/deep-learning-for-order-book-price-and-movement-predictions

Limit Order Book data analysis and modeling using LSTM network

Primary LanguageJupyter Notebook

Limit order book modelling with Deep Learning (LSTM network) for price and market movement predictions

Repo contains files and data for:

  • Cleaning limit order book data scraped from Binance
  • Exploratory Data Analysis on ETHBTC trades and orders
  • Price volatility calculation
  • Feature engineering the order book and trades data for Deep Learning
  • Modelling the order book using Long Short Term Memory (LSTM) network for market movement and price predictions
  • Shap values of the created models

Acknowledgments:

Parts of this project uses researches from:

D. T. Tran, M. Magris, J. Kanniainen, M. Gabbouj, A. Iosifidis, "Tensor Representation in High-Frequency Financial Data for Price Change Prediction" 2017. https://arxiv.org/pdf/1709.01268.pdf

J. Sirignano, R. Cont, "Universal features of price formation in financial markets: perspectives from Deep Learning" 2018. https://arxiv.org/pdf/1803.06917.pdf