This paper revisits the exchange rate disconnect puzzle using generative AI to forecast currency returns from economic fundamentals. Employing ChatGPT and DeepSeek, a comprehensive dataset of 544 indicators spanning 1996–2024 is analyzed to extract AI-derived fundamentals, showing significant cross-sectional predictability across major currencies. A simple AI-driven AIFX strategy—going long currencies with strong fundamentals and short those with weak fundamentals—achieves an annualized Sharpe ratio above 0.7, with alpha that remains significant after controlling for standard FX factors. Multiple robustness checks (GPT-4o and DeepSeek-V3, weighted signals, and time-series specs) support the central results. Four look-ahead bias tests indicate the results arise from AI reasoning rather than memorization, and the Taylor-rule framework linking fundamentals to monetary policy emerges as a key mechanism behind the predictability.
以下为对“AI and Exchange Rate Predictability”一篇关于用生成式人工智能(AI)预测汇率的金融研究报告的极其详尽解读与解构分析。分析覆盖元数据、章节要点、图表与表格解读、估值/回报分析、风险点、批判性视角,直到结论性综合。文中所有来自原文的关键结论或推断,均在句末附上溯源标识 [page::页码] 以便追溯。为便于溯源,文中涉及的图表与表格均按原文对应的页码给出对应图片引用。
一、元数据与概览(引言与研究要点)
标题、作者与机构、日期与主题
核心论点与结论(要点性摘要)
二、逐节深度解读(逐章精读与剖析)
4.1 Cross-sectional Trading Strategy(横截面交易策略)
4.2 Panel Regression(面板回归)
4.3 GPT-4o vs DeepSeek-V3(鲁棒性对比)
4.4 Alternative construction of AIFX(替代 AIFX 指标构建)
4.5 Time-Series Strategies(时间序列策略)
6.1 Predictive Power of Different Categories(不同类别基本面的预测力)
6.2 Decomposing Predictability: Positive vs. Negative News(正负新闻的分解)
6.3 A Potential Mechanism(潜在机制)
三、图表与表格深度解读(逐图解读与溯源)
Figure 1. Sharpe Ratios
Figure 2. Cumulative Returns
Figure 3. Portfolio Composition
Figure 4. t-statistics
Figure 5. Sharpe Ratios (GPT-4o vs DeepSeek-V3)
Figure 6. Alternative Specification of the AIFX index
Figure 7. Distribution of Data Releases
Figure 8. Distribution of AI's Guesses
Figure 9、Figure 10. AI Correlation Estimates 的分布
Figure 11–Figure 15 与 Table 1–Table 8(附录及主要表格)
表格要点(Table 1–Table 8)
互联网附录(Internet Appendix)
四、估值分析(在此文中以“估值/回报”的视角来解读)
五、风险因素评估(潜在偏误与局限)
六、批判性视角与细微之处(审慎评估)
七、结论性综合(综合要点、对研究结论的再陈述)
图/表引用与溯源整理(便于后续再现)
总体而言,该研究以系统、严格的实验设计及多维鲁棒性检验,较为充分地证实了 AI 驱动的基本面信号在 FX 市场的预测力,并给出机制性解释(尤以 Taylor Rule 及货币政策的异步反应为核心)。研究对于理论汇率模型的修正、对 AI 在金融研究中的应用方法、以及对实务投资的潜在启发,均提供了值得关注的新视角,尽管在数据源多样性、极端情景下的稳健性以及对 AI 内部推理过程的透明度方面仍有进一步提升的空间。
2608.00761.pdf