We set out to find whether value investing strategy such as Joel Greenblatt's earning yields and Piotroski's F-score can actually beat the Thai stock market. The fundamental data are obtained from Morningstar while the benchmark SET TRI is obtained from Stock Exchange of Thailand. We perform support vector machine, boosted logistic regression and random forest classification based on the data, as well as a backtest to verify the models. Most our models perform better than a coin toss with random forest achieving the highest accuracy and sensitivity of 85.13% and 79.25% respectively (20.08% annual return).
See the analysis here. Codes in index.Rmd
The codebook demonstrates how fundamental data on Thai stocks from the Morningstar and SET TRI from Stock Exchange of Thailand are transformed into the tidy format used in this analysis.
The tidy Thai stock dataset includes annual fundamental data of 463 Thai stocks in udf.csv and annual return of SET TRI in set_return.csv from 2002-2015. Monetary unit is THB.
Contains JSON format of all stocks in Thai market obtained from Monringstar.
Annual returns of SET TRI
List of symbols in Thai market
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent