Please use this identifier to cite or link to this item:
http://arks.princeton.edu/ark:/88435/dsp01zc77st29r
Title: | Constructing Equity Trading Strategies Based on Macroeconomic Event Analysis |
Authors: | Yin, Andre |
Advisors: | Fan, Jianqing |
Department: | Operations Research and Financial Engineering |
Certificate Program: | Applications of Computing Program Center for Statistics and Machine Learning |
Class Year: | 2022 |
Abstract: | The financial markets enrapture retail and institutional investors, especially with the heightened frenzy during the pandemic. Need we mention Gamestop? Over the past few decades, investors and traders have developed many strategies to beat the market using fundamental and technical analysis. While prior research has used macroeconomic event analysis to predict bond market returns, less attention has been devoted to predicting equity market returns with macro events. To address this gap, in this project, we develop trading strategies that predict US equity index market price changes surrounding macroeconomic news releases. Examples of these releases include the jobs report, unemployment claims, and GDP growth. To build our predictive model, we use inputs such as economic surprise, seasonality of the release, release frequency, and release importance, among other factors. Our output is then predicted percentage price change. For our baseline predictive model, we use ordinary least squares, then develop trading strategies using more nuanced models. Additionally, little if any research has combined technical analysis with macro event analysis in the equity markets. Thus, taking our research one step further, we introduce various technical indicators into our predictive models, then retrain those models and report updated results. We evaluate our strategies through various metrics, including cumulative profit and loss, sharpe ratio, and correlation with benchmarks such as the S&P 500 index. Trained on macro-only factors, our best-performing models beat the benchmark by 3x in cumulative PNL with a sharpe ratio of 0.6 while having near-zero correlation with the benchmark. Trained on combined macro and technical factors, our best-performing models beat the benchmark by almost 6x in cumulative PNL with a sharpe ratio of 0.9 while having near-zero correlation with the benchmark. |
URI: | http://arks.princeton.edu/ark:/88435/dsp01zc77st29r |
Type of Material: | Princeton University Senior Theses |
Language: | en |
Appears in Collections: | Operations Research and Financial Engineering, 2000-2024 |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
YIN-ANDRE-THESIS.pdf | 1.65 MB | Adobe PDF | Request a copy |
Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.