MetaTraderForecast A Project that involves training, testing, evaluating and forecasting time series forex data from MetaTrader using Sockets to connect with Python. An RNN Model of various architecture like LSTM, Bidirectional and GRU can be created. Supports CUDA Computation. What does the project do? MQL based Expert Advisors are attached to charts of trading platforms to make RNN based forecasts. All machine learning parameters can be tuned from the input interface of the EA. The model can be trained and exported as a serialized .hdf5 file. Tests are run simultaneously with initial train and the data is split based on Testing Part (in %) parameter. The model can be evaluated on any number of metrics as provided in the input screen of the EA. The model can be loaded to retrain or forecast any number of future bars. Recurrent Neural Network Specs The various parameter that can be opted for tuning the ML model are - RNN Architecture - Long Short Term Memory - Gated Recurrent Units - Bidirectional LSTM - Bidirectional GRU Losses - Mean Sqaure Error - R^2 Score Optimizers - RMSProp (R

⚠ No verified equity curve — no track-record source connected.

No score history is stored yet — only the current score is shown.

⚠ No live verification account connected — ask for proof before buying.

Alerts on changes: coming soon

Prop-firm compatibility not provided.

Data-completeness & trust index (not a profitability rating)