Algorithmic Trading System with Machine Learning
- An Algorithmic Trading System connecting to Robinhood to execute various algo trading strategies
- Stocks Recommendation System based on Machine Learning Models
These instructions will enable you to run an algo trading system on your local machine
You need a robinhood account with a two factor code setup using your phone number. At the point of executing AlgoRobinhood, you will put the account id, password and the two-factor codes in terminal to login
$ python AutoTrade.py
Enter your account id: [email protected]
Do you want to run recommendation code (y/n): y
Enter your password: *****
Please Enter SMS Code: ******
2019-09-30 23:22:16,785 __main__ Successfully logged in account
2019-09-30 23:22:16,785 __main__ Running stocks recommendation for today...
2019-09-30 23:28:54,145 __main__ The code is outside execution period.
2019-09-30 23:28:54,202 __main__ Successfully logged out account.
A step by step series of examples will be added
Example to be added
Simply git clone this codebase and execute the below command. After successfully login in, a log file will be generated under /log, with detailed information like recommendations made from a tensorflow model predction, as well as buy/sell operations being executed if any
python AutoTrade.py
Example of a log file
2019-09-30 23:22:04,500 __main__ Auto trading start at 2019-09-30 23:22:04
2019-09-30 23:22:04,500 __main__ The code is outside execution period.
2019-09-30 23:22:16,785 __main__ Successfully logged in account
2019-09-30 23:22:16,785 __main__ Running stocks recommendation for today...
2019-09-30 23:22:23,236 recommendation_system.recommendation Symbol SNAP is using LSTM training model and doing forecast...
2019-09-30 23:22:27,877 recommendation_system.recommendation Symbol SNAP is trained and validated with accuracy 60.27%, forecasted to price up by 5.0% over the 5 days with predicted probability of 24.43%
...
2019-09-30 23:28:40,817 recommendation_system.recommendation Symbol COF is using LSTM training model and doing forecast...
2019-09-30 23:28:54,144 recommendation_system.recommendation Symbol COF is trained and validated with accuracy 90.41%, forecasted to price up by 5.0% over the 5 days with predicted probability of 4.17%
2019-09-30 23:28:54,145 recommendation_system.recommendation Today's top 5 recommended stocks are:
2019-09-30 23:28:54,145 recommendation_system.recommendation Symbol GE: Rating 53.47% - Model Accuracy 87.67%
2019-09-30 23:28:54,145 recommendation_system.recommendation Symbol NFLX: Rating 49.46% - Model Accuracy 86.3%
2019-09-30 23:28:54,145 recommendation_system.recommendation Symbol FIT: Rating 38.27% - Model Accuracy 82.19%
2019-09-30 23:28:54,145 recommendation_system.recommendation Symbol AMZN: Rating 29.51% - Model Accuracy 91.78%
2019-09-30 23:28:54,145 recommendation_system.recommendation Symbol PVTL: Rating 27.16% - Model Accuracy 80.82%
2019-09-30 23:28:54,145 __main__ The code is outside execution period.
2019-09-30 23:28:54,202 __main__ Successfully logged out account.
Example will be added about how to deploy this on a user-friendly web based system
Example to be added
- Python 3.6 - Codebase development and execution
- Django - The web framework development
- Version 1.0
This project is licensed under the MIT License - see the LICENSE.md file for details
- Build Github page with more information in README.md - done on 20190929;
- Optimize the data pipeline for modeling, and do codebase refactoring; - done on 20190930
- add modeling class, which can be extented to RF and XGBoost; - done on 20190930
- Extend machine leaning models by adding Random Forest to do forecast; - done on 20191001
- Extend machine leaning models by adding XGboost to do forecast; - in progress
- Create Python package;
- Build Web-based UI;
- Build NLP on financial news, take it as additional features in ML models to enhance model performance;
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent