Paper: arXiv 2307.14887

Authors: Josef Teichmann, Hanna Wutte

Abstract

Introduced in the late 90s, the passport option gives its holder the right to trade in a market and receive any positive gain in the resulting traded account at maturity. Pricing the option amounts to solving a stochastic control problem that for \(d>1\) risky assets remains an open problem. Even in a correlated Black-Scholes (BS) market with \(d=2\) risky assets, no optimal trading strategy has been derived in closed form. In this paper, we derive a discrete-time solution for multi-dimensional BS markets with uncorrelated assets. Moreover, inspired by the success of deep reinforcement learning in, e.g., board games, we propose two machine learning-powered approaches to pricing general options on a portfolio value in general markets. These approaches prove to be successful for pricing the passport option in one-dimensional and multi-dimensional uncorrelated BS markets.

Complexity vs Empirical Score

  • Math Complexity: 9.0/10
  • Empirical Rigor: 4.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper introduces complex stochastic control and multi-dimensional partial differential equations with extensive mathematical derivations, but its empirical validation is limited to simulated data without reported backtests or statistical performance metrics.

Research Flowchart

  flowchart TD
  A["Research Goal: Price<br>Multi-dim Passport Options"] --> B{"Methodology Approach"}
  B --> C["Analytical: Discrete-Time Solution<br>Uncorrelated BS Assets"]
  B --> D["ML: Deep Reinforcement Learning<br>Agent-based Pricing"]
  
  C --> E["Computational Process:<br>Stochastic Control Optimization"]
  D --> F["Computational Process:<br>Reinforcement Learning Simulation"]
  
  E --> G["Key Findings/Outcomes"]
  F --> G
  
  G --> H["✓ Closed-form price for uncorrelated assets<br>✓ ML successfully prices 1D & multi-dim options<br>✓ Framework for general markets"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes two machine‑learning‑powered approaches—a policy‑gradient method and an advantage‑actor‑critic (A2C) method—for pricing passport options in multi‑dimensional markets.

Source sentence
Moreover, inspired by the success of deep reinforcement learning in, e.g., board games, we propose two machine learning-powered approaches to pricing general options on a portfolio value in general markets. These approaches prove to be successful for pricing the passport option in one-dimensional and multi-dimensional uncorrelated BS markets.

Assumes 3

  • The underlying market follows a Black‑Scholes model with uncorrelated risky assets. market
    Source sentence
    In this paper, we derive a discrete-time solution for multi-dimensional BS markets with uncorrelated assets.
  • Pricing is performed in discrete time. model
    Source sentence
    In this paper, we derive a discrete-time solution for multi-dimensional BS markets with uncorrelated assets.
  • The option holder can trade freely within pre‑specified constraints (passport option definition). other
    Source sentence
    Introduced in the late 90s, the passport option gives its holder the right to trade in a market and receive any positive gain in the resulting traded account at maturity. Introduction A passport option gives its holder the right to freely trade in a certain market within pre-specified constraints, over a predetermined time interval.

Data

  • Synthetic Black‑Scholes market simulations synthetic unknown
    one‑dimensional and two‑dimensional asset setups (uncorrelated and correlated)
    Source sentence
    Experiments In this section, we experimentally evaluate our proposed algorithms PG (Algorithms 1 and 2) and A2C (Algorithm 3) for pricing passport options in one- and two-dimensional BS markets. Section 5.1 and treats the one-dimensional case, where we show that both algorithms recover the well-known solution of Equation (17).

Costs

Not modeled A trading or execution result is reported without costs.

Finds 3

  • In the one‑dimensional Black‑Scholes case the PG and A2C algorithms recover the known optimal strategy q* = –sign(x). performance
    Source sentence
    Section 5.1 and treats the one-dimensional case, where we show that both algorithms recover the well-known solution of Equation (17). Thus, PG and A2C both manage to capture the optimal strategy q∗(t, s, x) = −sign(x).
  • In the two‑dimensional uncorrelated asset case the algorithms recover the solution derived analytically (Theorem 3). performance
    Source sentence
    In Section 5.2 we test the algorithms on the case of two uncorrelated assets and find that, also in this setting, they recover the solution derived in Theorem 3 in Section 5.2.1. Thus, PG and A2C both manage to capture the optimal strategy q∗(t, s, x) = −sign(x).
  • For correlated two‑asset markets, the learned strategies perform statistically on par with the optimal strategy obtained for the uncorrelated case, with overlapping confidence intervals of terminal payoff means. performance
    Source sentence
    We see that both PG and A2C yield strategies that perform statistically on par with the optimal one, as indicated by overlapping confidence intervals of the means of terminal payoff distributions (right sub-plot in Figure 8). In particular, we observe in Figure 16 that both the PG and the A2C algorithms yield strategies that perform comparably to the strategy of Theorem 3 that was optimal only in the market with ρ =

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 12 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

https://github.com/HannaSW/ML4PassportOptions — the authors' own repository

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.