Smart Stock Prediction

Chanjuan Wen

1 Background

Stock represents a portion of ownership of a company. Distinct from the property and the assets of a business, stock could fluctuate in quantity and value[1]. For stock traders, fluctuation in stock’s price directly determine the profit or loss. Therefore, to maximize the profit, it is important to choose stocks with good performance. However, accurately predicting the performance of a stock is very difficult for following reasons:

The public stock data we can access is limited. For example, we can only get the daily trading volume of a stock, but don’t know who sells it or buys it. Therefore, to get more accurate analysis of the performance of a stock, we need high quality financial data. But those data is usually too expensive to get.
Each stock has many properties, such as the company’s annual revenue, total asset, current assets, etc. The underlying relations between these properties and its performance is not easy to figure out.

In order to get a good prediction for stock price based on the limited public stock data, we can use machine learning algorithm to address this problem efficiently.

2 Objectives

Mainly based on the historic stock data of NASDAQ and S&P500 stock index, our goals for this project are the following:

classify an optimal set of stocks with great potential
predict the price of certain target stock accurately and efficiently

3 Methods

retrieve historic stock data from Yahoo Finance website through Yahoo Stock Quote API, and store them in local database.
use Support Vector Machine(SVM) to classify the relative performance of some stock against that of other stocks[2].
use linear regression method to make the price prediction of given stock[3].

4 Dataset

Our historic stock data is from Yahoo Finance. We will focus on the stock data of NASDAQ and S&P500 stock index, including the historic price, trading volume, company assets, total assets, etc. of each company.

5 Timeline

Period	Task
Apr. 13 - Apr. 22	Historic data of stocks in NASDAQ and S&P500
Apr. 23 - May 2	Define training set and analyze features vectors
May 2 - May 9	Implement the price prediction of target stock
May 10 - May 20	Implement the classifer for optimal set of stocks
May 21 - May 30	Optimization, testing and polishing results
May 31	Final report

References

[1] http://en.wikipedia.org/wiki/Stock

[2] Christopher King, Christophe Vandrot, John Weng. A SVM Approach to stock trading, 2009.

[3] Hung Pham, Andrew Chien, Youngwhan Lim. A Framework for Stock Prediction, 2009.