演讲人：刘扬（湖南大学金融与统计学院 助理教授）

时间：2023年12月26日（周二）11:30-12:30

地点：同济大厦A楼505室

同步#腾讯会议：960515625

会议密码：473466

链接：https://meeting.tencent.com/dm/bKRx4hcwKxJi

摘要：

While common machine learning algorithms focus on statistical model fit, we show that genetic programming, GP, is well-suited to maximize an economic objective, the Sharpe ratio of investment portfolios in two contexts. Firstly, in the cross-section of stock returns, we find that in contrast to popular methods such as LASSO and neural network, the spread portfolio of GP can double their performance in the US, and outperform them internationally. While the economic objective plays a role, GP captures nonlinearity in comparison with LASSO, and it requires smaller sample size than neural network. Secondly, we apply GP in the standard efficient portfolio optimization without estimating expected return and covariance matrix. GP again substantially outperforms conventional and other machine learning-based methods by almost doubling the out-of-sample performance.

个人简介：

刘扬，湖南大学金融与统计学院助理教授。清华大学金融学博士，中央财经大学经济学学士（金融工程），美国圣路易斯华盛顿大学奥林商学院访问博士生。主要研究领域为实证资产定价、机器学习与行为金融。论文发表（接收）于《Review of Asset Pricing Studies》等期刊，多篇工作论文曾入选美国金融学年会（AFA），中国金融国际年会（CICF）、美国财务管理年会（FMA）等高水平会议。