Using Machine Learning Techniques to Integrate Technical and Fundamental Analysis to Identify Trading Opportunities

(I) - Project Title:

Using Machine Learning Techniques to Integrate Technical and Fundamental Analysis to Identify Trading Opportunities (_Report.pdf)

(II) - People:

Longxiang Dai, longxiad@uci.edu

Yongheng Zhang, yonghenz@uci.edu

(III) - Our Dataset:

Clustering Dataset:

1-Time Series DTW Sample.csv

Records the cluster numbers of the time-series data based on K-means and Dynamic Time Wraping distance

2-Time Series Euclidean Sample.csv

Records the cluster numbers of the time-series data based on K-means and Euclidean distance

3-tweets_textblob.txt

Monthly sentiment indicator of stocks generated from tweets using TextBlob

Training Dataset:

1-Merged Daily Data Sample.csv

The dataset that we use for short-term prediction

2-Merged Quarterly Data Sample.csv

The dataset that we use for long-term prediction

(IV) - Our Notebooks:

Since the runtime requirement for the notebook is around 1 minute, we have deleted all the clustering and training steps. The results of those steps have already generated in files or final models. However, those steps are in the html files that we submited to Canvas.

1-get_data.ipynb

The notebook includes helpful functions to generate quarterly and daily features from SimFin and FRED datasets

2-Data Preprocessing.ipynb

Run the notebook to generate quarterly and daily and save the data into two csv files: merged quarterly data.csv, merged daily data.csv

3-QOQ Prediction.ipynb

Train models to make quarter over quarter (QOQ) predictions. To run the notebook using the sample data, you'll need to change the corresponding file path to the sample data path

4-WOW Prediction.ipynb

Train modesl to make week over week (WOW) predictions. To run the notebook using the sample data, you'll need to change the corresponding file path to the sample data path

5-One Year Movement and Clustering Application.ipynb

Train models to make one year movement predictions and apply clustering results in it. To run the notebook using the sample data, you'll need to change the corresponding file path to the sample data path