|
--- |
|
license: apache-2.0 |
|
--- |
|
|
|
The project on GitHub : |
|
https://github.com/reuniware/CryptoForex-Trader-Framework/tree/main/CCXT_ICHIMOKU/julie_scanner |
|
|
|
|
|
### How to Use `bluewenne8.py` |
|
|
|
1. **Install Dependencies**: |
|
Ensure you have the required libraries installed: |
|
```sh |
|
pip install ccxt pandas scikit-learn joblib argparse pytz |
|
``` |
|
|
|
2. **Script Overview**: |
|
`bluewenne8.py` performs cryptocurrency data analysis, trains a machine learning model, and makes predictions. |
|
|
|
### Command-Line Usage |
|
|
|
You run the script from the command line with various arguments to control its behavior: |
|
|
|
#### 1. **Fetch Data and Analyze Symbols** |
|
This command will fetch data for symbols, analyze the greatest candles, and save the results: |
|
|
|
```sh |
|
python bluewenne8.py --timeframe 1d |
|
``` |
|
|
|
- **`--timeframe`**: Required. Defines the candlestick timeframe, e.g., '1d' for daily candles, '1h' for hourly candles. |
|
|
|
#### 2. **Train the Model** |
|
If you want to train a model on historical data, use the following command: |
|
|
|
```sh |
|
python bluewenne8.py --timeframe 1d --train |
|
``` |
|
|
|
- **`--train`**: Optional. If included, the script will train a machine learning model using existing historical data. |
|
|
|
#### 3. **Use Existing Model to Make Predictions** |
|
To make predictions using an existing model: |
|
|
|
```sh |
|
python bluewenne8.py --timeframe 1d --use-existing |
|
``` |
|
|
|
- **`--use-existing`**: Optional. If included, the script will use the pre-trained model to make predictions based on existing historical data. |
|
|
|
### Detailed Steps for Each Mode |
|
|
|
#### A. **Fetch Data and Analyze Symbols** |
|
|
|
1. **Fetch Markets**: The script retrieves a list of available markets from the Binance exchange. |
|
2. **Fetch OHLCV Data**: Collects candlestick data for each symbol based on the provided timeframe. |
|
3. **Save Data**: Saves the fetched historical data to CSV files in the `downloaded_history` directory. |
|
4. **Analyze Symbols**: Identifies and logs the greatest candle for each symbol, including current prices. |
|
|
|
#### B. **Train the Model** |
|
|
|
1. **Load Historical Data**: Reads data from CSV files in the `downloaded_history` directory. |
|
2. **Preprocess Data**: Prepares data by formatting timestamps, setting indices, and splitting features and target variables. |
|
3. **Train Model**: Uses a RandomForestRegressor to train on the historical data. |
|
4. **Save Model**: Saves the trained model and scaler to disk (`model.pkl` and `scaler.pkl`). |
|
|
|
#### C. **Use Existing Model to Make Predictions** |
|
|
|
1. **Load Model and Data**: Loads the saved model and scaler, and reads historical data. |
|
2. **Predict Next Candle**: Uses the trained model to predict future price movements based on the latest data. |
|
3. **Save Predictions**: Writes predictions to a results file. |
|
|
|
### File Structure and Directories |
|
|
|
- **`downloaded_history/`**: Directory where historical data CSV files are saved. |
|
- **`scan_results_bluewenne8/`**: Directory where results and prediction files are saved. Created based on the script name. |
|
- **Model Files**: `model.pkl` and `scaler.pkl` are saved in the script's working directory when training. |
|
|
|
### Example Use Case |
|
|
|
1. **Fetch and Analyze Data**: |
|
```sh |
|
python bluewenne8.py --timeframe 1d |
|
``` |
|
This will fetch data for all available USDT pairs, analyze it, and save results. |
|
|
|
2. **Train Model**: |
|
```sh |
|
python bluewenne8.py --timeframe 1d --train |
|
``` |
|
This will train the model on data from files matching the filter `BTC_USDT`. |
|
|
|
3. **Predict with Existing Model**: |
|
```sh |
|
python bluewenne8.py --timeframe 1d --use-existing |
|
``` |
|
This uses the pre-trained model to make predictions based on the latest historical data. |
|
|
|
Feel free to adjust the timeframe and filters as needed for your specific analysis or training tasks. |