Abstract
This paper investigates the pricing of derivatives in a setting characterized by trading restrictions and a lack of probabilistic assumptions on the underlying asset model. By examining both discrete and continuous-time frameworks, the authors demonstrate how short-selling restrictions interact with market-observed option prices. Specifically, while call options maintain a standard pricing-hedging duality, the use of put options can lead to a "duality gap." This gap provides a robust, model-independent justification for the emergence of strict local martingale models, which are frequently used in the literature to represent financial bubbles. The results suggest that bubbles can be interpreted as a rational consequence of the interaction between market prices and structural trading constraints.
Introduction
In classical mathematical finance, derivative pricing typically begins with the assumption of a specific probability measure (e.g., the Black-Scholes model). However, "robust" or model-independent pricing seeks to derive bounds on option prices using only the observed market prices of liquid instruments, such as calls and puts, without assuming a specific stochastic process.
While most robust pricing literature assumes frictionless markets, this paper introduces a realistic constraint: short-selling restrictions. By incorporating a ban on short-selling, the authors explore how the fundamental relationship between hedging costs and model-based expectations—known as the pricing-hedging duality—is altered. Their findings reveal that these restrictions naturally give rise to phenomena that mirror financial bubbles, specifically through the emergence of strict local martingales.
Theoretical Framework: Robustness and Supermartingales
The authors utilize a discrete-time setup where the absence of short-selling implies that risk-neutral measures are no longer required to be martingales. Instead, they must be supermartingales, where the expected future price of an asset is less than or equal to its current price.
To maintain a model-independent approach, the paper defines a "prediction set" $\mathfrak{P}$, representing all possible paths the asset price might take. This framework allows for results that are valid across all consistent models rather than being tied to a single probability distribution. The core of the analysis involves comparing two values:
- The Super-replication Price ($V$): The minimum cost required to construct a foolproof hedge against all paths in $\mathfrak{P}$.
- The Primal Value ($P$): The highest expected payoff of the derivative across all calibrated supermartingale measures.
The Call-Put Asymmetry and the Duality Gap
A central insight of the paper is the breakdown of symmetry between call and put options. In frictionless markets, put-call parity allows one to be transformed into the other. However, under short-selling restrictions, this symmetry vanishes.
The authors prove that if call options are the primary traded instruments, the standard pricing-hedging duality holds ($V = P$). However, when only put options are traded, a duality gap can emerge. This gap occurs when the market-implied forward price (derived from put prices) is strictly lower than the current spot price.
The Bubble Formula
For a market where only puts are traded, the authors derive an explicit formula for this gap: $$V = P + \beta(s_0 - f_0)$$ Where:
- $s_0$ is the current spot price.
- $f_0$ is the "fundamental" forward price implied by the put options.
- $\beta$ represents the linear growth of the option's payoff.
This "bubble" term, $\beta(s_0 - f_0)$, represents the discrepancy between the cost of a physical hedge and the fundamental value of the asset. This discrepancy is shown to be the discrete-time analogue of the "strict local martingale" behavior seen in continuous-time models of financial bubbles.
Extension to Exotic Options
The paper extends this duality gap analysis to more complex derivatives, demonstrating that the "bubble" correction is pervasive:
- Asian Options: For options based on the average price over $n$ periods, the correction is $\frac{1}{n} \sum (s_0 - m_i)$, where $m_i$ are the implied means at different time steps.
- Lookback Options: For options based on the maximum price achieved, the gap is similarly linked to the difference between the spot price and the implied supermartingale expectations.
Connection to Financial Bubbles
In continuous-time finance, bubbles are often modeled using strict local martingales—processes that look like martingales locally but lose value on average over time. While mathematically useful, their application can sometimes feel arbitrary. This paper provides an intrinsic justification for their use: they are the natural mathematical limit of a market where short-selling is restricted and put prices suggest a lower future value than the current spot price.
Furthermore, the authors link these gaps to pathwise collateral requirements. In a robust setting, a hedging strategy must remain solvent across all possible future paths. The cost of maintaining this solvency in the face of trading restrictions is what ultimately drives the wedge between the fundamental price and the market price.
Conclusion
By relaxing the assumption of frictionless trading, this research provides a robust framework for understanding how market constraints generate price discrepancies. The emergence of a duality gap when short-selling is banned suggests that financial bubbles are not necessarily the result of irrationality, but can be a rational outcome of the interaction between hedging requirements and structural market limitations. Future work may expand these results to include partial short-selling bans or more complex dynamic trading environments.
Summary updated