Using the Heston Model to Estimate VaR and CVaR
Summary
The document discusses using the Heston stochastic volatility model to estimate Value at Risk and Conditional Value at Risk. It contrasts Heston’s joint stochastic modeling of asset prices and volatility with GARCH, which models conditional volatility from past returns and volatility. The response notes that Heston can represent volatility clustering and a leverage effect through the relationship between price and volatility shocks.
The proposed workflow is to calibrate model parameters to historical data, simulate asset paths, and calculate VaR and CVaR from the simulated outcomes. The response cautions that this can be more computationally demanding than GARCH because it requires simulating coupled price and volatility processes. It gives no empirical validation or implementation details, so model suitability depends on calibration quality and whether its assumptions fit the asset and risk horizon.
Key ideas
- Heston models asset prices and volatility as linked stochastic processes.
- GARCH estimates conditional volatility using past returns and volatility observations.
- VaR and CVaR can be estimated from simulated outcomes after calibrating Heston to historical data.
- Heston may capture volatility clustering and return-volatility correlation, at greater computational cost.
Tags
Full text
# Can Heston volatility model be used to calculate VaR or CVaR? # Can Heston volatility model be used to calculate VaR or CVaR? I'm just a beginner and third-year statistics major student. Based on what I read in some journal, most common model that used to calculate VaR or CVAR is GARCH. Is there any possibility that I can use Heston volatility model instead? Thank you. ## Answer by Amit Kumar Jha (score 4) https://quant.stackexchange.com/a/76762 Certainly! The Heston model is a well-known model in quantitative finance that describes the evolution of the volatility of an asset. It's a stochastic volatility model, meaning it assumes that the volatility of the asset is not constant but follows its own random process. The main distinction between the Heston model and the GARCH model is the way they model volatility: GARCH (Generalized Autoregressive Conditional Heteroskedasticity): This is a time series model that predicts volatility based on past returns and past volatilities. It assumes that volatility can be described by an autoregressive process, and it's often used in econometric studies. Heston Model: This is a stochastic volatility model that describes the evolution of an asset's price and its volatility using two stochastic differential equations (one for the asset price and one for the volatility). The Heston model allows for volatility clustering (periods of high volatility followed by periods of low volatility) and leverage effect (correlation between asset returns and volatility). You can definitely use the Heston model to compute Value at Risk (VaR) or Conditional Value at Risk (CVaR). The process would involve: - Calibrating the Heston model parameters to historical data. - Simulating many paths of asset returns using the calibrated Heston model. - Computing VaR or CVaR from the simulated paths . The Heston model might be more computationally intensive than GARCH due to the need to solve two coupled stochastic differential equations. However, it offers a more sophisticated representation of volatility dynamics. If you believe that the asset's volatility follows a more complex pattern that can't be captured by GARCH, then the Heston model might be a better choice. In practice, the choice between GARCH and Heston (or any other model) often comes down to a trade-off between model complexity, computational efficiency, and the ability to fit the observed data
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.