UROP Proceedings 2020-21

School of Business and Management Department of Accounting 189 Investment Analysis with Machine Learning Supervisor: YOU Haifeng / ACCT Student: WANG Jianghan / FINA Course : UROP1100, Spring We replicate Bartram and Grinblatt (2018) using python and make some improvements. In our research, we implement an agnostic fundamental analysis with minimal data snooping. Rather than applying highly stylized models (e.g., DCF valuation) to estimate fair value, we use techniques such as least squares to estimate peer-implied fair values from the market values of replicating portfolios with the same accounting statements as the company being valued. Consequently, we construct a mispricing signal which represents the divergence of a company’s estimated peer-implied fair value from its market value. Fama-MacBeth cross-sectional regressions are used to test if mispricing signal itself can predict future returns. As an improvement, we use LASSO to select the most influencing factors to construct peer-implied fair value. Information Collection from Annual Reports of Public Companies Supervisor: ZHENG Yue / ACCT Student: IP Ching Stephanie / FINA Course: UROP1100, Fall In this project, we attempt to collect product-related information from the annual reports of US high-tech public companies and get familiar with the online resource of filing that the US public companies file with the Securities and Exchange Commission (SEC). We are given an Excel file listing the 500 US companies’ 10k filings and we have to search the original 10-k file on the SECEdgar Website based on the CIK number and filing date. In the 10-k files, we extract paragraphs or sections describing the companies’ products and services, particularly from Item 1 (Business) into 500 txt format documents. Information Collection from Annual Reports of Public Companies Supervisor: ZHENG Yue / ACCT Student: KAN Ho Yin / FINA Course: UROP1100, Fall In this project, collection of information of US companies is required. During the information collection process, browsing the SEC Edgar Website, searching the 10-K annual reports of US high-tech public companies and reading Item 1 (Business) are needed. In a bid to gain a better understanding of the product and services these high-tech public companies offered and their developments of product space over years, sections or paragraphs describing the companies’ products and services are extracted and compiled into txt documents for reviews. At the end of this project, a total of 500 documents of product-related descriptions from Item 1 of 10-k filings have been made.

RkJQdWJsaXNoZXIy NDk5Njg=