/acropolis

Stock prediction using various regularization techniques on linear regression.

Primary LanguageMATLAB

acropolis

Stock prediction using various regularization techniques on linear regression.

Objectives

Predict % yield of next quarter

Features

  1. Split / Dividends - Ramya
  2. Market Cap (Share price times total shares) - Vronsky
  3. Volume - Ramya
  4. P/E (Price to Earnings Ratio)[Market value per share / Earnings per Share (EPS)] - Parker
  5. SMP (Standard & Poor rating) - Shawn
  6. Field of Business (Ag, Tech, Finance) - Dan
  7. Size (Employees / Stockholder) - Chris
  8. State - Dan
  9. Number of financial quarters this company has existed before our start date - Dan
  10. Kurtosis - Ramya

Notes

  1. Companies must have been created before 2000 and in USA
  2. We will train on 30 companies
  3. We are running 30 DISTINCT regressions - one per company.
  4. We are using a Neural Network from the DeepLearn Toolbox. https://github.com/rasmusbergpalm/DeepLearnToolbox
  5. All company data will start on the year of the most recently created company
  6. Features are taken on a per quarter basis

Tasks

Tue 11/18: Everyone is assigned a feature. Research it and present findings. Don't need data right now, just need explanation of how to use it.