https://github.com/philipperemy/stock-volatility-google-trends

Deep Learning Stock Volatility with Google Domestic Trends: https://arxiv.org/pdf/1512.04916.pdf
https://github.com/philipperemy/stock-volatility-google-trends

google-trends keras keras-neural-networks keras-tensorflow tensorflow tensorflow-examples tensorflow-experiments

Last synced: 4 months ago
JSON representation

Deep Learning Stock Volatility with Google Domestic Trends: https://arxiv.org/pdf/1512.04916.pdf

• Host: GitHub
• URL: https://github.com/philipperemy/stock-volatility-google-trends
• Owner: philipperemy
• Created: 2017-09-27T06:31:58.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2022-02-17T02:12:35.000Z (almost 3 years ago)
• Last Synced: 2024-05-02T05:11:13.596Z (9 months ago)
• Topics: google-trends, keras, keras-neural-networks, keras-tensorflow, tensorflow, tensorflow-examples, tensorflow-experiments
• Language: Python
• Homepage:
• Size: 1.29 MB
• Stars: 89
• Watchers: 12
• Forks: 40
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Deep Learning Stock Volatility with Google Domestic Trends

Tensorflow/Keras implementation of the [[paper](https://arxiv.org/abs/1512.04916)].



  


  Trend displayed in Google Domestic Trends



## Status?

Work accomplished so far:

- [x] End-to-end implementation

- [x] Test the data workflow pipeline

- [x] Sanity check of the model 

- [x] Train the models 

- [x] Reproduction of the results in the paper

## Plot



  


  MAPE of the train, val and test along with dummy (benchmark future value = last value)



A new predictor is added every 600 epochs. We start with only the historical volatility as a predictor.

Then, at 600 epochs, the second predictor is added: returns. At 1200, we had Trend COMPUT, 1800 Trend CRCARD, 2400 Trend INVEST and so forth.

We realized that we started to overfit after more than 5 predictors. The dataset is indeed incredibly small.

The lowest model MAPE loss on the test set coincide with the lowest MAPE loss on the validation set. They both have a comparable value (around 25), in agreement with the results found in the paper.

As a summary, the 5 best predictors are:

- sigma (volatility)

- returns

- Trend COMPUT

- Trend CRCARD

- Trend INVEST

As a conclusion, I'm a bit curious to know how they trained a model with 30 predictors on a such tiny dataset. They don't mention any information related to the model. It's just composed of a LSTM layer. I'm also a bit skeptical about this approach in general.

## How to run it?

```bash

# might require python3.6.

git clone https://github.com/philipperemy/stock-volatility-google-trends.git svgt

cd svgt

pip3 install -r requirements.txt

python3 run_model.py

```