LLM Agents and Strategy Expectations in Heterogeneous Markets

Summary

This study asks whether a generative agent powered by a large language model can reproduce changes in the mix of fundamentalist and trend-following behavior described by heterogeneous agent models. The agent uses current information to assign probabilities to adopting either strategy, and those probabilities are compared with findings from the literature for the S&P 500 from 1990 to 2020.

The reported comparison suggests that the agent’s expectations align with patterns documented in prior heterogeneous agent research. The authors also examine an artificial market to investigate the agent’s decision process. That analysis retains variation in expectations but finds a systematic tilt toward fundamentalist behavior. The abstract does not specify the model prompts, validation design, quantitative fit, or robustness checks, so the evidence supports a comparison of expectations rather than a demonstrated trading advantage.

Key ideas

• The generative agent assigns probabilities to fundamentalist and trend-following strategies using current information.
• Its strategy probabilities are compared with heterogeneous agent findings for the S&P 500 from 1990 to 2020.
• The reported expectations align with patterns in the prior literature.
• Artificial market analysis shows heterogeneous expectations alongside a systematic fundamentalist tilt.

Tags

# Generative Agents and Expectations: Do LLMs Align with Heterogeneous Agent Models?


# Generative Agents and Expectations: Do LLMs Align with Heterogeneous Agent Models?









Results in the Heterogeneous Agent Model (HAM) literature determine the proportion of fundamentalists and trend followers in the financial market. This proportion varies according to the periods analyzed. In this paper, we use a large language model (LLM) to construct a generative agent (GA) that determines the probability of adopting one of the two strategies based on current information. The probabilities of strategy adoption are compared with those in the HAM literature for the S\&P 500 index between 1990 and 2020. Our findings suggest that the resulting artificial intelligence (AI) expectations align with those reported in the HAM literature. At the same time, extending the analysis to artificial market data helps us to filter the decision-making process of the AI agent. In the artificial market, results confirm the heterogeneity in expectations but reveal systematic asymmetry toward the fundamentalist behavior.

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.