https://github.com/fziviello/forecast-trading
Trading Forecast Generator
https://github.com/fziviello/forecast-trading
ai data-science forex python stock
Last synced: 5 months ago
JSON representation
Trading Forecast Generator
- Host: GitHub
- URL: https://github.com/fziviello/forecast-trading
- Owner: fziviello
- License: other
- Created: 2024-10-31T10:31:51.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-12-09T11:25:11.000Z (over 1 year ago)
- Last Synced: 2025-04-16T02:47:04.828Z (about 1 year ago)
- Topics: ai, data-science, forex, python, stock
- Language: Python
- Homepage:
- Size: 594 KB
- Stars: 3
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Model Training
The script contains that schedules model training.
## Customizable Parameters
You can change them in the config.py file
- `TIME_MINUTE_REPEAT`: Interval expressed in minutes of the schedule
- `N_REPEAT`: Number of repetitions
## Args Parameters
- `SYMBOL`: The Name of Stock Exchange Symbol separated by comma for multi-currency training (--symbol) *REQUIRED
- `NOTIFY`: If True send the predictions to Telegram Channel
- `SEND_SERVER_SIGNAL`: If True send signal to MT5 Server
## Run
- Start Training : `python3 scripts/training.py --symbols AUDJPY,AUDNZD,AUDCHF,SGDJPY,AUDCAD --notify True --sendSignal True`
# DataSet Generator
This Python script retrieves historical Forex data for a specified symbol using the `yfinance` library
The data can be visualized in a candlestick chart and saved to a CSV file
## Features
- Retrieves historical Forex data for a specified symbol
- Visualizes the data in a candlestick chart if enabled
- Saves the data to a CSV file named `DATASET_{SYMBOL}.csv`
## Customizable Parameters
The script contains several parameters that you can modify to suit your needs:
- `SHOW_PLOT`: If True Show the plot
- `RETRY_LIMIT`: Times to retry in case of error
## Args Parameters
- `SYMBOL`: The Name of Stock Exchange Symbol (--symbol) *REQUIRED
- `INTERVAL`: Dataset Range (--interval)
- `GENERATE_PLOT`: If True Make the plot (--plot)
## Run
- Create DataSet: `python3 scripts/create_dataSet.py --symbol AUDJPY`
# Forecast BOT
This script implements a Long Short-Term Memory (LSTM) neural network for predicting trading signals in Forex markets. It leverages historical price data to generate buy/sell signals, calculate potential profits and losses, and save predictions to a CSV file. The model also includes functionality for plotting results and managing previous model states. Can you see running on https://www.mql5.com/it/users/fziviello87
## Features
- Data Processing: Loads and preprocesses Forex historical data
- Model Training: Trains an LSTM model to predict trading signals based on historical data
- Predictions: Generates buy/sell signals and calculates potential profits and losses
- CSV Management: Saves predictions to a CSV file with options to overwrite existing data
- Visualization: Generates plots to visualize trading signals against historical prices
- Early Stopping: Implements early stopping to prevent overfitting during model training
## Usage
Use the dataset created with the script `create_dataSet`
## Dynamic Parameters
- `REPEAT_TRAINING`: If True restarts model training
## Customizable Parameters
- `USE_SERVER_MT5`: If True required MT5 server ON
- `GENERATE_PLOT`: If True Make the plot
- `SHOW_PLOT`: If True Show the plot
### Business Parameters
You can change them in the config.py file
- `MAX_MARGIN`: Maximum margin on price
- `MIN_MARGIN`: Minimum margin on the price
- `LOT_SIZE`: The number of lots
- `CONTRACT_SIZE`: Standard volume for one Forex lot
- `EXCHANGE_RATE`: The exchange rate for profit calculations
- `FAVORITE_RATE`: Preferred conversion currency (EUR)
- `N_PREDICTIONS`: The maximum number of predictions to generate
- `VALIDATION_THRESHOLD`: Model Validation Threshold
- `INTERVAL_MINUTES`: Dataset interval in minutes
- `RETRY_LIMIT`: Maximum number of retry
- `INTERVAL_DATASET`: Dataset interval in desired format
- `FORECAST_VALIDITY_MINUTES`: Validity of the forecast
- `TIME_MINUTE_REPEAT`: Repeat Training time
- `N_REPEAT`: Number of Repeat Training
- `BOT_TOKEN`: Token API Bot Telagram
- `CHANNEL_TELEGRAM`: Telegram Channel Name with @
- `PARAM_GRID`: Neural Network Parameters
- `units`: The number of neurons in the LSTM layers
- `dropout`: The dropout rate to prevent overfitting
- `epochs`: The number of training epochs
- `batch_size`: The size of the batches used during training
- `learning_rate`: The learning rate to optimize the weights
- `optimizer`: The optimization algorithm (e.g. adam, rmsprop)
For env management you can create the `secret.env` file in the project root where you can create the production keys
## Args Parameters
- `SYMBOL`: The Name of Stock Exchange Symbol (--symbol) *REQUIRED
- `GYM`: If True bypass check status market
- `SEND_SERVER_SIGNAL`: If True send signal to MT5 Server
- `NOTIFY`: If True send the predictions to Telegram Channel
- `FAVORITE_RATE`: Favorite conversion rate (--favoriteRate) (default EUR)
- `INTERVAL_MINUTES`: Interval expressed in minutes to align with the dataset (--interval)
- `GENERATE_PLOT`: If True Make the plot (--plot)
## Run
- Start Forecast: `python3 scripts/forecast_bot.py --symbol AUDJPY --notify True --sendSignal True`
# Calculate Statistics
This script calculates the statistics obtained by the model during its training.
## Customizable Parameters
- `PREFIX_VALIDATION`: Validation file name prefix
## Args Parameters
- `SYMBOL`: The Name of Stock Exchange Symbol (--symbol) *REQUIRED
- `ALL`: Analyze all available validation files (--ALL) *REQUIRED
- `NOTIFY`: If True send the predictions to Telegram Channel
- `GENERATE_PLOT`: If True Make the plot (--plot)
## Run
- Start Get Statistics for ALL symbol availables: `python3 scripts/get_statistics.py --ALL --notify True`
- Start Get Statistics for single symbol: `python3 scripts/get_statistics.py --symbol AUDJPY --notify True`
### Use Venv
- `python3 -m venv .venv`
- `source .venv/bin/activate`
- `.venv/bin/python`
## Requirements
`pip3 install -r requirements.txt`
if you have problems installing ta-lib proceed as follows
- `brew install ta-lib`
- `TA_INCLUDE_PATH=$(brew --prefix ta-lib)/include`
- `TA_LIBRARY_PATH=$(brew --prefix ta-lib)/lib`
- `CFLAGS="-I$TA_INCLUDE_PATH" LDFLAGS="-L$TA_LIBRARY_PATH" pip install ta-lib`
if you use windows:
- Download ta-lib precompiled `https://sourceforge.net/projects/talib-whl` and move into venv dir
- Run `pip install path\to\ta_lib‑0.4.0‑cp310‑cp310‑win_amd64.whl`
To use the `SEND_SERVER_SIGNAL` functionality you can use APIM MT5
- `https://github.com/fziviello/APIM_MT5`
