He is truly a happy man who can, upon all occasions, reconcile himself to his fortune.
-NORMAN MACDONALD, Maxims and Moral ReflectionsThere is in the worst of fortune the best of chances for a happy change.
-EURIPIDES, Iphigenia in Tauris
In a merit-based society, randomness can be a troubling idea. We are raised to believe that a man’s lot in life is his own responsibility; success is a function of only talent and hard work, and failure a lack thereof. Central to the American Dream is the belief that one fully controls the outcomes of one’s life.
This, of course, is a fallacy that turns its back on a fundamental truth known to us for thousands of years: the river of life is crooked, bending to the whims of an uncertain force.
The force has had many names. The Greeks called her Tyche, the Romans Fortuna. The more “scientific” name is randomness. Whatever you wish to call it, fortune dictates many outcomes in life.
Take the most basic model in probability theory: the coin flip. A fair coin, when flipped, should have an equal chance of landing heads as it does of landing tails. Given 1,000,000 flips, a fair coin should land heads very close to 500,000 times, making up 50% of the outcomes. We can feel pretty confident about this. Given 1,000 flips, our confidence level drops. Given 10 flips, our confidence level drops still more. Given one single flip, things get very messy.
Theoretically, our coin should not tend to fall heads any more than it falls tails. Yet, at least on this planet, a fair coin will not fall on its side. The coin will land either heads or tails. Assuming that the physics of the toss are not manipulated in any meaningful way, the true force that pushes the coin to one outcome over the other may simply be called randomness. If we attempt to measure the fundamental nature of the coin after one toss, we will come to wild conclusions (for instance, that this fair coin will land heads 100% of the time). This is the fallacy that comes from drawing inference from an event in which the sample size is too small. In reality, the outcome of the first event has no bearing on the true nature of the coin. If the coin had a 50-50 chance of landing heads before the first toss, the same probability will hold for the second toss. In statistics, this is a handy little trend called regression toward the mean. It allows us to eliminate some of the noise of randomness when attempting to project future outcomes.
As is so often the case, the game of baseball perfectly mirrors life at large in this regard. Randomness creeps into the outcomes of a baseball game all the time, casting lots of varying favor on individual players. It is Fortune who guides an inordinate amount of balls in play to either fall for hits or find defenders for outs. It is Fortune who guides a few extra fly balls over the wall than expected in a month. Fans and analysts alike love to come to grand conclusions after Fortune plays her games in a small sample size. When forecasting future success in baseball, however, we need to account for unsustainable trends.
I should point out one issue I see frequently in attempting to account for mean-regression. It’s called the Gambler’s Fallacy, or the idea that, after a series of extreme outcomes in one direction, an outcome or series of outcomes equally extreme in the opposite direction is likely. For example, one might see a pitcher post a high BABIP through April and assume that he will post a proportionately low BABIP in the month of May. This expected “evening out” of outcomes is a perversion of the idea behind randomness and mean-regression. Since the events considered are independent, the probability of an outcome will revert back to average, rather than dipping in the opposite direction. If a coin has landed tails in ten consecutive flips, it is not “due” to land heads in the eleventh flip. If a pitcher has seen ten consecutive balls in play fall for hits, he is not “due” for an out on the eleventh play. These events are independent, and as such, carry their own true probabilities. That’s what mean-regression is all about. Over a large enough sample, yes, these things will tend to even out. But if there is one thing that randomness teaches us, it is that outliers, including long-lasting outliers, are likely. That’s the trouble with projecting fortune.
That said, here are five Orioles who have either reaped the benefits from or suffered the slings and arrows of outrageous fortune over the season’s first 20 games:
1. Chris Tillman – Chris Tillman has been hurt by the pattern of his balls in play. The 23 year old right-hander has posted the best peripheral statistics of his brief major league career – 7.58 K/9 rate, 2.84 BB/9 rate, and 2.67 K/BB rate – yet is sporting an unseemly 6.16 ERA. This is largely attributable to a .325 batting average on balls in play and some unfortunate sequencing. As a general rule in MLB, pitches put into play will fall for hits about 30 percent of the time. Individual pitchers may exert some small influence on this, as ground balls will tend to fall for hits at a slightly higher rate than fly balls. Beyond that, however, BABIP numbers tend to regress toward the mean over long periods. Tillman’s fly ball tendencies should keep his BABIP on the right side of .300 going forward.
It’s debatable how much influence pitchers exert over the sequence of their outcomes (I tend to think that “some influence” is a fairly liberal, but reasonable assumption). Still, Tillman is doing the two things that pitchers control the most (striking out batters and limiting walks) better than ever at the Major League level. Accordingly, he’s seeing a FIP of 3.68 and an xFIP of 4.03 on the young season.
