K Score-Dynamic Alpha Signal By Machine Learning - US Equities

Prediction power from weeks to 2-3 months in the future, NOT a quantitative relation to explain the past performance.

Evaluated over 1000 factors, and defined over 200 factors in house as input to train proprietary machine learning model.

Simplified rating that incorporated dynamic factor shifting and rotation.

Predictive equity rating from 1-9 for few weeks to 2-3 months in the future with incorporation of dynamic factor shifting. The development of K Scores takes in fundamental information, pricing and trading volumes, technical signals, and alternative data such as sentiment, and apply the best-of-breed proprietary learning algorithms, and uses methods such as regression, classification, model selection, ranking algorithms, deep learning and more to rank stocks. Our intelligent system identifies the intrinsic relationships, uncovers the dependency structure and calculates predictive analytics for future performance.

本模型的预测周期覆盖数周至2-3个月的未来时段，且并非用于解释过往业绩的定量关联。
该模型基于超1000个因子开展评估，并以自研定义的200余个因子作为输入，用于训练专属机器学习模型。
其为融合因子动态切换与轮换机制的简化评级体系。
该体系可针对未来数周至2-3个月的股价表现输出1-9分的预测性股票评级，且融入因子动态切换机制。K评分体系的研发纳入基本面信息、定价与成交量数据、技术指标信号，以及情绪等另类数据；其采用顶尖专属学习算法，并结合回归、分类、模型选择、排序算法、深度学习等多种方法对个股进行评级排序。本智能系统可识别内在关联、挖掘依赖结构，并计算用于预测未来业绩的分析指标。

搜集汇总

背景与挑战

背景概述

该数据集为美国股票提供基于机器学习的K评分动态阿尔法信号，旨在预测未来几周到2-3个月的股票表现，评级范围为1-9分。它通过评估超过1000个因素并定义200多个内部因素作为输入，利用专有机器学习算法（如回归、分类和深度学习）进行训练，同时整合动态因子轮换以优化预测。系统结合基本面、价格、交易量、技术信号及情感等替代数据，揭示内在关系并计算未来绩效的预测分析。