runs, both pretty unremarkable numbers.
We had mistaken an accurate description of what had happened in 2010 for an accurate description of the nature of David Price. Because of that, it was easy to miss the fact that Price has continuously improved himself and became an elite pitcher in 2011. But it was also easy to see. Just look at the big, emotionless number (my emotionless number of choice is SIERA), that understands how pitching has worked in the past, and expects it to continue to work the same way:
Year |
SIERA |
2009 |
4.46 |
2010 |
3.85 |
2011 |
3.27 |
Now take Shields. In 2010, he seemed terrible, with a 5.18 ERA. But he wasn't broken. He struck out a higher percentage of batters than he had in any season since 2007 while walking only slightly more than his norm, and while he did allow more line drives than in 2007 or 2008, it was the same percentage as he had allowed in 2009. No, fueling the collapse was a worst-in-baseball 13.8% HR/FB (with average once again at 9.5%), a meager 68.4% strand rate (72% is still average), and a putrid .341 BABIP (league average is right around Shields's career average of .299).
The blogosphere was full of certainty that Shields's stuff was so bad, and that his character was so weak and shaken that the set of rules which applies to major league pitchers as a whole should not be applied to an evaluation of his special case. People went so far as to assert that he shouldn't be on the postseason roster.
And yet in 2011 Shields was fantastic, with a 2.82 ERA. His HR/FB fell back to only slightly above average and he posted a Prician 79.6% strand rate to go with a very low .258 BABIP. Some here believe that the rebound was because he corrected a flaw in his delivery. He certainly did alter his pitch selection, and he did learn a nasty pickoff move. But there's a dispassionate number that paints a very different picture of random variation acting on a pitcher who has improved himself a little bit each year as he enters his prime.
Year |
SIERA |
2006 |
4.14 |
2007 |
3.60 |
2008 |
3.81 |
2009 |
3.99 |
2010 |
3.57 |
2011 |
3.29 |
I'm working on a project that will describe Price and Shields, along with the other Rays pitchers in gory, granular detail. I do like that stuff. But I think it's important to keep the big number in mind, and to always know when all we're doing is describing. Usually, things actually are how we already know them to be.