Applying
Game Theory Models to Games (Athletic Contests)
By Robert E.
Wright, Nef Family Chair of Political Economy, Augustana University for the 30th
Annual Teaching Economics Conference, McGraw-Hill Higher Education/Robert
Morris University, Moon Township, Pennsylvania, 22-23 February 2019.
“In theory, there is no difference between
theory and practice. In practice there is.” Or so baseball legend Yogi Berra
once reputedly quipped. He may have said that -- it sounds like Yogi -- but he
wasn’t the first to utter those words, an upperclassman at Yale in 1882 was (https://quoteinvestigator.com/2018/04/14/theory/). Similarly, I am not the
first to apply game theory to sport (see, e.g., Mottley 1954) but I hope to add
to the discussion by suggesting that game theory’s application to coaching
territorial team sports like basketball, football, hockey, lacrosse, rugby,
soccer, and water polo, can be powerful, but it is not a panacea and even can
be counterproductive. Game theory is often more powerfully applied to various
non-territorial sports, including baseball (see, e.g., Weinstein-Gould 2009;
Turocy 2014) and volley sports like tennis and volleyball (Lin 2014).
I would not call strategic problems in the
actual playing of territorial team sports “wicked,” in the technical sense of
“wicked social problems” that have no stopping rule, have solutions that are
only better or worse rather than right and wrong, and so forth (Peters 2017).
But they are certainly “complex” problems a la Nason (2017) and even “chaotic”
a la Fergus Connolly (Connolly and White 2017), who incidentally consults for
the Robert Morris University Colonials, among other elite level teams. His
hyper-interdisciplinary systems-within-systems approach to territorial sports
forms the core of the masters course that I teach at Augustana University on
the Business of Coaching.
But Connolly’s approach does not make absolutely
clear to readers that successful coaches must engage in strategic
competition, where strategy refers to the anticipation of the moves of their
opponents and not some vague notion of planning, as in the term “game plan.”
Game theory, which of course is intrinsically interesting for many students and
a key tool in the honing of strategic sensibilities (Dixit 2005), barely
registers in Connolly’s otherwise seminal/ovanal/gaminal 2017 opus, Game
Changers.
Two player/team/coach games, especially zero-sum
ones, seem like a natural way to apply game theory to sport. The realities of
on-field competition, however, quickly reveal the shortcomings of simultaneous
one-shot games, much as happened when Chicken was applied to the Cuban Missile
Crisis (Zagare 2014) and the Prisoner’s Dilemma (PD) was applied to the actual
behaviors of criminals (Khadjavi and Lange 2013). In both instances, it quickly
became apparent that PD and other simple games exist within larger game
structures and cannot in themselves always satiate real world decision makers.
For example, HBO’s The Wire showed that the simple PD description we all
work through in class (a la Tucker 1983) is, in the reality of Baltimore’s drug
scene, embedded in another game, actually referred to on the street as The
Game, where “snitches get stitches,” or worse, creating a payoff structure
where players keep their mouths shut no matter what (Cherrier 2012).
Simple games also break down due to the speed of competitive
play. Time to think through payoff structures does not exist; reactions must be
instinctive to be fast enough to matter. The best that can be done is for
coaches and players to think through and model various game scenarios and then
drill the rational responses, much as ex-NHL player Nicklas Lidstrom has done
regarding one-on-one plays in hockey, and so forth (Lennartsson, Lidstrom, and
Lindberg 2015).
Many games applicable to sport have mixed
strategy equilibria and hence dissolve into Minimax with random strategy
solutions (see, e.g., Flanagan 1998), a fact that the offensive coordinator of
the Los Angeles Rams seems not to have fully grasped during the recent Super
Bowl LIII. Several studies have shown that minimax predictions do not hold up
well in the laboratory (Levitt, List, and Reiley 2010) but do on the field, at
least where strategy randomization and outcomes can be precisely measured, as
in tennis serves and soccer penalty kicks (Palacios-Huerta 2003). Similarly,
McGarrity and Linnen (2010) leverage a natural experiment, the injury of a
starting quarterback, to show that NFL football teams play the equivalent of a
matching pennies game wherein the defense tries to match the offense’s decision
to run or pass and the offense tries to not match the defense’s decision. I
suggest that football teams actually run three types of plays -- run, pass, and
hybrids like draws, options, play action, and screens -- so a
rock-paper-scissors type game might be even more realistic (Spaniel 2011).
