Algorithmic Trading Bot

Algorithmic trading strategy that uses machine learning to automate the trade decisions

---

Summary

This project is trading strategy to make trading decisions, made with Jupyter Notebook and that contains:

• An algorithmic trading strategy that uses machine learning to automate the trade decisions.
• Adjust the input parameters to optimize the trading algorithm.
• A second machine learning model with adjusted input parameters.
• An analysis to compare the performance of both models.

---

Installation and Usage

git clone https://github.com/diegoalcocer/algorithmic-trading-bot.git
cd algorithmic-trading-bot/

To start the Jupyther Notebook you could launch jupyter lab:

jupyter lab

📚 resources/

This project uses the following resources required for analysis

• emerging_markets_ohlcv.csv:
  • 1/21/15 9:30 to 1/22/21 15:45
  • Contains the OHLCV dataset
• output1.png
  • Plot: SVM classifier returns vs actual returns
• output2.png
  • Plot: AdaBoost classifier returns vs actual returns

Summary Evaluation

The following is a comparison of performance between the two models created in this project. The models use the following:

1. SVC classifier from SKLearn's support vector machine (SVM) learning method to fit the training data and make predictions based on the testing data
2. AdaBoost classifier from SKLearn's ensemble to fit the data and make predictions

Model 1: SVC

The first model has the following performance statistics:

        precision        recall       f1-score   support

  -1.0        1.00        0.06        0.12        49
   1.0        0.63        1.00        0.77        79

  accuracy                            0.64       128
  macro avg       0.82      0.53      0.44       128
  weighted avg    0.77      0.64      0.52       128

As a result of a high accuracy score, the model was able to predict and generate higher returns than the actual returns:

Model 2: AdaBoost

The second model has the following performance statistics:

              precision    recall     f1-score   support

    -1.0       0.44       0.08        0.13      1804
     1.0       0.56       0.92        0.70      2288

  accuracy                            0.55      4092
  macro avg       0.50      0.50      0.41      4092
  weighted avg    0.51      0.55      0.45      4092

The AdaBoost classifier model had a low accuracy score and for that reason, the AdaBoost returns were in some cases lower than the actual returns

In conclusion the SVC model performs better than the AdaBoost model since it has a higher accuracy score

• SVC accuracy score: 0.64
• AdaBoost accuracy score: 0.55

---

License