报告简介：

AI for finance has moved from a narrow emphasis onfaster trading to a broader analytical framework for valuation, riskmanagement, portfolio construction, and market monitoring. Traditional financemodels rely on prices, accounting data, macroeconomic variables, and factorreturns. AI extends this toolkit by incorporating high-frequency market data,textual information, web traffic, search intensity, social media sentiment,satellite images, and other alternative data.

This seminar demonstrates how to utilize AI forconstructing investment and trading strategy that predicts returns from marketprices and Wikipedia page-view attention, and an efficient-frontier thatconverts expected returns, volatility, and correlation into portfolio assetallocations. The central purpose is that AI for finance should not beunderstood merely as prediction. It is a full decision pipeline: datacollection, feature engineering, out-of-sample validation, trading-rule design,and risk-reward optimization. The examples also show why careful train/testdesign and portfolio constraints are essential. Predictive signals can lookattractive in-sample but fail out-of-sample; portfolio optimization can improvediversification but can become unstable if inputs are noisy. A rigorousAI-in-finance framework therefore combines alternative data, transparentforecasting models, rolling validation, and mean-variance or constrainedoptimization to evaluate both return and risk.

报告人简介：

Vigdis W Boasson（王超）博士是中密歇根大学金融与法律系的资深金融学教授,她在纽约州立大学布法罗大学获取两个博士学位:金融博士和国际商务博士。她曾在最高层次的著名期刊上发表了许多高质量的学术论文，如Journal of Business 以及许多ISI 索引期刊杂志上。她的研究开创了一个基于跨学科的角度对企业财务分析深度剖析的方法，此方法可以解决众多在传统金融投资学上忽视或未能解决的问题。

Emil Boasson博士是中密歇根大学商学院工商信息系统专业的资深教授，同时也是中密歇根大学商学院工商信息系统研究生学位的教学主任。Emil Boasson教授在信息系统领域有着深厚的学历，他在纽约州立布法罗大学获取地理信息科学博士学位，在SSRN学者网站上是排名百分之十以上的学者。他的研究大多侧重于为区域经济发展的理论，研究地理信息系统和商业信息系统，并做出相关的发展研究框架和模型，这些模型可用以同时分析金融市场摩擦的时间和空间，空间统计，以及多媒体地理可视化空间和时间的商业动态提供丰厚的分析支持。