In any event, working through mixed strategy
examples can be helpful for aspiring coaches to see that randomness can be
optimal under specific conditions. That does not alleviate their angst
concerning their replacement by, if not just computers, then nerds using
computers (Davenport 2016; Jones 2018), but it can help them to overcome
behavioral biases like the risk aversion that apparently induces baseball
pitchers to throw too many fastballs (Kovash and Levitt 2009), football coaches
to punt too frequently on fourth down and to run the ball too much, and
basketball players not to attempt as many three-point shots as they should
(Fichman and O’Brien 2018).
The best applications of game theory to
territorial sports often occur off the field but can still be of immense importance
to coaches. Aspects of sport design, a huge arena ably surveyed, albeit over 15
years ago now, by Szymanski
(2003), are amenable to game theoretic modeling. How to keep up fan interest is
a core concern as it forms the basis for all sports funding, except in amateur
pay-to-play leagues, in which case maximization of player and/or parent utility
is paramount. Most research suggests that fans want the home team to win, but
in a close contest, rendering mechanisms for ensuring something like on-field parity
of prime interest. That leads in many fruitful directions, like rules for demoting
teams from the top tier, as in European soccer leagues, and drafting new
players. Forty years ago, for example, Brams and Straffin (1979) showed, with a
simple PD game and four fairly realistic assumptions about complete information
and incomplete collusion, that North American-style drafts could lead to Pareto
inefficient outcomes. As Syzmanski (2010) has shown, however, many early
treatments of competitive balance made unrealistic assumptions about the
correspondence between individual athletic talent and team wins. In sum, the
sum of individual talent can be greater than, equal to, or less than actual
team performance.
The need to maintain competitive balance also raises the largely intractable issue of doping, or the use of
performance-enhancing substances (Kirstein 2009). Frank Daumann of the
Institute for Sports Science in Jena, Germany, recently (2018) offered a game
theoretic analysis of the doping strategies of two athletes that can be easily
modified to a scenario where two head coaches must decide whether or not to
allow their players to use performance enhancing substances not explicitly
banned by the league or conference in which they compete. Benefits of doping
include a higher probability of winning and hence of job retention (Fizel and
D’Itri 1997), advancement, bonuses, and a burnished reputation. Costs include
the price of the substances themselves and reductions in athlete health, both
presumably small in present value terms, and the risk of a tarnished reputation
(Butler 2014). This, Daumann shows, could be modeled as a one-off PD such that
both coaches will decide to allow their players to dope although both teams
would be better off if they did not use performance enhancing substances.
But of course in team sports, especially the
territorial ones considered here, athletes may try to free ride on their
teammates. In other words, simply because a coach signals that athletes may
dope does not mean that they will choose to do so as players may hope that
enough of their fellows will bear the costs of doping to improve team
performance without having to bear the costs themselves.
The prospect of free-riding raises the specter
of coaches forcing their players to dope, or leveraging asymmetric information
to trick them into doping (Johnson 2003), either of which would radically
change the payoff structure the coach faces as he or she may have to bear all
of the cost of the enhancement substances and any negative social and reputational
effects if league officials, competitors, or fans discover the doping, which
seems more likely if the coach forces it than if he or she simply allows it
(Dunbar 2014). Coercion seems unlikely, moreover, because team sport coaches
regularly face free rider problems in a variety of areas and have techniques
for mitigating them analogous to the techniques used by those drug dealers in
Baltimore that I mentioned earlier.
Like the leaders of drug dealers and criminal gangs
(Spergel 1990), organized crime “families” (Shvarts 2002), and most military
units (Rose 1945-46; Montgomery 1946), coaches reduce free riding by creating a
culture of trust, an ideology of service to others, and pseudo-familial bonds
through various rituals and shared adversity. In effect, they try to reduce the
economic rationality of their athletes by convincing them that they love their
teammates more than they love themselves (Connolly and White 2017), thus
inducing them to place a large weight on their colleagues’ well-being in their
own utility functions (Bergstrom 1997). Game theory aficionados will recognize
the similarities of this with the Battle of the Sexes, the iconic version of
which features a husband who wants to go to a football game and a wife who
wants to go to the opera but both prefer that outcome only if they can persuade
the other spouse to attend with them (Hahn 2003).
