[Paper Review] Adaptive Competition, Market Efficiency, Phase Transitions and Spin-Glasses

Robert Savit, Radu Manuca+1 more|arXiv (Cornell University)|Dec 23, 1997

Complex Systems and Time Series AnalysisEconomics, Econometrics and Finance5 references20 citations

TL;DR

This paper introduces a minority game model of adaptive competition where agents choose between two groups, with the minority group receiving points. Using a fixed strategy pool size, the system exhibits a phase transition: small pools lead to an efficient market phase with low collective payoff and no predictive advantage, while large pools yield an inefficient market phase with higher total payoffs and exploitable information. The critical pool size for the transition scales simply with the number of agents, and the system displays spin-glass-like behavior, particularly in the glassy, efficient phase.

ABSTRACT

We analyze a simple model of adaptive competition which captures essential features of a variety of adaptive competitive systems in the social and biological sciences. Each of N agents, at each time step of a game, joins one of two groups. The agents in the minority group are awarded a point, while the agents in the majority group get nothing. Each agent has a fixed set of strategies drawn at the beginning of the game from a common pool, and chooses his current best-performing strategy to determine which group to join. For a fixed N, the system exhibits a phase change as a function of the size of the common strategy pool from which the agents initially draw their strategies. For small pool sizes, the system is in an efficient market phase. All information that can be used by the agents' strategies is traded away, no agent can accumulate more points than would an agent making random guesses, and thus the commons suffer,since relatively few points are awarded to the agents in total. For large initial strategy pool sizes, the system is in an inefficient market phase, in which there is predictive information available to the agents' strategies, and some agents can do better than random at accumulating points. In this phase, the total number of points awarded to the agents is greater than in a game in which all agents guess randomly, and so the commons do relatively well. At a critical size of the strategy pool marking the cross-over from the efficient market to the inefficient market phases, the commons do best. This critical size of the pool grows monotonically with N. The behavior of this system has some features reminiscent of a spin-glass.

Motivation & Objective

• To understand the collective dynamics of adaptive, competitive systems where agents use strategies to anticipate minority group choices.
• To investigate how the size of the initial strategy pool affects market efficiency and system performance.
• To explore the emergence of phase transitions in such systems, analogous to those in statistical physics.
• To examine the implications of these transitions for resource allocation, information efficiency, and collective welfare in social and biological systems.

Proposed method

• Agents choose between two groups at each time step, with the minority group receiving a point.
• Each agent maintains a fixed set of strategies drawn from a common pool, selecting the one with the best historical performance.
• Strategies are based on historical sequences of minority group outcomes (of length m), predicting the current minority group.
• The system is simulated over time, tracking the standard deviation σ of group choices as a measure of collective behavior.
• Phase transitions are identified by analyzing σ as a function of strategy pool size s and system size N.
• Statistical analysis compares the system's behavior to a random benchmark where agents choose groups independently with equal probability.

Experimental results

Research questions

• RQ1 How does the size of the initial strategy pool affect market efficiency and collective payoff in adaptive competitive systems?
• RQ2 What causes a phase transition between efficient and inefficient market phases in this model?
• RQ3 Why does the system perform best collectively at a critical pool size, and how does this critical size scale with the number of agents?
• RQ4 In what ways does the system's behavior resemble that of a spin-glass, particularly in the efficient market phase?
• RQ5 How does the predictability of minority group outcomes change across phases, and what does this imply for information efficiency?

Key findings

• For small strategy pool sizes, the system enters an efficient market phase where no agent can outperform random guessing, and collective payoff is low.
• For large strategy pool sizes, the system enters an inefficient market phase where some agents can achieve better than 50% success rates due to exploitable predictive information.
• The total number of points awarded is higher in the inefficient phase than in a random market, indicating better collective performance.
• The critical pool size marking the transition from efficient to inefficient market phases grows monotonically and simply with the number of agents N.
• The efficient market phase exhibits glassy dynamics reminiscent of spin-glasses, with high frustration and constrained strategy competition.
• The standard deviation σ of group choices scales as σ ∝ N for small m (e.g., m=3), indicating strong correlations, while σ ∝ N^1/2 for larger m (e.g., m=16), indicating more random-like behavior.