时态数据满⾜LLM -Explainable Financial Time Series ForecastingTemporal Data Meets LLM -Explainable Financ_temporal data meets llm–explainable financial time

本⽂提出了⼀项关于利⽤⼤型语⾔模型（LLM）出⾊的知识和推理能⼒进⾏ 可解释的⾦融时间序列预测的新颖研究。机器学习模型在⾦融时间序列中的 应⽤⾯临着⼀些挑战，包括跨序列推理和推理的难度，融合历史新闻、⾦融知 识图谱等多模态信号的障碍，以及起诉解释和解释模型结果。在本⽂中，我们 利⽤公开可访问的历史股价数据、公司元数据和历史经济/⾦融新闻，重点关 注纳斯达克 100 股票。我们进⾏实验来说明法学硕⼠在为上述挑战提供统⼀ 解决⽅案⽅⾯的潜⼒。我们的实验包括尝试使⽤ GPT-4 进⾏零样本/少样本 推理，以及使⽤公共 LLM 模型 Open LLaMA 进⾏基于指令的微调。我们证明 我们的⽅法优于⼀些基线，包括⼴泛应⽤的经典 ARMA-GARCH 模型和梯度 提升树模型。通过性能⽐较结果和⼀些例⼦，我们发现法学硕⼠可以通过对⽂ 本新闻和价格时间序列中的信息进⾏推理并提取⻅解、利⽤跨序列信息以及 利⽤嵌⼊的固有知识来做出深思熟虑的决定在法学硕⼠内。此外，我们还表 明，公开可⽤的 LLM（例如 Open-LLaMA）经过微调后，可以理解⽣成可解 释预测的指令并实现合理的性能，尽管与 GPT-4 相⽐相对较差。

过去⼏年，机器学习 (ML) 和⼈⼯智能 (AI) 技术的快速发展为各个领域带来 了众多机遇和挑战，包括⾦融市场领域 [4,32,49]。特别是，⾦融时间序列预测 任务是战略决策和政策制定的关键要素，它⻅证了从统计/计量经济时间序列 技术[2,7,21,46]到机器学习的重⼤技术创新。学习技术[31,33,68]，深度学习 [13,28,34,35,52]。尽管取得了这些进步，但在⾦融领域部署 ML/AI 模型仍 存在⼀些固有的挑战。⼀个挑战在于跨序列推理和推理领域，这是理解时间模式和做出准确预测 的重要⽅⾯。当前的⽅法包括时间序列相关分析[8,11,20,48]和聚类[1,3,50]。 最近，深度学习被⽤来学习时间序列之间复杂的潜在依赖关系[25,40,42,55]。

ABSTRACT

This paper presents a novel study on harnessing Large Language Models’ (LLMs) outstanding knowledge and reasoning abilities for explainable financial time series forecasting. The application of machine learning models to financial time series comes with several challenges, including the difficulty in cross-sequence reasoning and inference, the hurdle of incorporating multi-modal signals from historical news, financial knowledge graphs, etc., and the issue of interpreting and explaining the model results. In this paper, we focus on NASDAQ-100 stocks, making use of publicly accessible historical stock price data, company metadata, and historical economic/financial news. We conduct experiments to illustrate the potential of LLMs in offering a unified solution to the aforementioned challenges. Our experiments include trying zero-shot/fewshot inference with GPT-4 and instruction-based fine-tuning with a public LLM model Open LLaMA. We demonstrate our approach outperforms a few baselines, including the widely applied classic ARMA-GARCH model and a gradient-boosting tree model. Through the performance comparison results and a few examples, we find LLMs can make a well-thought decision by reasoning over information from both textual news and price time series and extracting insights, leveraging cross-sequence information, and utilizing the inherent knowledge embedded within the LLM. Additionally, we show that a publicly available LLM such as Open-LLaMA, after fine-tuning, can comprehend the instruction to generate explainable forecasts and achieve reasonable performance, albeit relatively inferior in comparison to GPT-4.