Wingman - an AI Volatility Trading Agent

Throughout finance, various analysts and engineers have tried to model markets and use those models to have an edge on the market. These methods and strategies vary across different asset classes, markets, and instruments. When it comes to options trading, the one model that stands out is the Black-Scholes-Merton model, taking into account the conditions of the contract, i.e. strike, price, time to expiry, as well as harder to measure risk-free rate and volatility. With the latter is where Wingman comes in. Wingman is an AI options trading model, designed to trade volatility on the market through calculating mispricings. It does this by using a modified version of the Black-Scholes-Merton model, called the "Mixture Dynamics" model. Through this model, it calculates an implied volatility surface from live market quotes rather than assuming a single flat volatility. Every 15 minutes, Wingman compares this model to the market: if a quote looks cheap relative to the model, it opens a long straddle; if it looks rich, it opens a short call vertical. If nothing clears the bar, it waits for the next cycle. Because a mispriced signal is not the same as a safe trade, every decision passes through independent checks before anything reaches the market: limits on position size and total exposure, protection against duplicate or conflicting trades, and a final AI-driven review that can reduce or block a trade based on current portfolio risk. Trades are then placed automatically through Alpaca's Trading API via its command-line interface, with no manual approval step. This makes Wingman suited to traders or desks who want a disciplined, risk-aware way to express a volatility view without watching the market themselves — and because the underlying approach isn't tied to any one stock, the same method could extend to other liquid options markets, from equity indices to FX and commodities.