Luck Is Not the Residue of Design In Baseball or In Life

Branch Rickey famously said that "luck is the residue of design." Branch Rickey was a baseball man and luck is part of baseball so it's worth examination in a baseball blog.

You might have noticed that only lucky people make this statement. They want to take credit for their good fortune. Partisan politics aside, it's not possible to view the life of George W. Bush -- who screwed up in college, over-imbibed in alcohol and cocaine in his 30s, bankrupted two companies, got himself removed from a corporate board in his 40s and traded away young Sammy Sosa as president of the Texas Rangers -- without recognizing the immense power of luck. Had he been George Tree, instead of the rich scion of a politically connected family, it seems pretty obvious that he wouldn't have been accepted to Yale, installed as a corporate CEO, served on well-paying boards, been installed atop an MLB team or won election to the local zoning board, much less the presidency.

Indeed, luck and design are utterly independent. Good design can mitigate bad fortune and exploit good, but overwhelmingly bad luck will ruin all but the most adept. Even the best horse-and-buggy makers were put out of business by Henry Ford. 

Which brings us to baseball.

Tigers' third baseman Nick Castellanos, a lifetime .258 hitter, leads the American League in batting average so far this season at a .378 clip. No doubt he is hot: he's already pounded half as many home runs (7) as last year in a fifth as many at bats. He's also batting .449 on balls in play, even though the distribution of the balls he's hit is only marginally better than in the previous two seasons. 

In other words, Catellanos's grounders are dribbling through a hole. His lazy flies are dropping between charging fielders. His line drives are not getting snared. He has been lucky.

There's a fella named Pujols whose batting average (.183) is less than half of Castellanos's. He's not ripping as many line drives as Castellanos but he's killing many more worms, the kind of batted balls that metamorphose into singles at a high rate. Combine his relative turtle-speed with a snoozing patron saint of serendipity and Albert's hitting just .160 on balls in play. That's horrible luck.

Perhaps Branch Rickey would blame design, but Pujols has fanned half as many times as Castellanos, while walking twice as often. That suggests that Castellanos is just plain lucky. We should think twice, for example, before voting him to the All-Star team.

And while on average these trends will regress to middle ground, sometimes they fester all season. In 2002, Cubs infielder Jose Hernandez batted .404 on balls in play for no apparent reason en route to a .288 batting average and an All Star selection. He never again came within 50 points of OPS over a whole season and retired with a .252 batting average.

We can overcome bad luck and allow good luck to slip by unnoticed. But refusing to acknowledge its mighty influence is just silly.

BABIP: From the White House to the Outhouse

Batting Average on Balls In Play (BABIP) burst onto the baseball landscape like a meteor, illuminating life on the diamond in its wake. By measuring the part of the game largely outside the control of the pitcher (and somewhat outside the control of the batter) it appeared to separate skill from luck.

If the Royals' Billy Bulter hits .341 on balls in play in 2012 and Shane Victorino hits .278, might that not suggest that Butler was crazy lucky and Victorino not so much? If we were making predictions about the following year, might we not temper Butler's prediction and enhance Victorino's on the notion that luck will average out?

The answer, it turns out, is: well, maybe. After cuddling up beside BABIP for several years, seamheads started to notice that it had some idiosyncracies. BABIP just likes some guys better than others. Speedsters are particular BABIP sweethearts because they can get on base without hitting the ball hard. Ichiro rocked a .357 BABIP through 2010, but as his speed has waned, he's suffered three sub-.300 BABIPs since.

Worm killers romp with BABIP because ground balls tend to find holes more often than fly balls. That's offset by the fly balls that do make safe landing, which tend to result in extra base hits. Pop-ups are BABIP death and line drives are BABIP medicine, for obvious reason. And batters can control their BABIP to some degree. The aforementioned Ichiro didn't just hit the ball; he often placed it, much to BABIP's delight.

BABIP can start the narrative, but not complete it. Certainly Butler doesn't have foot speed on Victorino, but he launched many more line drives and many fewer pop-ups than Victorino in 2012. That suggested there might be some staying power in those numbers; i.e., there might be some skill involved.

In 2013, Butler did, in fact, maintain an above-average BABIP, losing just 15 points from his 2012 rate. But Victorino, rejuvenated in Boston, "regressed" beyond the mean and up 43 points. Some of that was the result of doubling his line drive rate and much of it was pure serendipity, from a lot of bad in 2012 to a little good in 2013.