Of course, cultural manipulations sometimes fall
short or break down under stress, so coaches have other tools for reducing free
riding. One I call the wildebeest solution after a technique that wildebeest
herds reputedly use to cross crocodile-infested waters during their great
migrations across the African savannah. They force the putatively oldest,
weakest, least fertile member of the herd into the waters first. As the
voracious crocs devour her, the younger, stronger, healthier members of the
herd safely cross (Sapolsky 2017). No strategic choice is involved as the
wildebeests force one of their number into the river first, but costly signaling
may be involved. The key to survival is to appear not to be the oldest, weakest
member of the herd or, in our case, team, where death-by-crocodile is
substituted with getting cut from the team. Coaches, in other words, can reduce
defection, free riding, and shirking by making it clear that players who do not
signal that they are “team players,” who do not stand ready to forgo the
temptation to free ride, may end up making the ultimate sacrifice (Spence
1973).
Despite some recognition that Braess’s paradox
may apply to territorial team sports like basketball (Skinner 2010), superstar
athletes know that coaches cannot costlessly bench them, much less cut them
from the team, so some may shirk or defect by going for individual statistics
instead of wins (Berri and Krautmann 2006; Krautmann and Donley 2009), which is
why it is wise to make team performance, short of championship bonuses, a major
component of elite athlete compensation (Frick 2003). For most players, though,
fear of being labelled the wildebeest is enough to induce them to take one for
the team, even if that means injecting or imbibing some new or unusual
substance not yet banned.
In sum, coaches and other sports administrators
can leverage the insights of game theory to improve competitive outcomes but
they need to employ theory carefully and switch toolkits when need be, or face
becoming the expendable wildebeest themselves.
References
Audas, Rick, John Goddard,
and W. Glenn Rowe. (2006) “Modelling Employment Durations of NHL Head Coaches:
Turnover and Post-succession Performance.” Managerial and Decision Economics
27, 4: 293-306.
Azar, Ofer H. and Michael
Bar-Eli. (2010) “Do Soccer Players Play the Mixed-Strategy Nash Equilibrium?” Applied
Economics 43, 25: 3591-601.
Bergstrom, Theodore C.
(1997) “A Survey of Theories of the Family.” In Mark R. Rosenzweig and Oded
Start, eds. Handbook of Population and Family Economics, Vol. 1A, New
York: Elsevier.
Berri, David J. and Anthony
C. Krautmann. (2006) “Shirking on the Court: Testing for the Incentive Effects
of Guaranteed Pay.” Economic Inquiry 44, 3: 536-46.
Brams, Steven J. and Philip
D. Straffin, Jr. (1979) “Prisoners’ Dilemma and Professional Sports Drafts.” The American Mathematical Monthly 86, 2:
80-88.
Butler, Nick. (2014) “Coaches Can
Influence Whether Athletes Decide to Dope, Claims Scottish Study.” Inside the Games: The Inside Track on World
Sport 6 March.
Cherrier, Beatrice. (2012)
“To Teach or Not to Teach Economics with The Wire?” Institute for New
Economic Thinking Blog 1 November.
Connolly, Fergus and Phil
White. (2017) Game Changer: The Art of Sports Science. New York: Victory
Belt Publishing.
Daumann, Frank. (2018)
“Doping in High-Performance Sport -- The Economic Perspective.” In The
Palgrave Handbook on the Economics of Manipulation in Sport, Markus Breuer
and David Forrest, eds. New York: Palgrave Macmillan, 71-90.
Dixit, Avinash. (2005)
“Restoring Fun to Game Theory.” Journal of Economic Education 36, 3:
205-19.
Dunbar, Graham. (2014) “Lance
Armstrong’s Coach Banned 10 Years for Dope Promotion.” Las Vegas Review Journal 22 April.
Fichman, Mark and John
O’Brien. (2018) “Three Point Shooting and Efficient Mixed Strategies: A
Portfolio Management Approach.” Journal of Sports Analytics 4, 2:
107-120.
Fizel, John L. and Michael
P. D’Itri. (1997) “Managerial Efficiency, Managerial Succession and
Organizational Performance.” Managerial and Decision Economics 18, 4:
295-308.
Flanagan, Thomas. (1998)
“Game Theory and Professional Baseball: Mixed-Strategy Models.” Journal of
Sport Behavior 21, 2: 121-38.
Frick, Bernd. (2003)
“Contest Theory and Sport.” Oxford Review of Economic Policy 19, 4:
512-29.
Hahn, Sunku. (2003) “The
Long Run Equilibrium in a Game of ‘Battle of the Sexes’.” Hitotsubashi
Journal of Economics 44, 1: 23-35.
