Abstract
The public release of ChatGPT represents a significant milestone in generative AI technology, enabling the autonomous generation of content based on pre-training. This breakthrough presents new opportunities for advancements in the field of portfolio selection. This paper aims to introduce a comprehensive portfolio selection method by applying ChatGPT for stock selection and combining it with optimization algorithms to jointly optimize portfolio selection. Compared to randomly selected stocks, the portfolios optimized using ChatGPT-selected stocks and solved with the egret swarm optimization algorithm (ESOA) demonstrate higher diversification and lower volatility, leading to superior portfolio optimization results. Additionally, to validate ESOA’s superiority, its performance is compared against genetic algorithm (GA) and particle swarm optimization (PSO) on five metrics: risk, expected return, Sharpe ratio, objective value, and penalty term. Under equivalent experimental setting, ESOA exhibits a better ability to balance the relationship between risk and return.
| Original language | English |
|---|---|
| Article number | e1441 |
| Pages (from-to) | 6163-6179 |
| Journal | Neural Computing and Applications |
| Volume | 37 |
| DOIs | |
| Publication status | Published - Mar 2025 |
| MoE publication type | A1 Journal article-refereed |
Funding
This work was supported in part by the National Natural Science Foundation of China under Grants 62066015 and 62006095.
Keywords
- ChatGPT
- Egret swarm optimization algorithm
- Portfolio optimization
- Transaction cost