It's even more pronounced for pitchers, whose results are affected by the defense behind them, defense that remains largely intact over the course of a season. Rookie of the Year Jose Fernandez allowed a .240 batting average on balls in play, despite fairly pedestrian rates of line drives and pop-ups induced. Fernandez looks highly promising, but BABIP strongly suggests a return to earth in 2014. 

Conversely, all-world starter Justin Verlander played second fiddle to Max Scherzer in Detroit last year, in part because of a high BABIP-allowed (.316). Verlander allowed a lot of line drives, but probably not enough to explain all of BABIP. Likely the Tiger statues at third, short and first contributed to balls getting through. A bounceback may be in order, particularly with Miguel Cabrera moving to first, Prince Fielder and Jhonny Peralta leaving town and Ian Kinsler joining the infield.(Why didn't this matter for Scherzer? A. Luck. B. An 11% higher strikeout rate. Dr. Strangeglove can't flub your strikeout.)

So BABIP is useful, particularly at the extremes, but it isn't quite the revelation originally envisioned. Pure luck is hiding in every facet of the game,but in the nooks and crannies, not out in the open for BABIP to see in its entirety.

Lifting the Veil of Luck

It's appropriate on the weekend of Black Friday to discuss shopping frenzies. To some extent, every team is in one now as they kick the tires on free agents who might help them win in 2014 and beyond.

One of the most important advances in the world of baseball analytics is the ability of franchise brass to evaluate personnel realistically. When baseball GMs determine which free agents might help their teams in 2014, they will view 37-year-old Bruce Chen with a jaundiced eye.

Teams will know that Chen's 9-4, 3.27 effort for Kansas City in 2013 was a chimera. Nothing in his pitching profile suggested that he had a special season except the ultimate results. In 121 innings he fanned just 78. He relinquished just 107 hits, primarily because opponents batted only .255 on balls in play (BABIP) against him, 25 points lower than his career rate and 40 notches lower than league average. The chasm opens wider when examining only Chen's 15 starts; he made 19 relief appearances to begin the season before the Royals summoned him back into the rotation.

In addition, Chen  stranded 78% of runners who reached base against him, significantly above his lifetime and league average rates. Strand rates are rarely a matter of skill. Finally, an unusually low percentage of fly balls against him left the park, often a sign that a pitcher got lucky with wind, field configuration, and the like.  Taken together, the snapshot of Chen's results provides strong evidence that regression troops are arraying on the border and preparing to attack next season. Caveat emptor.

Likewise, Cubs brass won't be relying on Travis Wood to deliver another stellar (literally: he made the All-Star team) 3.29 ERA season in 2014. Wood also enjoyed a low BABIP (.248), high strand rate (77%) and low fly ball/HR ratio. Advanced statistical models suggest his actual performance looked more like a 4.50 ERA.

Instead, Cubs management might hope for bounce-back from Edwin Jackson. Despite an 8-18, 4.98 campaign in '13, Jackson was actually a strong performer. A .322 BABIP against and a frighteningly low (63%) strand rate conspired to denigrate his work. Fangraphs estimates Jackson was worth two wins above replacement, something an ERA of nearly-five would not suggest.

And don't look now, but 2014's Tiger pitching phenom could be Rick Porcello. The 25-year-old righty managed a 13-8, 4.32 line when everything was going against him; it'll be interesting to see what happens when he basks in the sunshine of serendipity. Like Jackson, Porcello suffered a high BABIP and low strand rate in '13, but big flies are what really did him in. One of every seven balls in the air against him cleared the fences, a miserable rate for which he might not be completely responsible. Fangraphs thinks Porcello was more like a 3.50 ERA hurler who was worth 3.5 wins last year.
Over the last four years, Porcello's strikeout rate has risen while his walk rate has held steady. If he can keep the ball in the park just a little better he has a chance to to some damage in 2014.

Remember, regression is a trend, not a guarantee. Jackson could stink up the joint again and Wood could again leave the bases full of runners. It's just not the way to bet.
 

And Now for Something Completely Different

Today we're going to talk about pop-up rate and Z-contact, their inverse correlation to BABIP and their effect on DIPS projections.

Ha! Just kidding! I mean, that's what we'll be discussing , but you won't know it. What you will know is that we've engaged in a new and fascinating discussion that might help you win your fantasy baseball league.