Hsu, Shih-Hsun, Chen-Ying
Huang, and Cheng-Tao Tang. (2007) “Minimax Play at Wimbledon: Comment.” American
Economic Review 97, 1: 517-23.
Johnson, Michael. (2003) “Athletes Must
Take Share of the Blame.” The Telegraph
25 October.
Khadjavi, Menusch and
Andreas Lange. (2013) “Prisoners and Their Dilemma.” Journal of Economic
Behavior and Organization 92: 163-75.
Kirstein, Roland. (2009)
“Doping, the Inspection Game, and Bayesian Monitoring.” Working Paper. 8
October.
Kovash, Kenneth and Steven D. Levitt.
(2009) “Professionals Do Not Play Minimax: Evidence from Major League Baseball
and the National Football League.” NBER Working Paper 15347.
Krautmann, Anthony C. and
Thomas D. Donley. (2009) “Shirking in Major League Baseball Revisited.” Journal
of Sports Economics 10, 3: 292-304.
Lennartsson, Jan, Nicklas Lidstrom, and
Carl Lindberg. (2015) “Game Intelligence in Team Sports.” PLOS One 13 May.
Levitt, Steven D., John A.
List, David H. Reiley. (2010) “Stays in the Field: Exploring Whether
Professionals Play Minimax in Laboratory Experiments.” Econometrica 78,
4: 1413-34.
Lin, Kai. (2014) “Applying
Game Theory to Volleyball Strategy.” International Journal of Performance
Analysis in Sport 14, 3: 761-74.
McGarrity, Joseph P. and
Brian Linnen. (2010) “Pass or Run: An Empirical Test of the Matching Pennies
Game Using Data from the National Football League.” Southern Economic
Journal 76, 3: 791-810.
Montgomery, Bernard. (1946)
“Morale in Battle: Address Given to the Royal Society of Medicine.” British
Medical Journal 2, 4479: 702-4.
Mottley, Charles M. (1954)
“Letter to the Editor -- The Application of Operations-Research Methods to
Athletic Games.” Journal of the Operations Research Society of America
2, 3.
Nason, Rick. (2017) It’s
Not Complicated: The Art and Science of Complexity in Business. Toronto:
UTP-Rotman Publishing.
Palacios-Huerta, Ignacio.
(2003) “Professionals Play Minimax.” The Review of Economic Studies
70, 2: 395-415.
Peters, B. Guy. (2017)
“What Is So Wicked About Wicked Problems? A Conceptual Analysis and a Research
Program.” Policy and Society 36, 3: 385-96.
Rose, Arnold. (1945-46)
“Bases of American Military Morale in World War II.” The Public Opinion
Quarterly 9, 4: 411-17.
Sapolsky, Robert W. (2017) Behave:
The Biology of Humans at Our Best and Worst. New York: Penguin.
Shvarts, Alexander. (2002)
“Russian Mafia: The Explanatory Power of Rational Choice Theory.”
International Review of Modern Sociology 30, 1/2: 69-113.
Skinner, Brian. (2010) “The
Price of Anarchy in Basketball.” Journal of Quantitative Analysis in Sports
6, 1.
Spaniel, William. (2011) Game
Theory 101: The Complete Textbook. Createspace IPP.
Spence, Michael. (1973)
“Job Market Signaling.” Quarterly Journal of Economics 87, 3: 355-74.
Spergel, Irving A. (1990)
“Youth Gangs: Continuity and Change.” Crime and Justice 12: 171-275.
Szymanski, Stefan. (2003) “The Economic
Design of Sporting Contests.” Journal of
Economic Literature 41, 4: 1,137-87.
_______. (2010) “Teaching Competition
in Professional Sports Leagues.” Journal
of Economic Education 41, 2: 150-68.
Tucker, Albert W. (1983)
“The Mathematics of Tucker: A Sampler.” The Two-Year College Mathematics
Journal 14, 3: 228-32.
Turocy, Theodore L. (2014) “An
Inspection Game Model of the Stolen Base in Baseball: A Theory of Theft.”
Working Paper. 22 August.
Weinstein-Gould, Jesse.
(2009) “Keeping the Hitter Off Balance: Mixed Strategies in Baseball.” Journal
of Quantitative Analysis in Sports 5, 2.
Zagare, Frank C. (2014) “A
Game-Theoretic History of the Cuban Missile Crisis.” Economies 2: 20-44.