Clustering Model for Stocks


Investment risk is loss of capital or reduced returns due to market fluctuations and stock’s performance. Technical and Fundamental Analysis of stocks can be used to identify stocks with low risk and better returns.

CANSLIM is one such techno-fundamental approach. CANSLIM stands for, C – Current Quarterly Earnings; A – Annual Earnings Growth; N – New Product or Service, New Management, New High; S – Supply & Demand; L – Leader or Laggard; I – Institutional Sponsorship; M – Market Trend or Market Direction

In this paper the authors would study the stocks listed in NASDAQ & NYSE. The stock performance data as well as the company financial data is for 3 years are collected from Yahoo Finance. A company’s stock gets a rating (0-7) basis the CANSLIM criteria, We use earnings per share this quarter vs same quarter previous year, annual earnings per share, number of outstanding stocks, investor concentration, debt to equity ratio, market share, market Indices like S&P 500 to rate the stocks. For analyzing stock performance, we calculate Risk/Reward ratio, 200 day moving average, 52 week change % etc. For day trading, we use the Open/Low/Close to calculate the Risk/Reward ratio. For position trading we use the Open/52 week low/ 52 week high to calculate the Risk/ Reward ratio. We then use Risk/ Reward Ratio to classify stocks into Low/Medium/High Risk Categories (Low > 0.5, Medium – 0.3 to 0.5, High <0.3).

We also use visualization techniques to study the stock performance: Histogram to see the distribution of returns of a given stock; OHCL Graphs; Bollinger bands; Identify what properties constitute a stock into Low/Medium/High Risk Categories: Study the relationship between risk categories and – Industry Type, Market Cap, Revenue, Beta, P/E Ratio.

This study will use clustering techniques to groups the stocks into Good, Average and Bad. This can be used to predict the behavior of stocks when they move from one cluster to another, or when there is a new stock in the market. This analysis can help traders and investors identify better stocks, which can help minimize risk and maximize the returns for day as well as position trading.

Presented and Published in: Proceedings of the Sixth International Conference on Business Analytics and Intelligence, December 2018, IISc, India.


Shravani Ponde

Senior Consultant, PwC

Ratnakar Pandey

Heading Data Science & Analytics for CX, Amazon

Leave a Reply

Your email address will not be published. Required fields are marked *