Proof of a conjecture about Parrondo's paradox for two-armed slot machines

Huaijin Liang,Zengjing Chen

2024-06-23

Abstract:The 1936 Mills Futurity slot machine had the feature that, if a player loses 10 times in a row, the 10 lost coins are returned. Ethier and Lee (2010) studied a generalized version of this machine, with 10 replaced by deterministic parameter J. They established the Parrondo effect for a hypothetical two-armed machine with the Futurity award. Specifically, arm A and arm B, played individually, are asymptotically fair, but when alternated ran-domly (the so-called random mixture strategy), the casino makes money in the long run. They also considered the nonrandom periodic pattern strategy for patterns with r As and s Bs (e.g., ABABB if r = 2 and s = 3). They established the Parrondo effect if r + s divides J, and conjectured it in four other situations, including the case J = 2 with r >= 1 and s >= 1. We prove the conjecture in the latter case.

Probability,Computer Science and Game Theory

What problem does this paper attempt to address?

The paper mainly discusses the application of the Parondo paradox in the two-armed bandit problem, particularly the proof of a conjecture by Ethier and Lee. The conjecture involves a modified Mills future-prize bandit model, where the loss of coins after losing ten times in a row is returned, and the number ten is replaced by a parameter J (J is greater than or equal to 2).

In the single-arm model, the bandit machine has an internal cam that generates I different non-future-prize patterns, which appear cyclically in sequence. In the two-armed model, each arm independently has a non-future-prize payout probability, but players can switch between the two arms randomly or according to a non-random periodic pattern. Ethier and Lee have previously proven that by using a mixed strategy, even if both arms are individually fair, the casino can still profit in the long run, which is known as the Parondo effect.

Ethier and Lee proposed a conjecture that even when J=2 and r and s (representing the number of times A and B are played in the strategy) are greater than or equal to 1, the Parondo effect still exists even if the condition of r+s being divisible by J is not satisfied. The main contribution of the paper is to prove this conjecture, showing that the casino can achieve profits in the long run regardless of which non-random periodic pattern strategy the player adopts.

The authors demonstrate how the casino achieves profits through this non-random periodic pattern strategy by expanding and analyzing the expected profits of the casino under different strategies. They also propose a new method to accurately calculate the long-term profits of the casino when adopting a specific non-random periodic pattern strategy, and prove that these profits are always positive, thereby confirming Ethier and Lee's conjecture. The proof details of the paper are included in the appendix.