“Options data add a crucial dimension:  they represent forward-looking view on volatility and risk that are not directly observable in cash equities.” ¹

Traditional research suggests that stock features anticipate both stock and options performance, while recent research indicates a change: option’s data may now serve as a more accurate and longer-term indicator of options returns, as well as stock performance.  

A key factor causing this shift is ‘informed trading’ – using information from alternative sources that is not currently or efficiently reflected in prices. Informed traders often prefer the options market due to its leverage opportunities and reduced short-selling constraints. This strongly suggests that options markets play a measurable and leading role in price discovery, often embedding information before it is reflected in the underlying equity market.

In a recent article, “Generative AI for Stock Selection” written by Keywan Rasekhschaffe ¹  – Product Manager at Code Willing and former Director of Quantitative Research at Quantbot Technologies – investigated whether large language models can automate the traditionally manual process of feature engineering. Using SpiderRock’s Option Features Dataset alongside analyst and price-volume data, Rasekhschaffe found that AI-generated features delivered Sharpe ratio improvements of 14 to 91% over traditional baselines. Notably, these signals showed low correlation with conventional factors, suggesting they capture genuinely new information. The research points to a practical path forward: when retrieval quality is controlled, generative AI can produce interpretable, additive signals while significantly reducing engineering effort.

Key Features of the SpiderRock Option Features Dataset

Key features include an earnings indicator for identifying trading days with earnings announcements, retail and institutional flow estimator proxies, risk-neutral implied metrics like skewness and kurtosis, and behavioral trading metrics such as speculative call volume and Put-Call ratios. Market regime measures include mean-reversion/trending indicators, implied volatility term structure slope, normalized Delta skew ratios, and underlying asset closing auction imbalance data.

The Option Features Dataset supports dynamic analysis, cluster analysis, strategy development, and alpha factor augmentation. It enables the construction of market sentiment indicators, prediction of directional or implied moves, and analysis of short versus long-term flow via open interest, volume, and ratios by expiry. Applications include using VWKS to predict returns, combining VWKS with earnings indicators to forecast news flow, analyzing implied skewness for tail risk, and using implied volatility to anticipate future volatility.

Key Users

Both buy-side and sell-side professionals use the Option Features Dataset to incorporate options market data into their workflows, run overlay strategies, model forward volatility, and enrich risk models or macro-driven strategies. Daily data is available from January 2014 and is accessible via AWS S3 and Snowflake.

 Historical Data Insights

The Option Features Dataset uses content from various historical datasets offered by SpiderRock.  SpiderRock is unique in that it offers not only end of day historical data but also intraday sets that all include both market data marked up with SpiderRock analytics including volatility surfaces, implied pricing and volatility and Greeks

• Option Intraday History (5-minute & 30-minute intervals)

Provides point in time snapshots of option strikes, across all U.S.-listed options. Each interval includes price, volume, size, implied volatilities, and full Greek metrics.

• Option Print Set

Contains every option print along with quote, surface, and SR probability details at print time. Each trade includes associated quote, volatility surface, trade size and volume, and a full slate of Greeks.

• Surface Curve Intraday History (5-minute & 30-minute snapshots)

Captures fitted volatility surface curves intraday, including spline parameters, ATM volatility points, skew or term‑structure metrics, and associated Greeks. Snaps are taken at both 5‑minute and 30‑minute frequencies.

These datasets are delivered in standardized tabular formats and validated daily for accuracy and completeness, ensuring institutional-grade reliability. Because institutions trade on the same analytics that power the SpiderRock engine in real time, the data is battled-tested and reflects actual market activity. This alignment enables a frictionless flow from back testing models to live trading environments, enabling strategy development and execution on a consistent data foundation.

These records are ideal for quant researchers, options strategists, traders, and risk teams who require actionable and accurate historical reference data for modeling volatility, analyzing gamma exposure, or understanding skew movements.

About SpiderRock Data and Analytics 

SpiderRock Data & Analytics is a division of SpiderRock Technology Solutions, a provider of industry-leading options trading solutions. SpiderRock Data and Analytics is an exchange-licensed redistributor of market data, providing US stocks, options, and futures market data in a raw and normalized format.  

SpiderRock’s proprietary live analytics offer low-cost delivery of market data and options analytics without requiring clients to make a significant investment in infrastructure. In addition, SpiderRock’s robust historical datasets updated daily from live markets are ideal for research, back testing, and making data-driven decisions.

For more information, please email DataSales@SpiderRock.net, visit www.spiderrock.net, or visit our LinkedIn page.

 About Code Willing, Inc.

Code Willing delivers a comprehensive quant solution, empowering hedge funds and asset managers to capture alpha and achieve business success through advanced data management and compute optimization. The company’s flagship product, the CWIQ Platform, simplifies the integration of critical capabilities into a seamless, end-to-end solution. 

To learn more about Code Willing, please email sales@codewilling.com, visit CodeWilling.com or their LinkedIn page.

¹”Generative AI for Stock Selection”, Keywan Rasekhschaffe, January 2026

Have any questions?