Ha! If anything in this blog were actually fascinating, Braindrizzling would be positively thrumming with annoying but remunerative pop-up ads and flashing banners. Far more likely, you'll find it mildly relevant if you manage to read to the end.

First things first: this blog post owes everything to Dave Cameron and Voros McCracken. It is a recapitulation of this fine post by Cameron at Fangraphs.

Fourth things second: there will be no math and you won't be quizzed on it next week. However, there will be a wee bit of trailblazing that will help us understand yet a skooch better how pitchers succeed or fail.

Second things third: the abridged version of the back story. Sabermetricians, led by McCracken in '99, realized that pitchers can control Ks, BBs, HBP and HRs, and not much else. Fielding, ballpark and luck are much more relevant than pitcher aptitude to whether a grounder punches through for a single or gets gobbled up by the shortstop for a 6-4-3. Their BABIP-against, that is, the batting average on all those other balls put into play, tends to be random. Sort of. Hurlers with high BABIP one year often have low ones the next and vice versa. Not all and not always, but more or less. Void where prohibited by law.

Except: flyball pitchers give up fewer safeties than worm killers. On the other hand, those fly ball hits more often fall in as doubles and triples and those four hoppers through the infield rarely push the hitter past first.

So when the SABR dudes attempt to project pitching ERAs, they don't look at a pitcher's previous ERA but at the constituent parts -- HR, K, BB, HBP -- and assume an average BABIP going forward. These projections have bested projections using just previous ERA by a significant and consistent amount over the last decade, particularly as further refinements -- like ground ball and fly ball proclivities -- have helped salt the projections.

In English, if you want a prediction of Johnny Cueto's ERA in 2013 because you're considering making a bid on him in your fantasy league, his 2012 ERA of 2.78 says GO! But his constituent parts whisper FLUKE! They project a 3.74 ERA in 2013, assuming a league-average BABIP.

And now for something completely different: Suppose Dave Cameron tells you that he has a formula that's NEW AND IMPROVED! It can account 15% better for the batting average against a pitcher on balls put in play. That means he can get you a better idea of the pitcher's ERA and WHIP, and give you a leg up on the innumerate competition. Has Dave Cameron ever lied to you before?

In a nutshell, what Cameron found is that pitchers who induce a lot of pop-ups and who get batters to swing and miss on more pitches in the strike zone, reduce their BABIPs. It's partly why consistently good pitchers, like Justin Verlander and Jared Weaver, post consistently low BABIPs. In other words, we've found another way that BABIP is not entirely random for pitchers. (It's not random at all for batters, as anyone who's watched Ichiro knows intuitively.)

If Cameron's conclusion is supported by more research -- because this is how science advances -- pitching projections will get 15% better. How big a deal is that? Five advances of 15% each double the accuracy of projections. (Trust me on the math; I stayed in a Holiday Inn near MIT.) But even without four more Eureka moments among the seamheads, you can look up a pitcher's pop-up rate last year and squeeze out an edge on the competition in your fantasy league.

The naysayers will still neigh because advances of these stripe -- small and complicated -- are hard to see. The projections still won't identify the lucky dogs, the breakthroughs and the cliff divers. They won't be able to predict which pitcher tamed his control tiger, altered his mechanics and added 4 mph to his heater, or just figured something out. Truth is, luck is still way too big a factor in baseball to overcome with science. But if science can help you win your fantasy league you probably want to listen to the scientists.

BABIP Regressors: Pitchers to Watch

If you're in the market to sign a starting pitcher, it would be helpful to know what performance to expect from the various options. Unfortunately, while there are plenty of rear view mirrors, only Mr. Peabody & Sherman can tell you what lies ahead, and their time machine stopped working in 1971.

Nonetheless, we gaze into the past and salt our observations with what it all portends. Which brings us to pitcher BABIP.

Research over the years since Voros McCracken elucidated the theory that pitchers don't control the results once a ball is put in play has uncovered a grain of truth in his finding. While some pitchers can maintain a lower BABIP over their entire career, it's still true that large variations from their own established levels can often be attributed to a smiling deity. Because deities are fickle, this element of a player's performance tends not to repeat itself over time.

All of which is a long way of saying that some pitchers appeared to be particularly lucky or unlucky in 2012 and may be expected to deliver more or less of the goods in 2013. 

Let's take a look:

League average BABIP is about .298. That is, batters hit just under .300 when they aren't striking out, homering, walking or getting dinged by pitches. In 2011, Braves lefty Mike Minor suffered through a season of .350 BABIP, so while he pitched solidly, his ERA of 4.14 belied it.

