Choosing Option Models for Delta Hedging
Summary
The document considers whether a practical option model can produce better delta hedges than Black–Scholes for equity options. One response says the Heston model is relatively quick to implement analytically, but may still miss important features of volatility dynamics. Rough volatility models may better reflect observed behavior, though their complexity and computational cost can make them difficult to use for fast calculations; Monte Carlo simulation is suggested as an alternative to closed-form solutions.
A second response describes the binomial tree model, which steps through possible up and down stock-price moves and works backward from expiration to value the option. It can accommodate changing volatility and early exercise for American options. The discussion offers implementation guidance but no hedge-performance comparisons or empirical results. It also does not specify a market calibration, hedge frequency, or how model risk affects realized hedging outcomes, so it does not establish that either alternative will outperform Black–Scholes in practice.
Key ideas
- Heston is presented as a comparatively accessible model, but its volatility dynamics may remain unrealistic.
- Rough volatility models may better capture volatility behavior, with greater computational and implementation demands.
- Monte Carlo simulation can be used when a closed-form solution is difficult to implement.
- A binomial tree values options by stepping through up and down price moves and working backward from expiry.
- The binomial model can represent early exercise for American options.
Tags
Full text
# A decent model to calculate hedges # A decent model to calculate hedges Is there an option pricing model that wouldn't be too time consuming to set up in Python (for example) and that would provide better delta hedges than Black-Scholes? This would be mainly for equity options. ## Answer by Elyes Mahjoubi (score 1) https://quant.stackexchange.com/a/63875 You sometimes can’t have both simplicity and robustness.However,the Heston Model analytical solution shouldn’t take you too much time to code but isn’t a great one,mainly because the volatility is not a diffusion/super-diffusion.Newer and more complex models like rough volatility models are closer to the reality but their generation is still not efficient enough to allow very fast computation.But you can still try them even if the analytical closed forms are hard to code for you.You can still use a Monte-Carlo generation (rough volatility models are martingale so we remove the drift for the volatility process) ## Answer by Amit Kumar Jha (score 0) https://quant.stackexchange.com/a/76763 Certainly! The Black-Scholes model, while pioneering and foundational, is based on several simplifying assumptions that might not hold in the real world. One of its main assumptions is that volatility is constant, which is often not the case. Several models have been developed to address its limitations. One such model that's relatively straightforward to implement and provides more realistic dynamics than Black-Scholes is the Binomial Option Pricing Model. This model is particularly useful for American-style options, which can be exercised before the expiration date. Binomial Option Pricing Model: The binomial model breaks down the time to expiration into potentially very many time intervals, or steps. In each step, the stock price can move up or down. The model then computes the option value at each step, starting from the expiration date and working backward to the present. Advantages: Can handle varying volatility. Suitable for American options. Provides an intuitive representation of the option pricing process. To implement the Binomial Option Pricing Model in Python: Set up the binomial price tree for the stock. Calculate option values at the final nodes (at expiration). Work backwards, adjusting for the risk-free rate and the potential early exercise for American options, to get the option price at the root.
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.