Paper: arXiv 2409.11540
Authors: Shuaiyu Chen, T. Clifton Green, Huseyin Gulen, Dexin Zhou
Abstract
We examine how large language models (LLMs) interpret historical stock returns and compare their forecasts with estimates from a crowd-sourced platform for ranking stocks. While stock returns exhibit short-term reversals, LLM forecasts over-extrapolate, placing excessive weight on recent performance similar to humans. LLM forecasts appear optimistic relative to historical and future realized returns. When prompted for 80% confidence interval predictions, LLM responses are better calibrated than survey evidence but are pessimistic about outliers, leading to skewed forecast distributions. The findings suggest LLMs manifest common behavioral biases when forecasting expected returns but are better at gauging risks than humans.
Complexity vs Empirical Score
- Math Complexity: 3.0/10
- Empirical Rigor: 6.0/10 (capped from 8.0: no out-of-sample test; see penalty flags)
- Quadrant: Street Traders — practical and empirical, lighter on theory
Why this score: This paper presents a novel behavioral analysis evaluating whether LLMs replicate human cognitive biases when forecasting stock returns using numeric and visual inputs. It features strong empirical rigor through extensive regression analyses against historical returns and survey data, though mathematical derivations are minimal.
Research Flowchart
flowchart TD
A[Research Goal: Assess LLM Biases & Calibration in Stock Return Forecasts] --> B[Data Inputs: Historical Stock Returns & Crowd-Sourced Ranking Estimates]
B --> C[Methodology: Prompt LLMs for Expected Returns & 80% Confidence Intervals]
C --> D[Analysis: Compare LLM Predictions vs. Historical Reversals & Human Survey Data]
D --> E[Outcome 1: Return Forecasts Over-Extrapolate Recent Trends & Show Optimism]
D --> F[Outcome 2: Risk Forecasts Show Better Calibration Than Humans but Skew Outliers]