Stochastic Volatility Models

Modelling Uncertainty as a Dynamic Process in Financial Markets

Volatility lies at the centre of financial markets.

It represents uncertainty, risk, and the dispersion of possible outcomes. Within classical frameworks such as Black–Scholes, volatility is treated as a constant parameter. This assumption enables mathematical tractability and provides a clear structure for pricing and hedging.

However, empirical evidence suggests that volatility is neither constant nor stable. It evolves over time, clusters in periods of heightened activity, and responds to both market dynamics and external shocks. These characteristics cannot be fully captured within models that assume fixed volatility.

Stochastic volatility models address this limitation by treating volatility itself as a random process. In doing so, they provide a richer and more realistic framework for understanding financial markets.

From Constant to Random Volatility

The key innovation of stochastic volatility models is conceptual. Rather than assuming that volatility is known and constant, it is modelled as an evolving variable.

In its simplest form, the price of an asset and its volatility are described by two coupled stochastic processes. The asset price evolves according to a diffusion process, while volatility follows its own dynamics, often characterised by mean reversion and random fluctuations.

This introduces an additional layer of uncertainty. Markets are no longer described by a single source of randomness, but by multiple interacting sources.

Volatility as a State Variable

In this framework, volatility becomes a state variable. It is no longer an input to the model, but a quantity that evolves alongside price.

This has several implications. First, volatility can change independently of price movements. Periods of relative price stability may still exhibit changing levels of uncertainty. Second, volatility can exhibit persistence. High-volatility periods tend to be followed by further high volatility, a phenomenon known as volatility clustering. Third, volatility can revert toward a long-term average, reflecting structural properties of the market.

These features align more closely with observed market behaviour.

The Structure of Stochastic Volatility Models

While many specific formulations exist, most stochastic volatility models share common elements.

They typically include:

• a stochastic process for the underlying asset price

• a separate stochastic process for volatility

• a correlation between the two processes

This correlation is particularly important. As in equity markets, negative correlation between price and volatility is often observed. Declining prices tend to coincide with rising volatility, a phenomenon sometimes referred to as the leverage effect. Capturing this relationship is essential for realistic modelling.

Examples of Stochastic Volatility Frameworks

Several models have been developed to formalise these ideas.

The Heston model is one of the most widely used. It assumes that volatility follows a mean-reverting square-root process, allowing for analytical tractability while capturing key features of market behaviour. Other approaches include models with log-normal volatility dynamics or those incorporating additional factors.

While the mathematical details differ, the underlying principle remains consistent:

volatility is treated as a dynamic, uncertain quantity.

Implications for Option Pricing

The introduction of stochastic volatility has significant implications for option pricing.

Under constant volatility, models such as Black–Scholes imply flat volatility surfaces. In reality, implied volatility varies across strike prices and maturities, producing patterns such as the volatility smile and skew.

Stochastic volatility models provide a mechanism for generating these patterns. By allowing volatility to evolve and interact with price, they produce option prices that reflect the observed structure of implied volatility.

This represents a key advancement. It moves models closer to empirical reality without abandoning the analytical framework of derivative pricing.

Risk and Uncertainty

Modelling volatility as stochastic introduces a more nuanced view of risk. Risk is no longer captured solely by the variability of prices. It includes uncertainty about the variability itself.

This second-order uncertainty has important consequences.

• hedging becomes more complex

• risk exposure depends on multiple factors

• model outputs become more sensitive to assumptions

Understanding and managing this layered uncertainty is central to modern quantitative finance.

Calibration and Practical Challenges

While stochastic volatility models provide greater realism, they also introduce complexity. Parameters must be estimated from data, often requiring calibration to observed market prices.

This process can be sensitive.

• multiple parameter combinations may fit the data

• calibration may change across time and conditions

• computational requirements increase

As a result, the practical application of these models requires careful implementation and interpretation.

Limitations and Extensions

Despite their advantages, stochastic volatility models are not complete representations of market behaviour.

They typically assume continuous price evolution and may not fully capture:

• sudden jumps in prices

• extreme tail events

• structural breaks

To address these limitations, models have been extended to include jump components or additional stochastic factors. However, each extension increases complexity. This reflects a broader trade-off in modelling. Greater realism often comes at the cost of tractability.

The MorMag Perspective

At MorMag, stochastic volatility models are viewed as an important step toward capturing the dynamic nature of markets.

They align with the broader principle that:

• uncertainty evolves

• relationships change

• models must reflect this variability

Within the Market Scanner and broader analytical framework, volatility is not treated as static. It is monitored, interpreted, and incorporated into decision-making as a dynamic factor.

At the same time, the limitations of any single model are recognised. Quantitative outputs are evaluated within a system that integrates probabilistic reasoning, behavioural insights, and structural analysis.

From Static Assumptions to Dynamic Systems

The transition from constant to stochastic volatility reflects a deeper shift in financial modelling. It moves from a static representation of uncertainty to a dynamic one. This shift aligns with the broader view of markets as complex, adaptive systems.

In such systems:

• variables evolve

• interactions matter

• outcomes are not fully predictable

Stochastic volatility models provide a framework for engaging with this complexity.

Conclusion

Stochastic volatility models represent a significant advancement in the modelling of financial markets.

By treating volatility as a dynamic and uncertain process, they capture key features of real-world behaviour, including clustering, mean reversion, and correlation with asset prices. While they introduce additional complexity, they provide a more realistic representation of uncertainty and its impact on pricing and risk.

At MorMag, this perspective informs a disciplined approach to quantitative analysis. Models are used to structure understanding, but they are applied within a broader framework that recognises the evolving and uncertain nature of financial markets.

In modern finance, volatility is not a fixed input, it is a process. Understanding that process is essential for navigating markets with clarity and precision.