^{1}

^{2}

This article was submitted to Artificial Intelligence in Finance, a section of the journal Frontiers in Artificial Intelligence

This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

High-frequency trading (HFT) offers an excellent use case and a potential killer application of the commercially available, first generation quasi-quantum computers. To this end, we offer here a simple game-theoretic model of HFT as the famous two player game, Prisoner’s Dilemma. We explore the implementation of HFT as an instance of Prisoner’s Dilemma on the (quasi) quantum cloud using the Eisert, Wilkens, and Lewenstein quantum mediated communication protocol, and how this implementation can not only increase transaction speed but also improve the lot of the players in HFT. Using cooperative game-theoretic reasoning, we also note that in the near future when the internet is properly quantum, players will be able to achieve Pareto-optimality in HFT as an instance of reinforced machine learning.

Non-cooperative game theory is the art of strategic interaction between individuals competing for joint stakes over which they entertain differing preferences. Game-theoretic reasoning can formally be traced back the Ancient Chinese General Sun Tzu (circa 500 BCE) and the ancient Indian minister, Chanakya (circa 300 BCE).

Mathematical formalization of non-cooperative game theory in the 20th century goes back to the work of von Neumann and Nash. The publication the seminal work of von Neumann and Morgenstern titled Theory of Games and Economic Behavior (

With the recent advent of commercially viable quantum computation and communication technologies, the confluence of ideas from game theory and quantum information processing has gained strong interest. This interest has given birth to the subject known as quantum game theory (

Consider the non-cooperative game called Prisoner’s Dilemma, a 2-player non-cooperative game in which each of the two players (prisoners) who committed a crime together are given the opportunity to reduce their time served in prison by helping authorities implicate the other player for the crime. This game is presented in tabular form in

Prisoner’s dilemma.

The players have disparate preferences over the outcomes of the game, which are captured below using the symbol ≻ to denote the notion of “preferred over”:

It is assumed that the players are rational, that is, each player will play the game in way that is consistent with his preferences. The game is played by employing strategies to optimize the payoffs. The two strategies available to both players are to either cooperate with the authorities to implicate the other player (C), or to defect from offer to help the authorities (D). The question is: what is the outcome of the game (or the play of the game)?

The answer is provided in the form of Nash equilibrium, a profile of strategies, one per player, in which no player has motivation to deviate from his strategic choice. In other words, Nash equilibrium is a strategy profile in which each player’s strategy is a best reply (with respect to the players’ preferences) to all others. Not all games have a Nash equilibrium.

For Prisoners’ Dilemma,

Nash equilibrium versus Pareto-optimal outcomes in Prisoner’s Dilemma.

When Nash equilibrium is not present in a game, or if it is sub-optimal, game-theorists suggest that players employ randomization over the outcomes as a mechanism for introducing or improving Nash equilibrium. To this end, players are allowed to independently randomize over their respective strategies, a notion referred to as mixed strategies, to produce probability distributions over the outcomes. The resulting mixed game will have at least one Nash equilibrium outcome (John Nash’s Nobel prize winning result

Further refinement of the Nash equilibrium may be possible if a referee is inducted into the game at negligible cost. This proper extension of a game is know as the game with mediated communication. In such games, the referee creates a probability distribution over the outcomes of the game that the players could not using mixed strategies. The referee then tells each player in confidence which strategy he should employ. Each player than checks the viability of the referee’s advice with respect to his preferences and the 50–50 chance of the other player agreeing to the advice given to him by the referee. If the viability checks out, the player agrees with the referee. When both players agree to the referees advice, the resulting Nash equilibrium is known as a correlated equilibrium.

Even further refinements of Nash equilibrium are conceivable by simply extending the domain of the game from Euclidean space to more exotic (and non-trivial) mathematical spaces such as Hilbert space and Banach space. The challenge then becomes how to keep the mathematical extensions grounded in physical reality. For the case of games extended to complex projective Hilbert space, the physical context is quantum mechanics. The result of this extension is the theory of “quantum games.”

Foreseeing the rise of quantum technologies like quantum computers and quantum communication devices and protocols, Meyer offered the first game-theoretic model of quantum algorithms. In his seminal work (

The quantum circuit implementation of the EWL quantum game protocol. The referee consists of two quantum logic gates, ^{†}. In the middle of these two operations are the players’ independent quantum strategic choices that each of them enacts on her qubit as unitary operations. We assume the top qubit is Player I’s and the second one is Player II’s.

The EWL protocol is a quantum circuit that takes in as input the two-qubit state

With each qubit belonging to one player. This state is acted upon by the referee to produce a higher-order randomization in the form of a quantum superposition followed by measurement. In particular, the referee entangles the two qubits using a general entangling operator_{
x
} is the Pauli-spin flip operator, and

For

Therefore,

The referee forwards the state in

The remarkable implication of the EWL protocol for Prisoner’s Dilemma is that under the right subset of quantum strategies, this quantum extension of the game eliminates the dilemma and the resulting Nash equilibrium is Pareto-optimal! The quantum strategies that allow this are the two-parameter subset of the set of one qubit gates:

However, when the full set of quantum strategies is made available to the players (

The dilemma reappears in the quantum version of the game and the Nash equilibrium solution is the same as of the classical game. This is because a best reply to a quantum strategy from set

Note that while the EWL “quantum” Prisoner’s Dilemma is a game with quantum mediated communication, the equilibrium in the game is referred to as Nash equilibrium rather than correlated equilibrium. This is because mediated communication attempts to produce randomization over a game’s outcomes that cannot be produced by the player’s mixed strategies only, and therefore, one can view the game with mediated communication as an enlargement of the mixed game. A Nash equilibrium in this larger non-cooperative game is what is called a correlated equilibrium in the original game.

