Order book information contains ask, bid prices and corresponding quantities at each level.
Note! price jumps : current bid price > previous bid price (within short period of time)
- Example JuypterNotebook-Rnn
- "./data" contains sample data from upbit, tick-tick information of L2 orderbook (KRW-ADA)
- features are from https://github.com/dzitkowskik/StockPredictionRNN/blob/master/docs/project.pdf
from nn import NeuralNetwork
from rnn import RNN
timestep = 50
n_cross_validation = 3
# for order book info only
data = data_prep.get_test_data(timestep, predict_step=5, filename="upbit_l2_orderbook_ADA")
# input_shape <- (timestep, n_features)
# output_shape <- n_classes
nn = NeuralNetwork(RNN(input_shape=data.x.shape[1:], output_dim=data.y.shape[1]), class_weight={0: 1., 1: 1., 2: 1.})
print("TRAIN")
nn.train(data)
print("TEST")
nn.test(data)
print("TRAIN WITH CROSS-VALIDATION")
nn.run_with_cross_validation(data, n_cross_validation)keras, tensorflow, sklearn, nuumpy, pandas ...
pip install -r requirements.txt[LSTM]https://github.com/miroblog/limit_orderbook_prediction/blob/master/rnn.py
[CNN-LSTM] https://github.com/miroblog/limit_orderbook_prediction/blob/master/cnn_lstm.py
- Lee Hankyol - Initial work - L2-Prediction
This project is licensed under the MIT License - see the LICENSE.md file for details
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent