School of Business and Management Department of Accounting 188 Investment Analysis with Machine Learning Supervisor: YOU Haifeng / ACCT Student: SHIH Jackson / QFIN SO Man Kit / QFIN TAM Shing Hang Boris / COGBM Course: UROP1100, Fall UROP1100, Fall UROP1100, Fall Analyzing investment decisions with machine learning techniques is getting increasingly popular. In this report, we explored some simple machine learning models which make use of technical indicators to try to predict future market return in the short term. Our testing and training data are taken from over 1000 A-share stocks in around a 4 year timespan (from the beginning of 2000 to December 2003). We have chosen 8 useful technical indicators by looking at their t-statistics and used these indicators to fit with the Decision Tree and Logistic Regression Model. The 2 models take the indicators’ value and output binary signal on buying or selling. Investment Analysis with Machine Learning Supervisor: YOU Haifeng / ACCT Student: SUN Beini / RMBI ZHANG Yichen / RMBI YIU Kwong Chun / QFIN Course: UROP1100, Fall UROP1100, Fall UROP2100, Spring UROP1100, Fall This paper has referred to Bartram and Grinblatt (2018) and tried to estimate peer-implied fair values of stocks from their accounting items and the market value of other stocks with different machine learning techniques (both linear regression and non-linear regression such as decision tree, random forest, support vector regression, Gradient boosted regression trees). The research has also manipulated input of the data by determining the significance level of the variables using VIF, correlation parameters etc. and filter out useless factors for prediction. Assuming no short-selling constraint or trading costs, divergence between the stock’s market value and peer-implied value represents mispricing and convergence trading opportunities with over 10% annual return. Investment Analysis with Machine Learning Supervisor: YOU Haifeng / ACCT Student: LIU Zhetian / MAEC CHENG Hongyu / FINA YANG Wenting / COGBM Course : UROP1100, Spring UROP1100, Spring UROP1100, Spring Our UROP topic is “Investment Analysis with Machine Learning” and our group job is to do technical analysis. We first did some literature review on papers, and then tried to replicate the paper (Han, Zhou & Zhu, 2016). We notice that there remains differences between a variety of stock species, and specifically, Shenzhen GEM performs better than other stock sectors. Furthermore, some evidence implies that the performance of trend factor is not severely affected when prohibiting short-selling, as a suggestion that it could be applied in A share market where short-selling is relatively difficult. This report will include the whole process of research and concrete analysis of the results.