Niner

"Go West, young man" - Horace Greeley

• Project overview:
  • train deep nueral network (DNN) model on historical cryptocurrency market data (BTC-USD, ETH-USD, etc.)
  • features of the historical market data:
    • date
    • close (i.e., closing price of crypto product of choice)
  • the model's hidden layers primarily utilize the LSTM architecture
  • the DNN model's training process uses the following loss function(s):
    • MSE -> temporary
    • A3C -> weird, custom loss function not implemented yet
  • predict crypto prices given market data input (i.e., x_test)
    • x_test is fed to the model manually for now, but will eventually pull from our internal DB of realtime prices
  • make various trade decisions (buy/sell/hold) based on the confidence internal of our prediction
  • make 95% successful trade predictions, barring global chaos

Disclaimer

• This is an active side project that I maintain in my free time, don't use it in production environments.

Installation

If available in Hex, the package can be installed by adding niner to your list of dependencies in mix.exs:

def deps do
  [
    {:niner, "~> 0.1.0"}
  ]
end

Documentation can be generated with ExDoc and published on HexDocs. Once published, the docs can be found at https://hexdocs.pm/niner.