Bitcoin price analysis

Getting started

• Initialise

python3 -m venv .venv

---

• Activate

source .venv/bin/activate

---

• Install dependencies

pip3 install -r requirements.txt

---

• Generate model

python3 gen.py

NOTE: this wil fail on the first run, fix the following manually.

• the BTCUSD_1H.csv chart needs to have the url removed from the first line of the csv.
• the BTC_sentiment.csv will be wrapped in json and the columns need reordering (date is first column and it has it as the last)

---

• Run data collection

python3 collect.py

To run collect.py in the background as a persistent process (so it doesn't stop when you close your terminal):

nohup ./.venv/bin/python3 collect.py &

This command uses nohup to prevent the process from being terminated when the controlling terminal is closed, and & to run it in the background. Output will be redirected to nohup.out by default.

This script connects to the Binance WebSocket stream to collect real-time BTCUSDT data and calculate features, logging everything to BTCUSD_trading.csv.

To stop the collect.py process if it was run with nohup:

Stopping the collect process:

pgrep -f "python3 collect.py" | xargs kill

This command finds the PID of the running collect.py script and kills it in one step.

---

• Run trading bot

python3 trade.py

This script reads the BTCUSD_trading.csv file, loads the pre-trained model, makes price predictions, and logs trade actions to BTCUSD_predictions_trades.csv.

View Trade Log

To view the trade.html page, which displays the BTCUSD_predictions_trades.csv data, you need to run a local web server. Navigate to the project's root directory in your terminal and execute the following command:

python3 -m http.server 4242

After starting the server, open your web browser and go to http://localhost:8000/trade.html. The page will automatically update with the latest data from BTCUSD_predictions_trades.csv every 5 seconds.

Running Tests

To run the tests for this project, use the following command:

python3 -m unittest test_features.py

Research

[] Gradient boosting machines (XGBoost, LightGBM or CatBoost), these models are good for regression tasks, predicting future prices, or understanding feature importance.

[] Deep learning (LSTM or GRUs), specialised for sequential data. Great for capturing long-term dependencies and complex patterns in price data. Particularly useful with large datasets.

[] Recurrent neural nets (temporal data, RNNs and variations). Designed for handling time series data and sequential dependencies effectively.

[] GANS (Generative Adversarial Networks) can be used to simulate and generate synthetic price data for testing trading strategies and risk analysis.

[] Bayesian machine learning - can help in probabalistic modeling, providing a range of possible outcomes and quantifying uncertainty in predictions. Especially good for volatile assets like Bitcoin.

Resources

• machine-learning-for-trading.