In 2012, the league adjusted to Minor and his performance wasn't nearly as strong. But his ERA held steady (4.12) thanks to a .256 BABIP. Nothing in his underlying performance, or in the defense behind him, suggested such a change. That means Atlanta should hold the reins on expectations for the 25-year-old Vanderbilt man.

If Minor inspires warning flags, the regression sirens have to be blaring around Kyle Lohse. Over the last two years, Lohse is 30-11, 3.10, following a 10-year career of 88-98, 4.50. It's likely that Lohse has figured something out, but BABIPs of .269 and .262, compared to a lifetime BABIP right around league average, hasn't hurt. The music eventually stops, and 2013 is as good a time as any. Add to that his age-34 season and Lohse is in for major regression.

Jeremy Hellickson enjoyed a .262 BABIP last season, but is he in for regression? Hellickson's three-year career in St. Petersburg has been a study in confounding the numbers. Over 403 innings, Hellickson has a career BABIP of .244, which suggests that he might be doing something special (or playing behind a special defense) that's tamping down opponent batting average. Either that or the dude has been abidingly lucky and is going to crash in 2013. Hellickson has a lifetime ERA 0f 3.06, but the statheads suggest he's more like a .450 ERA pitcher. We'll see.

Then there's Gio Gonzalez, who entered MLB in 2008 with Oakland. Here is the arc of Gio's career, expressed in ERA: 
2008 - 7.68
2009 - 5.65
2010 - 3.23
2011 - 3.12
2012 - 2.89

So the fact that batted balls fell safely at just a .267 clip last year doesn't dull his prospects much. Gonzalez has a lifetime BABIP of just .286; perhaps in part because his two home parks, Oakland and Washington, are BABIP dampers; so if he regresses even to his 2011 performance, he's still a star in 2013.

Max Scherzer had a terrific 2012, going 16-7, 3.74 with 231 strikeouts. That last figure helped dull the impact of a .333 BABIP. That's fortunate, because it's likely that a chunk of Scherzer's BABIP lies at the feet of immobile infielders who will be back with the Tigers, barring a major surprise, in 2013.

It's just the opposite for Nathan Eovaldi, the young Marlin righty who relies on his defense more than average. His .316 BABIP in L.A. and Miami was particularly harmful to his stat line, a painful 4-13, 4.30 that overstates his struggles. Given his youth, Eovaldi is headed for what will look like a breakout in 2013.

One last interesting case -- in nearly every respect -- is R.A. Dickey. Kuncklers have a history of outfoxing the whole BABIP system, but Dickey has maintained nearly identical BABIPs around .275 since returning with the scroogie in 2010. It's quite possible that Dickey as not been particularly lucky, just particularly adept, and that his run of extraordinary performances could continue.

Diving Deeper In the BABIP Pool

A pitcher throws a pitch and it's strike three. Bully for him. He shares the credit with no one, except perhaps the batter, who deserves the entirety of the blame. It's part of what makes baseball largely a team game of individual performances.

Or, the batter takes ball four. The blame-credit dispensation is switched. Demerit pitcher and no one else. Credit batter alone.

Same hurler and hitter, but this time the bat makes contact and the ball sails into the warm, starry night. The moundsman can't blame his defense for a home run, unless it bounces off Jose Canseco's noggin and over the fence.

Walks, HBPs, Ks and home runs are all simply mano-a-mano events. When we examine the records of the combatants, we know exactly who did what by the result. Other kinds of results leave much more room for error because defense and luck play significant roles. 

Surely a pitcher who induces lots of pop-ups has the right to expect to pile up the outs. Those that drop between fielders are the exception, rather than the rule. Conversely, serving up a barrage of line drives is likely to lead to a quick hook.

Which brings us to Vernon Wells. The man with the toxic contract hit a miserable .218/.248/.412 for his 26 million,187 thousand, five hundred dollars in 2011. Wells isn't worth that money in three good years, but with that line he wasn't even worth a roster spot. You might espy a Major League Baseball-worst BABIP of .214 and conclude that he's ripe for a 2012 rebound. Alas, a deeper dive into his record and the kinds of balls he hit suggests that he suffered more from too many pop-ups and dribblers than from bad karma for his inflated paycheck.

