Managing an option book is not trivial. There is considerable research on hedging errors, optimal hedging frequency, and the choice of volatility inputs used in hedging algorithms. Reference [1] continues this line of research by examining the effectiveness of hedging strategies under five different volatility measures:
- Flat ATM implied volatility
- Stochastic Volatility Inspired (SVI) implied volatility
- Close-to-close realized volatility
- Parkinson realized volatility
- Yang–Zhang realized volatility.
The study uses OptionMetrics SPX options data from 2019 to 2024, including the COVID crash and the 2022 hiking cycle, and analyzes 2000 stratified options across four VIX regimes. The author pointed out,
This thesis provides the first empirical evaluation of SVI-calibrated implied volatility as a delta-hedging input on real S&P 500 index options. The results challenge three common assumptions.
First, more information does not automatically improve hedging. The SVI surface, despite encoding the full implied volatility smile with high calibration quality (median RMSE of 19.5 bps), produces 9.4% higher hedging error variance than flat ATM implied volatility. The calibration noise embedded in the five-parameter SVI fit outweighs the informational benefit of strike-specific volatilities for most option types.
Second, statistical efficiency in volatility estimation does not translate into hedging efficiency. The simple close-to-close realized volatility estimator, which uses only closing prices, outperforms both the Parkinson and Yang–Zhang estimators, which incorporate intraday range information. For hedging purposes, smoothness (low sensitivity to intraday noise) appears more valuable than efficiency (low estimation variance under GBM).
Third, and most importantly, the optimal volatility input is strongly dependent on option moneyness and market regime. SVI surface IV reduces hedging error for out-of-the-money calls by 6–12%, where the smile slope carries genuine hedgeable information. Realized volatility dominates for out-of-the-money puts, where the skew premium makes implied-based deltas biased. No single input wins everywhere, suggesting that a moneyness-conditional hedging strategy, using different volatility inputs for different regions of the option space, would outperform any static approach.
In short, the results show that, contrary to intuition, SVI surface volatility does not improve aggregate delta hedging performance. In fact, SVI increases the standard deviation of hedging errors by 9.4% relative to flat ATM implied volatility. Close-to-close realized volatility performs best overall, reducing hedging error standard deviation by 5.8%, while the Parkinson estimator performs worst. The effectiveness of volatility measures, however, is also found to depend on regime and moneyness.
Another important finding is that higher-order effects, including vanna, volga, jumps, and model misspecification, dominate hedging errors.
This paper provides valuable insights for both traders and risk managers. Let us know what you think in the comments below or in the discussion forum.
References
[1] Annigeri, Z. (2026), Regime-Dependent Delta Hedging with SVI-Calibrated Volatility Surfaces: An Empirical Analysis of SPX Index Options, Rutgers Business School, SSRN 6465741
Further questions
What's your question? Ask it in the discussion forum
Have an answer to the questions below? Post it here or in the forum
