search

nav3van / RNN-Trading-Bot

Artifical neural network that executes trades on the BTC-e bitcoin exchange upon analysis of historical trade data.
MIT License
19 stars 11 forks source link

Did you only use Recurrent Neural Network to predict price of bitcoin? #3

Open ll36771 opened 8 years ago

ll36771 commented 8 years ago

I also did a lot work on this project.I use LSTM->SRNN->Linear to train to predict the price of bitcoin.But the accuracy is only about 55%.What's accuracy can you get?

alexcoy commented 7 years ago

ms.exel =forecas lebih falit *average kesalahan forecas sebelumnya

alexcoy commented 7 years ago

=FORECAST(C1137,$C$2:C1137,$C$2:C1137)*J1138

j1138=average lose history FORECAST

Byte1122 commented 7 years ago

Hi @ll36771 @alexcoy can you look into this, I hope you can help me out... when running trading_floor.py

All required directories and files detected
Waiting for first prediction from neural network 0:0:25Traceback (most recent call last):
  File "trading_floor.py", line 63, in <module>
    usd_to_btc_qty = get_order_size(err_rate) * bal.usd_bal()
                     │              │           └ <balances.Balance object at 0x76a1c6b0>
                     │              └ 'inf'
                     └ <function get_order_size at 0x767ccfa8>
  File "/home/DDcc/neuralnet/RNN-Trading-Bot/pyTrader/analyst.py", line 80, in get_order_size
    if _error_rate < 0.0005:
TypeError: unorderable types: str() < float()