As such, Nash equilibrium in the EWL game with quantum mediated communication is a type of correalted equilibrium in the original Prisoner’s Dilemma. A discussion on the relationship between Nash and correlated equilibrium for classical games can be found in (

Emulating mixed strategies, a further natural quantum extension is possible by allowing players to randomize over their quantum strategies, giving rise to the notion of mixed quantum strategies. Eisert et al. showed that while the players cannot solve the dilemma by resorting to mixed quantum strategies in Prisoner’s Dilemma, they can come close to it. By using mixed quantum strategies, the players can affect a Nash equilibrium in which the payoff is

Motivated by the results of the seminal works of Meyer and Eisert et al., quantum game theory has become a major area of research since the seminal papers of Meyer and Eisert et al. A relatively recent and comprehensive review of the subject can be found in (

High-frequency trading (HFT) is defined by Gomber et al. in (

HFT relates to the implementation of proprietary trading strategies by technologically advanced market participants…. HFT enable market participants to dramatically speed up the reception of market data, internal calculation procedures, order submission and reception of execution confirmations.

Our aim here is to show that quantum computing via the cloud can be used to implement HFT as a quantum game. For this, first note that HFT is an instance of Prisoner’s Dilemma where Player I and Player II represent the trading mindset of a market, buying and selling of commodities using the two strategies Buy or Sell. Assuming that in markets there is a preference toward being part of a mass-buy versus a mass-sell, we set the following preferences for the players over the four possible strategy profiles as reasonably reflecting the mood of any market,

These preferences are identical to those in Prisoners’ Dilemma when the numerical payoff values from expression (1) are faithfully substituted into expression (8).

High-frequency trading as an instance of Prisoners’ Dilemma, as per the preferences described in expression (8).

Today, the internet is quasi-quantum, meaning that users can access third party, first generation quantum processors via the cloud (the quantum cloud), which can offer transnational speed up. More importantly, the quasi-quantum internet can offer enhanced payoffs in the transaction when implemented using the EWL protocol for Prisoner’s Dilemma.

Due to the quasi-quantum nature of the internet, only noisy quantum communications are possible to date. Therefore, the referee will likely only be able to create limited entanglement between the qubits. This means that HFT on the quantum cloud will improve the lot of the players to only a near Pareto-optimal Nash equilibrium, the upper-limit of which for the moment is the appropriate equivalent of the notional (2.5, 2.5) payoff. Nonetheless, even these small improvement in the payoffs will be worthwhile given the large amounts of money being traded.

In the near future, the internet will be fully quantum, and improved fidelity of the transmission of the quantum information will mean that quantum entanglement between the players’ qubits will be maintained for longer duration. This will allow the realization of the upper limit of the mixed quantum strategy Nash equilibrium, (2.5, 2.5).

From a non-cooperative game theory perspective, the pure quantum strategy Nash equilibrium that resolves the dilemma and produces the Pareto-optimal Nash equilibrium (3, 3) is fundamentally irrational. This is due to the fact that the best reply to any strategy from the set

We use the term cooperative to describe any possible outcome of a game, as long as no player can guarantee a better outcome for himself. It is important to emphasize that in general, a cooperative out-come is not in equilibrium; it’s the result of an agreement. For example, in the well-known “prisoner’s dilemma” game, the outcome in which neither prisoner confesses is a cooperative outcome; it is in neither player’s best interests, though it is better for both than the unique equilibrium.

Hence, the solution lies in the notion of agreement contracts and the ability to enforce them. For this, the game has to be played repeatedly and the behavioral history of the players collected and used to develop the contracts and the enforcement methods (incentives and disincentives). It is noteworthy then that quantum games such as the quantum prisoner’s dilemma can be thought of as the available policy space for an agent undergoing reinforcement learning. Here, however, it is known that the quantum policy options, in for example the quantum prisoner’s dilemma, are Pareto-optimal over the classical policy options. Therefore, if the task undertaken in quantum reinforcement learning can be thought of as having instances of the prisoner’s dilemma as subtasks, an agent with quantum strategies available to them will perform strictly better than one with only classical policy options, as observed by Meyer in his seminal work.

We established a game-theoretic interpretation of high-frequency trading as the game Prisoner’s Dilemma, and showed how it can be implemented as a quantum game using quantum computing processors available over the cloud. We argue that even today’s nascent quantum technology infrastructure allows substantial improvement in the payoffs of the players of this game, and that in the near future, a fully quantum internet and better performing quantum processors will allow players to completely avoid the dilemma via reinforced learning of contracts, as predicted by cooperative game theory.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

FSK crafted the majority of the narrative of the manuscript, produced the tables and figures, and compiled the references. NB offered insights into how to map HFT into Prisoner’s Dilemma.

FSK is employed by Dark Star Quantum Lab Inc.

The remaining author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

We would like to thank Nathan Benjamin, James Sully, Nathan Urban, for useful discussions. NB is supported by the Computational Science Initiative at Brookhaven National Laboratory.