Question 1

How to reduce lag in LSTM trading signals for crypto?

Accepted Answer

The README suggests retraining the network with better features or exploring reinforcement learning. Optimizing the model architecture or adding more timely technical indicators might also help improve response speed.

Question 2

LSTM or GRU for Bitcoin price prediction?

Accepted Answer

This project uses LSTM-RNN, but GRUs could be experimented with for potentially faster training and similar performance. The choice depends on balancing complexity with accuracy for your specific dataset and trading goals.

Question 3

How to add more technical indicators to this model?

Accepted Answer

You can extend the feature engineering by integrating additional indicators from libraries like TA-Lib, following the existing code structure for MACD and RSI in the features section of the README.

Question 4

Can this bot be used with Kraken instead of Binance?

Accepted Answer

The code uses the python-binance API, so you'd need to adapt it for Kraken's API. The core LSTM model remains reusable, but API integration and data fetching logic would require changes.

Question 5

What's the best way to handle trading fees in simulations?

Accepted Answer

The project includes fee calculation in the trading test code, applying a 0.001 fee per trade as per Binance. Adjust this value based on your exchange and account type to simulate real-world costs accurately.

Question 6

How accurate is this LSTM model for predicting Bitcoin trends?

Accepted Answer

Validation accuracy is around 80%, but trading simulations show losses, indicating that accuracy alone isn't sufficient for profitability; feature refinement and better loss functions are needed, as noted in the results.

LSTM-Crypto-Price-Prediction

What is LSTM-Crypto-Price-Prediction?

Overview

Use Cases

Best For

Related Projects

Found a gem we're missing?

Not Ideal For

Pros & Cons

Pros

Cons

Frequently Asked Questions