主办单位

太阳集团tyc151

主讲嘉宾

李仲飞

南方科技大学金融系讲席教授

会议信息

时间

2022年11月9日（星期三）

北京时间13:30-15:00

主持人

太阳集团tyc151 潘伟

线上地点

腾讯会议号：771-476-277

主讲人简介

李仲飞，男，南方科技大学金融系讲席教授，国务院学位委员会学科评议组成员，长江学者特聘教授，国家杰青，全国模范教师，国务院政府特殊津贴专家，全国百篇优秀博士学位论文获得者，新中国成立70周年观礼嘉宾，Elsevier 中国高被引学者，全球前2%顶尖科学家，国家自科创新研究群体项目主持人，国家社科基金学科评审组专家，中国系统工程学会副理事长，中国优选法统筹法与经济数学研究会副理事长及其量化金融与保险分会理事长，十多个国内外期刊的领域主编、副主编或编委。历任中山大学社科处处长，管理学院执行院长，创业学院院长。研究领域包括绿色金融与碳经济，金融科技与数字金融，金融工程与风险管理，金融市场与投资。

讲座题目

Prediction of Mutual Funds’ Returns Based on Machine-learning Methods基于机器学习方法的中国公募基金收益预测

讲座摘要

With the rapid development of artificial intelligence and information technology, human psychology has many shortcomings compared with machine algorithm when making decisions and judgments, and is prone to be affected by emotions. Through the management modeling of existing knowledge, in the psychological perspective of knowledge management, compare human psychology and machine learning algorithms. Machine learning algorithm is used to predict the yield of the fund. For fund investors, one of the most important issues is whether the return of mutual funds is predictable. This study uses eight machine learning methods to predict mutual fund returns in China. We combine the rich features into four categories: funds, managers, stocks and macroeconomics. Some of our models generate very high R2 outside the sample, which can reach 66%. As far as we know, this is the highest achievement of journal papers that predict the performance of mutual funds. Our scale weighted long portfolio can achieve an annual return of 49.7%, an annual alpha of 40.3% and a sharp ratio of 1.07. We also compared the predictive power of characteristics and discussed the nonlinear relationship between characteristics and future returns of funds. Our research provides insights for fund investors and financial analysts.

图文/供稿：文丽

设计 责编：曹乐涵 马文林

审核：潘伟 宋枫