这里的目标则是利用新闻来预测更广泛的市场趋势。从这一角度来看，新闻一点也不稀疏。每天交易 日都会产生并记录数以万计的文章。然而，当我们从关注具体公司新闻转向对美国市场整体进行聚合 时，一个新的问题出现了。 我们发现，稀疏性的挑战被序列相关性所取代。 正如图1所示，美国每日 整体新闻情绪在相对较长的滞后期内都表现出高度的自相关性。 图1：美国每日整体新闻情绪的滞后自相关 每日整体数据通过对所有与美国上市股票相关的文章的情绪评分进行求和来计算。 这似乎对我们既 定的目标提出了真正的挑战——利用整体新闻信号来预测美国市场的剧烈变化。要实现这一点，我 们需要一个尽可能接近"无记忆"的指标。这样的指标才更有可能及时感知市场的转折。 Amit Das LSEG 新闻数据与API业务 在过去几十年里，美国展现出极强的增长韧性，期间伴随少数几次急剧的熊市。然而，总体趋势保持 积极，熊市后的复苏通常也很迅速。我们可以用S&P500来说明这一点。在COVID-19疫情初期的冲 击下，S&P500市值下跌约30%。然而，仅一个月后它便开始稳步回升。在六个月内，它完全收复了 失地。到年底时，指数较疫情初始已上涨超过10%。 在本篇洞 ...

Core Viewpoint - The emergence of new AI predictive models is transforming the way investors approach global macroeconomic forecasting, providing more precise methods for achieving excess returns [2][5]. Group 1: Investment Strategies and Data Management - Financial services are investing unprecedented resources in data management and AI to build more accurate macroeconomic forecasting models, as asset allocation decisions significantly impact portfolio performance [3][5]. - The complexity of global markets and the vast amount of data involved have historically made reliable macroeconomic forecasting a challenging goal for investors [3][5]. - The use of clean and accurate training data, along with techniques like Teacher forcing, enhances the accuracy of AI predictions by ensuring optimal information is used at each step of the training process [3][5]. Group 2: AI and Predictive Models - Many large buy-side and sell-side firms are increasing their investments in data management and AI resources, although traditional methods involving human resources for data handling are costly and difficult to scale [4][5]. - LSEG's global macroeconomic forecasting integrates advanced technology and AI reasoning, providing a range of indicators that help traders and asset managers make informed decisions [7]. - The predictive service can be delivered via APIs, allowing companies to trade based on forecasts before official government data releases, such as the Consumer Price Index (CPI) [7][12]. Group 3: Performance and Insights - LSEG's scalable, centralized data warehouse and automated framework support cutting-edge AI reasoning and automation, enhancing performance when data management and AI work in tandem [6]. - The integration of various data sources, including machine-readable news and point-in-time data, allows for the discovery of complex relationships and the generation of forward-looking economic indicators [5][12]. - Users can access predictive results and related data through a universal platform, enabling them to overlay this information onto existing portfolios for better identification of inter-data relationships [12].