Find the results of the experiment in the jupyter_tensorflow folder. This is forked from the original authors project here: https://github.com/zcakhaa/DeepLOB-Deep-Convolutional-Neural-Networks-for-Limit-Order-Books.
Baseline File: run_train_tensorflow-version2_baseline.ipynb
Experiment with VisionLOB data augmentation step: run_train_tensorflow-version2-Vision.ipynb
In the FI-2010 data set we have 5 targets (prediction horizons). This experiment shows just the result on the 5th target, adjust variable k for other targets.
This jupyter notebook is used to demonstrate our recent work, "DeepLOB: Deep Convolutional Neural Networks for Limit Order Books", published in IEEE Transactions on Singal Processing. We use FI-2010 dataset and present how model architecture is constructed here. The FI-2010 is publicly available and interested readers can check out their paper.
Both tensorflow (version 1 and 2) and pytorch are available.