Enhancing the Wheel Strategy with Bayesian Networks

The option wheel strategy is a systematic approach that combines selling cash-secured puts and covered calls. The process begins by selling puts on a stock the investor is willing to own; if assigned, the investor acquires the shares and then sells covered calls against the position to collect additional premium. The cycle repeats, though returns depend heavily on underlying volatility, assignment risk, and disciplined position management.

This is another popular options strategy among investors and was widely promoted by trading educators. However, experienced investors recognize that it suffers from the same drawback as the covered call strategy.

Reference [1] revisits the wheel strategy, but with a twist: it applies an LLM-based Bayesian network on top of the wheel framework. Essentially, this Bayesian network is used to characterize market regimes and guide position sizing and strike selection. The authors pointed out,

This paper introduces a novel model-first hybrid AI architecture that overcomes key limitations of using LLMs directly for quantitative financial decision-making, specifically in options wheel strategy decisions. Instead of employing LLMs as decision-makers, we use them as intelligent model constructors. This approach yields strong and stable returns with enhanced downside protection, achieving a Sharpe ratio of 1.08 and a maximum drawdown of -8.2%. The strategy delivers 15.3% annualized returns over 18.75 years (2007–September 2025), including volatile periods such as 2020–2022. Additionally, the model provides full transparency through 27 decision factors per trade… Our comprehensive baseline comparisons demonstrate the effectiveness of the model-first architecture. Pure LLM approaches yield 8.7% returns with a 0.45 Sharpe ratio. Static Bayesian networks achieve 11.2% returns and a 0.67 Sharpe ratio. Rules-based systems produce 9.8% returns with a 0.52 Sharpe ratio. In contrast, our hybrid approach attains 15.3% returns and a 1.08 Sharpe ratio, while maintaining superior risk management.

In short, using the LLM-based Bayesian network, the performance of the wheel strategy improved significantly.

We find the results appear unusually impressive and [glossary_exclude]warrant [/glossary_exclude]caution, but the underlying design and architecture are worth examining.

Let us know what you think in the comments below or in the discussion forum.

References

[1] Xiaoting Kuang, Boken Lin, A Hybrid Architecture for Options Wheel Strategy Decisions: LLM-Generated Bayesian Networks for Transparent Trading, arXiv:2512.01123

Post Source Here: Enhancing the Wheel Strategy with Bayesian Networks

source https://harbourfronts.com/enhancing-wheel-strategy-bayesian-networks/