Skip to content

A small, experimental LSTM for predicting stock prices.

Notifications You must be signed in to change notification settings

jdawson4/stockLSTM

Repository files navigation

stockLSTM

A small, experimental LSTM for predicting stock prices. Author: Jacob Dawson

Example of sequential output

Example of single output

Requirements:

  1. Tensorflow (best performance requires GPU speedup!)
  2. Numpy
  3. Sklearn
  4. Matplotlib
  5. Python (I hope that's obvious)
  6. robin_stocks, an API for accessing RobinHood's historical data. Requires RobinHood sign-in credentials!

Files:

  1. datasetMaker.py, which creates a file called "dataset.json" with data about stocks
  2. smallLSTM.py, which initializes and trains a neural network. It also saves the weights and network to this folder.
  3. make_sequence_prediction.py and make_single_prediction.py, which both load the pretrained network and use matplotlib to show some graphs about their predictions.
  4. simpleEvaluation.py, which will simply import the whole dataset and report the some metrics about the model's abilities on it.

About

A small, experimental LSTM for predicting stock prices.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages