Be the house. That is the central theme of Moneyball and the mantra of those engaged in empirically-based approaches to baseball management. Take as many small advantages as possible so that, over time, you win and the poor suckers who risked their hard earned dollars on the roulette wheel, the blackjack table, and Carl Pavano have no idea how it happened.
If you had not noticed, there is a very good Indian Casino operating in Cleveland right now that is putting on a veritable “be the house” management clinic. Need a relief ace and a future star third baseman? Just make sure the customer leaves the craps table with a waitress named Coco and a Crisp stack of bills and he will be hailing a taxi at the Logan Airport arrival gates before he realizes that he gave up too much. Now you need to replace Coco? Word is that a there is an underused outfielder (with a career .291/.380/.442 line!) working in Philly and the management there has a weakness for geriatric loogys, so you send one down the Rhode(s) in exchange for the soon-to-be emancipated outfielder. Before you know it, you have lost an outfielder with a 102 career OPS+ and gained an outfielder with a 113 career OPS+, you have lost a relief pitcher with a 105 career ERA+ and gained a relief pitcher with a 116 career ERA+. Oh, and you gained Baseball America’s second highest ranked third base prospect (and fourteenth overall) to boot.
They - the 'marks' - just keep coming back for more and the Cleveland Indian Casino keeps turning a tidy profit while maintaining the illusion that the house can be beaten. It is an unassuming operation, specializing in compounding tiny advantages into large aggregate edges.
[Author's note: I am well aware that there is a joke available here wherein the setup involves a general manager walking into the casino and assuming that he can gain an advantage by picking up a cue or tell from the stoic and drab employee working the poker table. The middle of the joke is cloudy, but the punch line is 'nobody reads the Cleveland plain dealer'.]
The sportsbook, of course, is the casino anomaly. Whereas the house - provided balanced action and adequate vigorish - never loses, the astute and disciplined player can, conceivably, win over a statistically significant amount of time. Any time that a player can identify the situations in which a market inefficiency outweighs the vigorish, there is an opportunity for profit. The fact, however, is that the market is complex and the trappings so vast that entering into it for anything more than 'pizza money' is an absolutely foolhardy enterprise for virtually everybody. Ironically, the discipline required to bet on sports successfully is displayed only by the most risk-averse of people, a section of society that is least likely to be attracted to the sportsbook in the first place. All of which is to say that virtually all players in the sportsbook lose.
With that caveat in mind, and in keeping with our casino theme, I thought that it would be interesting to look into some of the available futures and props markets as they pertain to Cleveland. Armed with the knowledge that the betting market is as efficient a prognosticator as any, it is interesting to compare those lines with some of the more popular projection systems.
The over/under on wins for Cleveland this season currently sits at 89.5, which seems high relative to the amount of wins predicted for them by PECOTA, Diamond Mind and ZiPS (83, 87 and 86, respectively) in the simulations conducted by SG, using the three systems, at the Replacement Level Yankees Weblog. Cleveland's 2005 was a great improvement on the previous season, and this feels like a perfect situation for the application of Bill James' 'plexiglass principle', which stipulates that a team will regress to the league mean somewhat in the season following a great improvement. I would tend to agree with the projection system consensus that 89.5 is a little high. I say this having been an ardent Cleveland-backer at this time last year, when my preview asked the absolutely prescient question: "... is this the best bullpen in baseball?" It was, you know. Of course, I take full responsibility for not having foreseen Kenny Williams' success at alchemy.
The line for Cleveland to win the American League Central is +198. Both ZiPS and PECOTA give Cleveland (roughly) a 30% probability of winning the division, based on the aforementioned simulations. At less than a one in three chance for less than a two to one return, the proposition falls short of profitability (a 33% chance of winning at +200 would be offer a break-even expectation). They are +942 to win the pennant and +1874 to win the World Series, which correspond roughly as expected to their odds of winning the division.
Some of the more interesting futures markets involve two teams' win totals. In the table below, you will find the line, the team being matched up with Cleveland and the pick as provided by each of PECOTA, Diamond Mind and ZiPS.
