Prediction in Stock Trading J S Roger Jang

Prediction in Stock Trading J. -S. Roger Jang (張智星) MIR Lab, CSIE Dept. National Taiwan University jang@mirlab. org, http: //mirlab. org/jang 2021/2/24

Types of Predictions in Stock Trading Up/down prediction Classification � Use historical data as GT (groundtruth) Price prediction Regression � Use historical data as GT Best timings for “buy” and “sell” � How to set up GT? Heuristic methods Optimal methods � Use historical data for prediction and backtest Historical data from 技術面、消息面、基本面… 2/13

Prediction of GT Derived by Heuristics 4/13

GT for “Buy” and “Sell”: Optimal Approach Problem definition � Given the price curve over a long period of time, find “buy” and “sell” points (where you can look into the future) such that the overall profit is maximized. Assumptions � 每一次都全買或全賣 � 根據開盤價格來買或賣，一定會成交 � 必須計算交易費 Approaches � Optimal search: Dynamic programming � Heuristic search: Genetic algorithms, simulated annealing, randomized search? 5/13

Prediction of GT Derived by Optimal Methods 6/13

References and Resources Python � 7 methods to perform Time Series forecasting � A comprehensive beginner’s guide to create a Time Series Forecast � 11 Classical Time Series Forecasting Methods in Python �… MATLAB � Time Series Forecasting Using Deep Learning � Time Series Prediction and Forecasting for Prognosis � Time Series Analysis �… Search keywords: time series prediction python/matlab 7/13