With 86 punchouts in 529 plate appearances, we could expect Wells to bat .234 with his career BABIP of .282. But Wells was making weak contact in 2011, resulting in fewer line drives, more infield rollers, fewer ground balls, more shallow pops and more flies in general than his career average. In fact, he lined just one-eighth of his fair batted balls, the lowest percentage of his career. 

All this suggests Wells had no business batting much above .218, and unless he discovers a performance-enhancing formula, will have to stave off the effect of turning 33 since the last season. Mark Teixeira, now he's another story.

The slugging Yankee first-bagger suffered the third worst BABIP in baseball last season at .239. Teixeira hurt himself by pulling the ball too much, but hit just .215 on ground balls and just .646 on line drives. League average for each is about .320 and .750 respectively. Expect him to improve on his .248/.341/.494 in 2012, scary as that is.

Here's the list, courtesy of Baseball Reference, of the hitters who qualified for the batting title with the lowest BABIP in 2011. Expect most of them to improve at least their batting average and on base percentage next season.

Rk	Player	BAbip	Tm	PA	AB	H	2B	3B	HR	BB	SO	BA	OBP	SLG	OPS	Pos
1	Vernon Wells	.214	LAA	529	505	110	15	4	25	20	86	.218	.248	.412	.660	*789/D
2	Alexis Rios	.237	CHW	570	537	122	22	2	13	27	68	.227	.265	.348	.613	*8/D
3	Mark Teixeira	.239	NYY	684	589	146	26	1	39	76	110	.248	.341	.494	.835	*3/D
4	Evan Longoria	.239	TBR	574	483	118	26	1	31	80	93	.244	.355	.495	.850	*5/D
5	Ian Kinsler	.243	TEX	723	620	158	34	4	32	89	71	.255	.355	.477	.832	*4D
6	Kurt Suzuki	.244	OAK	515	460	109	26	0	14	38	64	.237	.301	.385	.686	*2/D
7	Casey McGehee	.249	MIL	600	546	122	24	2	13	45	104	.223	.280	.346	.626	*5/3D
8	Dan Uggla	.253	ATL	672	600	140	22	1	36	62	156	.233	.311	.453	.764	*4
9	Yuniesky Betancourt	.259	MIL	584	556	140	27	3	13	16	63	.252	.271	.381	.652	*6
10	Carlos Santana	.263	CLE	658	552	132	35	2	27	97	133	.239	.351	.457	.808	*23/D

Batters do have some ability to affect BABIP; Ichiro in his productive years regularly batted .350 on balls in play. Wells, on the other hand, hits just .282 lifetime even when he's not striking out. Here are the guys with highest BABIP in 2011.

Rk	Player	BAbip	Tm	PA	AB	H	2B	3B	HR	BB	SO	BA	OBP	SLG	OPS	Pos
1	Matt Kemp	.380	LAD	689	602	195	33	4	39	74	159	.324	.399	.586	.986	*8/D
2	Adrian Gonzalez	.380	BOS	715	630	213	45	3	27	74	119	.338	.410	.548	.957	*3/D9
3	Emilio Bonifacio	.372	FLA	641	565	167	26	7	5	59	129	.296	.360	.393	.753	65789/4
4	Michael Bourn	.369	TOT	722	656	193	34	10	2	53	140	.294	.349	.386	.734	*8
5	Michael Young	.367	TEX	689	631	213	41	6	11	47	78	.338	.380	.474	.854	D534/6
6	Alex Avila	.366	DET	551	464	137	33	4	19	73	131	.295	.389	.506	.895	*2/D5
7	Miguel Cabrera	.365	DET	688	572	197	48	0	30	108	89	.344	.448	.586	1.033	*3/D
8	Hunter Pence	.361	TOT	668	606	190	38	5	22	56	124	.314	.370	.502	.871	*9
9	Alex Gordon	.358	KCR	688	611	185	45	4	23	67	139	.303	.376	.502	.879	*7/3
10	Dexter Fowler	.354	COL	563	481	128	35	15	5	68	130	.266	.363	.432	.796	*8

How about Dexter Fowler, whose .354 BABIP still left him hitting .266. When his luck runs out, so will his MLBPA union card. On the other hand, we've seen this before from Miguel Cabrera and Adrian Gonzalez. Cabrera's just 18 points above his lifetime mark of .347, which is why a) the bronze sculptor in Cooperstown is practicing carving Cabrera's face and b) no one is suggesting a fall to earth in 2012 for him.

We'll look at pitcher BABIP-against next time, and see who may be fixing to soar or belly-flop next season.

b