2. Nick Markakis – The “skill” of getting hits on balls put in play, while largely owing to fortune regardless of the players in question, seems to lie more squarely in the hands of hitters than it does in those of pitchers. Individual hitters generate bat speed, loft, and other mechanical tendencies that remain particular to their skill-sets, so we tend to see larger variance in the batted ball data of hitters. Pitchers, on the other hand, face all types of hitters throughout the year, and the thought is that these things tend to neutralize over a long enough period.
When assessing BABIP for hitters, it’s necessary to account for how often they are hitting the ball on the ground, in the air and on a line. Line drives tend to fall for hits most often, followed by ground balls and fly balls, respectively. So, a hitter’s BABIP should largely be tied to his line drive percentage and career norms. As this article goes to press, Nick Markakis is sporting a 16.7 LD%, not too far off from his career 18.4% and not below his lowest season-long rate (16.6% in 2009). Yet Markakis’ BABIP is sitting at an abysmal .203, down from a career .325 mark. Much of this has to do with an absurd 12% infield-fly-ball-rate, which has never before deviated more than half a percentage point from his career 6.8% mark. All of this is driving his troublesome .208 batting average and .284 OBP. Neutralize his BABIP to career levels and he’d have an OBP around .400. His early BABIP slump will hurt his overall line in 2011, but it’s reasonable to expect his BABIP to creep back to career levels from here on out.
3. Koji Uehara – Just as Fortuna taketh, she giveth as well. Uehara, Baltimore’s best reliever in 2010, has gotten by on smoke and mirrors in his 6 appearances of 2011. Despite allowing 73.3% (!) of his balls in play to be hit into the air, Uehara is yet to surrender a home run, leaving him with a HR/FB rate of, you guessed it, 0%. Not only have Uehara’s balls in play stayed in the ballpark, they’ve met the gloves of defenders with unlikely frequency. Neither his fly ball tightrope act nor his .200 BABIP is a stable trend. He’ll need to keep the ball on the ground more to be successful going forward.
4. Kevin Gregg - On the other side of the reliever fortune scale there is Kevin Gregg, the recently penned, and already much maligned, free agent “closer.” Gregg has already drawn the ire of Orioles fans, having come out of the gate with poor late-inning performances and a deferential attitude. But, not only is Gregg a better pitcher that he’s been so far, his performance alone has probably been a bit better than it seems. Hurting the right-hander is his .389 BABIP and 14.3% HR/FB rate. Expect both of those rates to be lower going forward.
The problem, however, is one of opportunity. Gregg has already pitched close to one tenth of the innings he will probably amass this season, and he’s running out of time to neutralize his trends. Production from relievers is not just volatile because of the opportunity for fortune to rear its head in small sample sizes, but simply because it’s difficult to recover from a rough start when you’re only pitching 70 innings a year. Gregg was never likely to fulfill his 2 year, 10 million dollar contract anyway, but unless manager Buck Showalter adopts a radically different approach to bullpen management, it will be damn near impossible for him now. All of this is why it’s foolish to commit multi-millions and mult-years to relievers.
Mark Reynold's HR/FB rate has dropped from a career 20% to 5%.
5. Derrek Lee and Mark Reynolds – I’ve lumped these two together because they’ve mainly suffered the same fate – an artificially low home run rate relative to the balls they’ve hit in the air. The two right-handed bats acquired this off-season in an effort to bolster the Orioles’ lineup have combined for just two home runs (three after Reynolds opposite field blast in Sunday’s game). Despite hitting fly balls at rates consistent with career norms, Derrek Lee’s HR/FB rate has dropped from a career 16.3% to 4.3%. Reynolds’ numbers have slipped from 20% to 5%. This drop-off is the difference between 5-6 homers for Reynolds as opposed to 1-2. Lee could have been expected to hit 4 or 5 HRs instead of 1. The extra 16 total bases would have brought Lee’s slugging percentage from .296 to .521 and Reynolds’ rate from .306 to .564. With Reynolds making the adjustment to the American League and Lee both returning from thumb surgery and facing a natural age-related decline, it’s reasonable to be a bit worried about either. Still, both hitters should see an adjustment to their HR/FB rates and overall slugging as the season continues.
I’ll check in again at the end of May to see how the fates have aligned for these players and the rest of the Orioles. Again, as the Gambler’s Fallacy shows, extreme positive or negative outcomes are no guarantee of the opposite going forward. We’ll just have to wait and see what Mother Fortune has in store for the Birds.
