Volatility is not just stochastic. It is rough.

Classical stochastic volatility models assume volatility evolves smoothly through time, reverting toward a long-run mean with relatively regular dynamics. High-frequency data tells a different story. Volatility is jagged, persistent, and exhibits memory across multiple time scales.

The Rough Heston model combines two powerful ideas: the tractability of the Heston framework and the empirical realism of rough volatility. By introducing fractional dynamics into the variance process, it captures both stochastic volatility and the roughness observed in real markets.

The result is a model that retains much of Heston's analytical appeal while producing volatility dynamics that look far more like the world we actually trade.

📚 Model Definition

This following structure is an example of a Markov chain...

At a high level, the Rough Heston model combines three ideas: stochastic volatility, mean reversion, and roughness.

The asset price, Sₜ, evolves according to the instantaneous variance Vₜ, just as in the classical Heston model. Higher variance produces larger price fluctuations, linking the behavior of the underlying directly to the volatility process.

The variance process itself has two components. The first is a mean-reverting term, controlled by λ and θ, which pulls volatility back toward a long-run level. The second is a stochastic shock term, scaled by ξ, which introduces randomness into the evolution of volatility.

The key difference is the kernel K(t). Rather than allowing volatility to evolve with smooth, Markovian dynamics, the power-law kernel introduces memory through the 

Hurst parameter H. When H is small, volatility becomes rough, exhibiting the jagged, persistent behavior observed in empirical data.

Finally, the Brownian motions W and B are correlated through ρ, generating the familiar leverage effect where falling prices tend to coincide with rising volatility.

Effectively, the Rough Heston takes the classical Heston framework and replaces smooth volatility dynamics with a fractional, memory-driven process. The result is a model that captures both stochastic volatility and the roughness that real markets appear to exhibit.

📈 Model Applications

In practice, Rough Heston is used for pricing and hedging derivatives in markets where volatility exhibits both stochastic behavior and rough dynamics.

It is particularly valuable for equity and index options, where classical models often struggle to reproduce the short-maturity volatility smile. By incorporating memory through the rough kernel, Rough Heston produces implied volatility surfaces that align much more closely with observed market data.

The model is also used in Monte Carlo simulation, volatility surface construction, and model calibration. Institutions can fit the parameters to option prices and use the resulting dynamics for pricing exotic derivatives and managing volatility risk.

Of course, this realism comes at a cost. Rough Heston is significantly more computationally demanding than its classical counterpart, and calibration can be challenging. But that tradeoff is common in quantitative finance: better empirical fit often means greater mathematical and computational complexity.

Ultimately, Rough Heston represents a shift in philosophy. Instead of assuming volatility is smooth because it is convenient, it attempts to model volatility as it appears to behave in reality—rough, persistent, and far from memoryless.

🎓 A Little Story

The first time I learned about long memory processes, I thought they were straight up fucking cool.

Seriously. You spend years being told that everything is Markovian. The future depends on the present. Memoryless dynamics. Nice, clean mathematics.

And then someone comes along and says, "Actually, what if the process remembers?"

That completely broke my brain.

The idea that events from the distant past could still influence the present through a power-law decay felt less like mathematics and more like science fiction. And the craziest part was that the empirical evidence suggested markets might actually behave this way.

I remember going down an absolute rabbit hole: fractional Brownian motion, rough volatility, Hurst exponents, power-law kernels. Every paper I read made me more convinced that this was one of the coolest ideas in quantitative finance.

Whether rough volatility is the final answer is almost beside the point.

The first time you realize that stochastic processes do not have to forget, you start looking at randomness very differently.

💡 Takeaway

Rough Heston is a reminder that markets are often more complicated than our models would like them to be.

Classical volatility models assume smoothness because it is mathematically convenient. Rough Heston asks what happens if we instead model volatility as it appears to behave: irregular, persistent, and influenced by the distant past.

Whether rough volatility is the final word remains an open question. But the broader lesson is an important one: sometimes the path to better models is not adding more parameters, but relaxing the assumptions we have taken for granted.

After all, if volatility remembers, perhaps our models should too.

Need to study up on topics in math, probability, and finance?

👉 Learn to solve problems like this on Quant Guild

 — the platform I wish I had when I was studying to become a quant.

The Black-Scholes model enables us to produce market implied volatilities, but it is incapable of calibrating to the volatility surface for each option on the grid simultaneously as the underlying process assumes constant volatility.