Delta Hedging Performance Under Different Volatility Measures
A recent study using OptionMetrics SPX data from 2019‑2024 compares five volatility inputs for delta‑hedging. The SVI‑calibrated implied volatility surface, despite fitting the smile, increases hedging error variance 9.4% over a flat ATM IV. Close‑to‑close realized volatility cuts the error standard deviation by 5.8%, outperforming Parkinson and Yang‑Zhang estimators. Effectiveness varies by option moneyness and market regime, indicating that conditional volatility inputs can beat static approaches.
Volatility Derivatives and VIX Market Dynamics
The article reviews recent research on VIX market dynamics, highlighting how the lead‑lag relationship between the VIX index and VIX futures has shifted from early dominance by the spot index to a more balanced, sometimes futures‑leading, pattern after the introduction...
Regime-Aware Trading Strategies with Machine Learning
The study introduces a regime‑aware trading framework that couples a Hidden Markov Model with LightGBM, a gradient‑boosting algorithm, to forecast equity markets. Using 63 features—including technical, macro, and cross‑asset data—the model outperforms XGBoost, logistic regression, SMA crossovers and momentum strategies....
Gamma Exposure and S&P500 Return Predictability
A recent study finds that changes in aggregate gamma exposure (GEX) in S&P 500 options predict next‑day equity returns. The derivative of GEX shows a statistically significant relationship with subsequent performance, robust both before and after 2020 though slightly weaker in...
Determining Implied Volatilities of American Options Using the Willow Tree Method
Researchers propose using the Willow tree method to compute implied volatilities for American options, overcoming the speed limitations of traditional bisection techniques that apply only to European options. The approach requires calculating transition probabilities once, then updating tree nodes as...
Why Backtests Decay: Regime Dependence and Crowding
A new study of 1,726 commercially marketed strategies from ten global institutions (2009‑2025) finds that backtested performance typically erodes by 2‑3% per year once the strategy goes live. The analysis shows that most of the apparent skill in backtests stems...
Forecasting Earnings and Returns
Recent advances in data science and machine learning have transformed fields like computer vision and natural language processing, yet financial forecasting remains stubbornly difficult. The literature highlights three core challenges—unpredictable earnings and returns, noisy explanatory variables, and model uncertainty—driven by...
Volatility Risk Premium Dynamics Through the Heston Framework
The paper links the volatility risk premium (VRP) to Heston model parameters using market‑based proxies such as variance swaps, VIX futures, and straddles. Over 7‑day horizons, a one‑standard‑deviation rise in initial variance (v0) cuts next‑day variance‑swap returns by roughly 730...
Where Do Options Returns Come From
The paper by Broadie, Chernov and Johannes examines index option returns and concludes that options are not systematically mispriced. Individual put returns are highly dispersed and offer little diagnostic value, while straddle portfolios expose risk premiums that standard Black‑Scholes and...
Overnight vs Daytime Returns in Sector ETFs
A new study of SPY and nine sector ETFs from 1999‑2025 finds that overnight returns generate a strong, exploitable momentum, while daytime returns are weak or negative. A simple long‑only overnight strategy (Strategy #1) outperformed buy‑and‑hold in every ETF, delivering...
Dispersion Trading Using Principal Component Analysis
Dispersion trading leverages statistical arbitrage by using options on a select group of stocks rather than the whole index. The authors applied Principal Component Analysis to the S&P 500, creating a replicating option basket of as few as five securities. Their...
Regime Classification Framework for Mean-Reverting and Trending Markets
The paper introduces a regime‑classification framework that labels markets as mean‑reverting or trending, using return thresholds of 0.5%, 0.75% and 1% on SPY, QQQ, DIA and IWM from 2000‑2024. Three machine‑learning models—Random Forest, Neural Network (MLP) and XGBoost—are tested with...
Volatility Risk Premium and Clustering: Intraday vs Overnight Dynamics
Recent academic work separates the variance risk premium (VRP) into overnight and intraday components, revealing that the overall negative VRP documented in prior literature is driven almost entirely by the overnight period. Overnight VRP remains significantly negative and predicts longer‑term...
Price Dynamics of Volatility Indices
A 2022 study applied an ARFIMA model to daily data from 2007‑2020 for nine volatility indices, including VIX, VXN, VXO, VXD, RVX, VPD, OVX, VVIX and SKEW. The analysis revealed persistent long‑memory behavior, supporting the fractal market hypothesis and suggesting...
Variational Autoencoders in Volatility and Option Pricing
A new semi‑parametric framework combines a Variational Autoencoder (VAE) with a LightGBM‑based implied volatility model and a Multi‑Level Monte Carlo (MLMC) pricing engine to price options on the NIFTY50 index. The VAE learns non‑Gaussian return distributions, preserving tail events, while...