EECS 339 Project B DindaProject B: Portfolio Manager Now that you've had the experience of extending an existing database-backed webapplication (microblog), you're ready to design and implement your own. In this project,you will do so, developing an application from scratch that meets the specifications laidout here. This application combines requirements that are common to most interactiveweb applications with some elements of datamining and analysis using database engines. You should do this project in teams of 2-3 people. Please see us if you can't find a team. This is a brand new project for 2007. We may adjust expectations as needed during thecourse of the project.Overall RequirementsYou are to develop a web application that lets a user do the following:● Register for an account and log in. ● Access her portfolios. A user may have one or more portfolios.● Track a portfolio of the user's stocks, including their current value and other aspects oftheir performance. This part of the project will be most similar to microblog. ● Analyze and predict stock and portfolio performance based on historical performance.You will have access to about 10 years of historical daily stock data for this purpose.This part of the project will give you the opportunity to play with simple data mining,and rather more sophisticated prediction techniques. ● Evaluate automated trading strategies using the historic data. In the following, we'll explain a bit more about what each of these items mean, and givethe concrete specifications for each.What are Portfolios and Portfolio Management?A portfolio is a collection of investments that is intended to serve some purpose for theportfolio holder. For the context of this project, investments will consist purely of stocksand cash. We will also ignore stock dividends, taxes, margin, and trading costs. Thereare many other kinds of investments that could be included in a portfolio. Page 1 of 10EECS 339 Project B DindaWhile the stocks actually held by an individual investor certainly constitute a portfolio,portfolios are put together for other reasons too, for example to analyze how a particularcollection of stocks has done in the past, to predict how well it may do in the future, or toevaluate how well an automated trading strategy might work for the portfolio. An important investing problem is how to choose investments and their relativeproportions such that the risk (and reward) of the portfolio as a whole is controlled. Forexample, a 20-something computer scientist may be willing to have a much riskierportfolio than a retired 70-something teacher. The intuition behind designing a portfolio with a given “risk profile” is pretty simple. Theamount of risk of an investment (a stock here) is basically the variance of the value of theinvestment, sometimes normalized to the average value of the investment (standarddeviation over mean, or “coefficient of variation” (COV)). A high COV means the stockis “volatile” and thus riskier. Now, suppose you are trying to choose two stocks. Youcan not only compute their individual variances, but also their covariation (and thuscorrelation). If the two stocks are positively correlated, then this means that if both ofthem are in your portfolio, the combination will be more volatile (higher risk). If theyare negatively correlated, then the combination will be less volatile (lower risk). So,“portfolio optimization” is the process of choosing a collection of stocks such that theircovariances/correlations with each other combine to give you the variance (risk) that youwant while maximizing the likely return (reward). One simplification is to just considerthe correlation of each stock with the market as a whole (this is called the “Betacoefficient”) in building a portfolio. The Beta of the whole portfolio can thus be madelarger or smaller than the market as a whole by choosing the right stocks.The devil in the details is that in order to build a portfolio like this, we would need toknow the future values of volatility, covariance, Beta, etc, or of the stock pricesthemselves. We only have the past ones. So, a very important discipline is prediction,determining how a stock is likely to move in the future based on how it and all otherstocks have moved in the past, as well as predicting what the future values of the otherstatistical measures will be. Since the statistics are almost certainly nonstationary, wewill occasionally fail completely in predicting the future. The best we can do is muddlethrough, but there is a huge range of possibility, and a big part of any serious tradingenterprise is datamining historical data to develop better and better predictors. Another important consideration is automation. We would like to have a computerprogram that continuously adapts the portfolio holdings in pursuit of maximizing returnPage 2 of 10EECS 339 Project B Dindawhile controlling risk. These programs are called “trading strategies”, and anotherimportant goal of datamining of historical financial data is to find them.StocksA share of stock represents a tiny bit of ownership of a company. Companies paydividends (cash money) on their outstanding shares. The value of a stock is essentiallythe (discounted) sum of all of the future dividends it will pay. Since no one knows whatthat is, markets try to estimate it by buying and selling. The price at which a stock issold (the “strike price”) is an estimate of its value---the seller thinks the stock's value isless than the strike price, while the buyer thinks it's more. Notice that a “price event”happens on every sale, and there may be 1000s of sales per day of a stock. If you look atthe “stock price” in some typical free web stock quoting service, you're seeing an averageof these sales over some interval of time, or, in some cases, the most recent sale. For the purpose of this project, we will consider only information about the “day range”of a stock. In particular, for each day and stock, we shall have:● Day● Symbol (the alphabetic string representing the company – for example, AAPLrepresents Apple Computer)● Open (the strike price of the first trade of the day)● High (the highest strike price during the day)● Low (the lowest strike price during the day)● Close (the strike price of the last trade of the day)● Volume (the total number of shares traded during the