Line Match Up PECOTA Says D.M. Says ZiPS Says -2 Braves Push Indians -2 Indians -2 -15.5 Reds Reds +15.5 Reds +15.5 Indians -15.5 -6 Astros Astros +6 Indians -6 Indians -6 -4.5 Dodgers Dodgers +4.5 Dodgers +4.5 Dodgers +4.5 -8 Brewers Brewers +8 Brewers +8 Brewers +8 -7.5 Phillies Phillies +7.5 Phillies +7.5 Phillies +7.5 -6 Giants Giants +6 Giants +6 Giants +6 +3.5 Cardinals Cards -3.5 Cards -3.5 Cards -3.5 -14.5 Nationals Nats +14.5 Nats +14.5 Nats +14.5
It seems relatively clear that the sportsbooks - or at least the investors in the market - are overrating Cleveland relative to the simulation systems. I will take Philadelphia +7.5 for the limit, please.
One of my favourite bets available is the "first American League manager to be fired or resign" proposition. Cleveland skipper Eric Wedge comes in at +1251, a significantly longer shot than favourites Buck Showalter (+511), Ron Gardenhire (+615) and Mike Hargrove (+666). With regard to Wedge, I would think that the manager of any team that is likely to fall prey to the 'plexiglass principle' could be on shakier ground than most; I call it the Tosca Factor.
For the fantasy player, the most enthralling of proposition bets are those that pertain to individual players, and there are no shortage of markets available for betting that involve Cleveland players. In the table below, you will find the prop, the line, and the picks as provided by PECOTA and ZiPS.
Prop Line PECOTA ZiPS Hits T. Hafner -1 v. J. Bay Bay +1 Bay +1 Hits G. Sizemore +5.5 v. M. Tejada Tejada -5.5 Tejada -5.5 HR T. Hafner -2.5 v. R. Howard Howard +2.5 Howard +2.5 RBI T. Hafner -6 v. A. Dunn Dunn +6 Dunn +6 Runs G. Sizemore +1 v. J. Rollins Rollins -1 Rollins -1 Runs T. Hafner +5 v. A. Dunn Dunn +5 Dunn +5 SB G. Sizemore -1 v. C. Beltran Beltran +1 Beltran +1 Wins C. Lee +1 v. C. Zambrano Zambrano -1 Push
It cannot be coincidental that both projection models agree in each of the eight propositions that the non-Cleveland player is the wise pick (with the exception that ZiPS has Zambrano pegged for exactly one more win than Lee). As we have seen with both the player props and the team win futures, it appears as though the market is overplaying Cleveland to a certain extent. This fits in perfectly with the idea that scientific projection systems, like PECOTA and ZiPS, are much more adept at regressing expectation toward the mean and eschewing trendy picks than are public prognosticators and bettors who are not so impervious to subjective indicia of future performance.
Finally, there is one line that I saw at an offshore sportsbook that made me giggle and I would like to share it with you all.
Team to End World Series Drought First Team Line Cubs +325 Indians +180 Giants +250 Rangers +375 Astros +350
With all due respect to the long-suffering fans of these teams, let me say this: do not, under any circumstances, wire your hard-earned money to Costa Rica to bet into a market that offends the 'rule against perpetuities'!
[Authors note: I just realized that most readers are fortunate enough to have never had to sit through a semester of Wills and Estates class, so here is a definition of the 'rule against perpetuities', courtesy of Black's Law Dictionary: "[the] principle that no interest in property is good unless it must vest, if at all, not later than 21 years, plus period of gestation, after some life or lives in being at time of creation of interest." This is not to be confused with the 'fertile octogenarian' rule, which refers to the remote possibility that any woman who was alive during Julio Franco's rookie season is still pre-menopausal.]
Finally, in order to beat this casino theme into submission, it is time for us to discuss the single greatest thing about casinos: free drinks, free rooms, free parking... Comps!
Player comparables, using baseball-reference's fantastic similarity scores, are a relatively blunt instrument for putting a player's past in historical context in order to gain insight into his future. For each player, I have considered each of his three most comparable players up to that age and discarded the median, leaving us with a good scenario and a bad scenario. Good comp/bad comp, if you will. Part of the reason for my doing this is to demonstrate both the excitement of having a core of good, young players and the inherent folly of expecting them all to progress, in unison, and improve upon previous seasons. In what has - organically - become somewhat of a secondary theme for this essay, we can experience the highs of the ascent to superstardom and the accompanying lows of regression to the mean that coexist in the range of probabilities for young players who have recently had a breakthrough season. Consider the Greatest Jay, Carlos Delgado. After his first full season at age 24 in 1996, his list of most comparable players through that age included both Hall of Famer Willie Stargell and perpetual fourth outfielder Brian Hunter. Remember, Cleveland fans, that they will not all be Stargells.
