12/08/2021

INTRODUCTION

  • In 2021, 56% of Americans reporting that they own stock
  • Financial planning and decision making
  • Popularity of algorithmic trading

PEER COMMENT SUMMARY

  • Why 2 year data -> We have included 20+ years of data

  • Technical indicators -> we haven’t changed proposal w.r.t this

  • Does your project have a plan to use NLP? -> No,We have only stock prices and related data to consider for prediction

  • Not sure what prices will be compared. -> Closing price of each day is what we are considering.

PROPOSAL MODIFICATION

  • Initial plan was to use 15 min interval data for analysis.

  • Since 15 min interval is not much relevant unless it’s for intra day trading. we changed our plan and we are considering only daily data

  • only the content frequency has changed. Data remains the same

DATA SOURCE

Alphavantage API - https://www.alphavantage.co/

  • Time Series Stock APIs : Time series covering 20+ years of historical data
  • We are choosing Intraday data ( 15 min interval ) for our analysis

DATA

  • Package : Alphavantagr

  • DAILY ADJUSTED DATA

  • TARGET VARIABLE : ADJUSTED_CLOSE

Data pre processing

  • Convert to timeseries

  • Scaling

DATA VISUALIZATIONS

DATA VISUALIZATIONS

Data Modeling

LSTM ( KERAS ):

Parameters

One input and one output per data point

Epochs : 10

layers : 2

LSTM RESULTS ( COSTCO ) :

Mean Error ( 7 days ) : 2.5%

LSTM RESULTS ( TARGET ) :

Mean Error ( 7 days ) : 1.6%

Conclusion

  • Improve the models

  • Include technical indicators/ other metrics