Optimal Trading with Learning in Latent Alpha Models

Summary

This paper studies optimal trading when a statistical arbitrage signal comes from hidden factors that make prices jump and diffuse. The trader’s orders can also affect quoted prices and execution prices, so the problem includes both uncertain market states and the trader’s own price impact. The proposed method tracks the posterior distribution of latent states and uses it to solve the trading problem explicitly. A verification theorem supports the solution, while a variation of expectation-maximization provides a way to calibrate the model.

Simulations compare the learning-aware strategy with strategies that ignore latent-state learning, and the paper reports calibration results using Intel stock. These examples illustrate the approach, but the supplied description does not give numerical performance results or detail the model’s assumptions. The findings therefore establish a modeling and solution framework rather than showing how it performs across markets or under live trading conditions.

Key ideas

• Latent factors can drive both jumps and diffusion in prices used for statistical arbitrage.
• A trader’s orders may affect quoted prices and the prices received during execution.
• The strategy learns a posterior distribution over hidden states and uses it to choose trades.
• A verification theorem and an expectation-maximization variant support solution and calibration.
• Simulations compare the approach with strategies that omit latent-state learning.

Tags

# Trading algorithms with learning in latent alpha models


# Trading algorithms with learning in latent alpha models









Alpha signals for statistical arbitrage strategies are often driven by latent factors. This paper analyses how to optimally trade with latent factors that cause prices to jump and diffuse. Moreover, we account for the effect of the trader's actions on quoted prices and the prices they receive from trading. Under fairly general assumptions, we demonstrate how the trader can learn the posterior distribution over the latent states, and explicitly solve the latent optimal trading problem. We provide a verification theorem, and a methodology for calibrating the model by deriving a variation of the expectation-maximization algorithm. To illustrate the efficacy of the optimal strategy, we demonstrate its performance through simulations and compare it to strategies which ignore learning in the latent factors. We also provide calibration results for a particular model using Intel Corporation stock as an example.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.