Generative AI and Fundamentals-Based Exchange Rate Forecasting
Generative AI analyzes economic data to forecast currency returns, addressing the exchange rate puzzle and eliminating look-ahead bias.
What it examines
This study re-examines the exchange rate disconnect puzzle using generative AI. By harnessing ChatGPT and DeepSeek, it constructs variables from extensive economic data releases to capture currency fundamentals. The approach uses AI-generated signals, rigorous tests, and multiple strategies to determine if AI reasoning, not memorization, drives predictive power.
What it concludes
The research shows that AI-derived signals effectively predict currency returns by linking economic fundamentals with monetary policy. Results imply applications in FX trading, risk management, and economic analysis. Future studies may refine AI methods for broader financial forecasting and practical decision-making in real-time markets.
Evidence objects
The study introduces novel AI technologies like ChatGPT and DeepSeek, generating financial signals over two decades with the innovative $$\text{AI-FX ratio}$$, strength ratio, and weighted AI-FX ratio, surpassing traditional models robustly.
key_findings bullet 2 · key_findings · validation V0
Researchers harness generative AI to reveal striking predictive signals in currency returns, linking economic fundamentals with exchange rate movements, challenging the long-standing disconnect puzzle with robust statistical Sharpe ratios remarkably.
key_findings bullet 1 · key_findings · validation V0
Employing rigorous panel regressions, cross-sectional and time-series trading strategies, the study validates forecasts while warning about model overfitting and data limitations, offering insights into monetary policy effects on currency movements.
key_findings bullet 3 · key_findings · validation V0
Bridging generative AI and FX forecasting, this paper introduces innovative techniques by leveraging LLMs such as $\text{ChatGPT}$ to extract fundamental variables. Its originality lies in addressing longstanding puzzles with modern AI. Meticulous look-ahead bias tests ensure robustness, making the methodology novel, compelling, and a significant incremental extension of existing literature.
key_findings bullet 4 · key_findings · validation V0
Raw abstract and provenance
- … This section begins by constructing cross-sectional trading strategies that use the AI-FX ratio as a signal, evaluated across a range of lookback periods. The performance …
Source row: 972 · abstract type: snippet