In the following table, you will see the principle players on Cleveland, along with each of their 'good scenario' comparable players through their current age as determined by the above-described method, and how each fared in the following season (which represents an expectation for the Cleveland player for 2006). A final note: the similarity scores are calculated based on raw statistics, rather than adjusted ones such as OPS+. For that reason, you need not consider the era or park for the comparable player listed when considering the application of his statistics line to the current Cleveland player - the raw numbers will suffice. For a full explanation of similarity scores, check out baseball-reference.
Good Comp Player Comparable Year Line V. Martinez Ernie Lombardi 1935 .343/.379/.539 B. Broussard Paul Sorrento 1995 .235/.336/.511 R. Belliard Todd Walker 2004 .274/.352/.468 A. Boone Randy Jackson 1959 .235/.326/.358 J. Peralta Ernie Banks 1955 .295/.345/.596 J. Michaels Pat Mullin 1948 .288/.385/.504 G. Sizemore Duke Snider 1950 .321/.379/.553 C. Blake Mike Blowers 1997 .293/.376/.427 T. Hafner Brian Giles 2000 .315/.432/.594 C.C. Sabathia Tony Cloninger 1966 1.53 K/BB, 4.12 ERA J. Westbrook Dave Guisti 1968 2.78 K/BB, 3.19 ERA C. Lee Bartolo Colon 2000 2.16 K/BB, 3.88 ERA P. Byrd Woody Williams 2002 3.04 K/BB, 2.53 ERA J. Johnson Glenn Abbott 1983 2.23 K/BB, 3.63 ERA B. Wickman Todd Jones 2005 4.43 K/BB, 2.10 ERA G. Mota Steve Farr 1989 2.55 K/BB, 4.12 ERA R. Betancourt Tom Ferrick 1946 2.44 K/BB, 3.58 ERA
That, my friends, is a 110 win team.
Bad Comp Player Comparable Year Line V. Martinez Terry Kennedy 1983 .284/.342/.434 B. Broussard Henry Rodriguez 1997 .244/.306/.479 R. Belliard Damion Easley 2001 .250/.323/.376 A. Boone Chris Sabo 1995 .238/.283/.345 J. Peralta Eddie Miller 1941 .239/.288/.326 J. Michaels Ernie Koy 1940 .301/.357/.452 G. Sizemore Carlos Beltran 2000 .247/.309/.366 C. Blake Jerry Martin 1981 .241/.308/.336 T. Hafner Dick Stuart 1962 .228/.286/.398 C.C. Sabathia Chuck Stobbs 1955 1.05 K/BB, 5.00 ERA J. Westbrook Phil Regan 1965 1.85 K/BB, 5.05 ERA C. Lee Darren Oliver 1998 1.32 K/BB, 5.73 ERA P. Byrd Mike Bielecki 1995 1.45 K/BB, 5.97 ERA J. Johnson Pat Rapp 2000 1.28 K/BB, 5.90 ERA B. Wickman Greg Minton 1989 1.14 K/BB, 2.20 ERA G. Mota Mike Maddux 1994 2.46 K/BB, 5.11 ERA R. Betancourt Ace Adams 1943 0.84 K/BB, 2.82 ERA
That, my friends, is at best a 95 loss team.
The point to be illustrated here is that neither the all good comp scenario nor the all bad comp scenario is likely... in fact either of those contingencies borders on the impossible, probabilistically considered. Some of these players will, in fact, achieve the lofty standards set by their respective good case comparable players. Likewise, some will regress toward the mean or worse (drift away from the mean on the negative side) as did many of the bad case comparable players. Cleveland is 'the house' in terms of management savvy precisely because of how well it manages that risk by judiciously apportioning resources among a wide array of players that could become stars, but assuredly resigned to the fact that some will not. It is a classic employment of risk-aversion, that very trait that the casino has and the customer lacks.
In conclusion, I think that the glimpse into comparable players and the juxtaposition of public betting markets with the prognostications of some empirically based projection systems lead us to one conclusion: while this is a very good, young team, the public expectations are likely too high because it fails to consider the possibility of collapse and the reality of regression. What was that 'plexiglass principle' again?