Researched Models

This page presents evaluations of our market machine-learning models on unseen historical data. Each model was tested on price data it had not encountered during training to approximate real-world usage. Results are presented for research and analytical purposes only.

How to use this page

You should
  • View model descriptions and outputs.
  • See historical evaluation and saved artifacts.
  • Compare each model against.
You should not
  • Expect guaranteed performance.
  • Assume a single best model always wins.

LSTM (Long Short-Term Memory)

Deep Learning

A recurrent neural network baseline trained on sequential OHLCV data. Saved artifacts are reused for forward prediction.

How It Works

  • Builds sliding windows of OHLCV inputs over the chosen lookback period
  • Standardizes features using training data
  • Trains a single LSTM layer with a small dense regression head
  • Stores model + scalers for future prediction runs

Testing & Validation

Time Period
1 month
Dataset
Crypto + equities (varies by symbol)
Training Samples
Varies by window and timeframe

Outputs & Limits

  • RMSE, MAE, MAPE metrics per run
  • Saved model artifacts for reuse
  • Forecasts 10-20 future candles

Visualizations

LSTM (Long Short-Term Memory) visualization
LSTM (Long Short-Term Memory) visualization

Model Configuration

Checking access...

Note: All models have been tested on out-of-sample data to ensure realistic performance metrics. Past performance does not guarantee future results. These models are implementations based on academic research and have been adapted for our trading platform.