Arbitrage-free prediction of the implied volatility smile

Πέτρος Δελλαπόρτας, Aleksandar MijatovićarXiv 2014 · DOI · arXiv

Abstract

This paper gives an arbitrage-free prediction for future prices of an arbitrary co-terminal set of options with a given maturity, based on the observed time series of these option prices. The statistical analysis of such a multi-dimensional time series of option prices corresponding to $n$ strikes (with $n$ large, e.g. $n\geq 40$) and the same maturity, is a difficult task due to the fact that option prices at any moment in time satisfy non-linear and non-explicit no-arbitrage restrictions. Hence any $n$-dimensional time series model also has to satisfy these implicit restrictions at each time step, a condition that is impossible to meet since the model innovations can take arbitrary values. We solve this problem for any $n\in\mathbb{N}$ in the context of Foreign Exchange (FX) by first encoding the option prices at each time step in terms of the parameters of the corresponding risk-neutral measure and then performing the time series analysis in the parameter space. The option price predictions are obtained from the predicted risk-neutral measure by effectively integrating it against the corresponding option payoffs. The non-linear transformation between option prices and the risk-neutral parameters applied here is not arbitrary: it is the standard mapping used by market makers in the FX option markets (the SABR parameterisation) and is given explicitly in closed form. Our method is not restricted to the FX asset class nor does it depend on the type of parameterisation used. Statistical analysis of FX market data illustrates that our arbitrage-free predictions outperform the naive random walk forecasts, suggesting a potential for building management strategies for portfolios of derivative products, akin to the ones widely used in the underlying equity and futures markets.

Introduction

In financial markets, predicting the evolution of the "volatility smile"—the U-shaped pattern formed by implied volatilities across different strikes—is a critical yet complex task. Standard statistical models often fail because they do not account for no-arbitrage constraints. In a liquid market, prices must satisfy strict mathematical properties to prevent "free money" strategies. Traditional time-series models like ARMA introduce random noise that can easily push predictions into states that allow for arbitrage, rendering them useless for practical trading.

Methodology

The authors propose a "parameter-space" modeling framework to solve this. Instead of modeling the high-dimensional vector of option prices directly, they model the parameters of a risk-neutral measure (specifically the SABR model). This ensures that any prediction generated is mathematically consistent and arbitrage-free by construction.

Schematic of the parameter-space prediction framework.

As shown in the framework above, the process follows four distinct steps:

  1. Encoding: Daily implied volatility (IV) data is mapped to the parameters of the SABR model ($\alpha$, $\nu$, and $\rho$).
  2. Transformation: Parameters are mapped onto the real line using log and logit-like transformations to ensure they remain within valid bounds (e.g., positive volatility).
  3. Time Series Modeling: An ARMA-GARCH model with Student-t errors is fitted to these transformed parameters to capture momentum and volatility clustering.
  4. Decoding: The model projects these values forward, which are then transformed back into SABR parameters to generate a guaranteed arbitrage-free volatility smile prediction ($\hat{\mathbf{Y}}_{t+1}$).

Data and Experimental Setup

The study utilizes a high-quality proprietary dataset of USDJPY (US Dollar / Japanese Yen) liquid one-month options provided by the RBS FX options desk. The dataset spans from September 2006 to December 2011, comprising 1,360 daily observations. The authors used the first 1,000 days for model training and the final 360 days for out-of-sample testing to evaluate predictive accuracy against a standard random walk benchmark.

Results and Discussion

The methodology consistently outperformed the naive random walk forecast across all strikes. Key findings include:

  • Predictive Accuracy: The model achieved a Mean Absolute Error (MAE) of approximately 28 basis points for one-day-ahead predictions.
  • Tail Sensitivity: The greatest improvements (up to 4 basis points) were observed in "out-of-the-money" options, which are highly sensitive to the extreme events captured by the SABR parameters.
  • Trading Performance: Simulated trading strategies using strangles and risk-reversals yielded consistent positive returns, suggesting the model captures genuine market signals rather than statistical noise.

Conclusion

By modeling in the parameter space of a risk-neutral measure, this paper provides a robust method for generating arbitrage-free volatility predictions. While the study focused on FX markets, the framework is theoretically applicable to other asset classes like equities. Future work may explore using the information within the volatility smile to improve predictions of the underlying asset's forward rates.

Summary updated

Abstract

This paper gives an arbitrage-free prediction for future prices of an arbitrary co-terminal set of options with a given maturity, based on the observed time series of these option prices. The statistical analysis of such a multi-dimensional time series of option prices corresponding to $n$ strikes (with $n$ large, e.g. $n\geq 40$) and the same maturity, is a difficult task due to the fact that option prices at any moment in time satisfy non-linear and non-explicit no-arbitrage restrictions. Hence any $n$-dimensional time series model also has to satisfy these implicit restrictions at each time step, a condition that is impossible to meet since the model innovations can take arbitrary values. We solve this problem for any $n\in\NN$ in the context of Foreign Exchange (FX) by first encoding the option prices at each time step in terms of the parameters of the corresponding risk-neutral measure and then performing the time series analysis in the parameter space. The option price predictions are obtained from the predicted risk-neutral measure by effectively integrating it against the corresponding option payoffs. The non-linear transformation between option prices and the risk-neutral parameters applied here is \textit{not} arbitrary: it is the standard mapping used by market makers in the FX option markets (the SABR parameterisation) and is given explicitly in closed form. Our method is not restricted to the FX asset class nor does it depend on the type of parameterisation used. Statistical analysis of FX market data illustrates that our arbitrage-free predictions outperform the naive random walk forecasts, suggesting a potential for building management strategies for portfolios of derivative products, akin to the ones widely used in the underlying equity and futures markets.
BibTeX
@misc{2014arbitrage,
  author = {Πέτρος Δελλαπόρτας and Aleksandar Mijatović},
  title = {{Arbitrage-free prediction of the implied volatility smile}},
  year = {2014},
  eprint = {1407.5528},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1407.5528},
}

Loading related content...

Figure