Hey folks!!I will take you guys through the world of finances with this blog where I will show you how to predict the stock shares of a particular organization using R.

Ours is a period of unpredictable economy. An organization earning big bucks today might not be even in the frame of contention in a period of few years. Of late, the stock market has gained utmost attention from all possible sectors because of the high returns it promises. But stock market is no place for a novice. The last decade has seen a number of big stock market players biting the dust due to lack of proper insights into the stock trends. Such is the nature of the stock arena, huge revenue generator one day, major reason for downfall the other day. As such, there is a need for a comprehensive stock value prediction system.

In layman terms, stock market prediction is nothing but trying to determine the future stock prices of a company based on historic and presently available data. For a more formal definition, refer wiki. This calls for a proper statistical analysis framework. As I have made myself pretty clear through the previous posts, when it comes to statistical modeling, I always prefer R due to the abundance of statistical analysis packages it provides. So let’s get started…

Let us look into the stock prizes of Google, Inc (GOOG). GOOG has been traded on the NASDAQ stock exchange since their IPO on August 19^{th}, 2004 at $85 per share. Now, it has seen a rise of about 1,294%. Quite a marvelous accomplishment one would say.

There are quite a few finance sites which would give you the historic stock prices of every company listed on NASDAQ, such as Yahoo and Google finance. Let me introduce some irony here and download the GOOG stock price from the Yahoo finance site. If you will open the link provided, you will see that the stock listing has a number of parameters such as:

- Date
- Opening and closing price
- High and low of the day
- The volume(number of stocks bought and sold)

Also, the beginning date is August 19, 2004, and the ending date is April 25, 2016 (when I retrieved the data). For the sake of convenience, I would just consider the stock prices from August 19, 2013 to April 25, 2016 on a monthly basis. You can see “Download to Spreadsheet” option at the bottom of the screen. Select that and obtain a CSV file of the stock prizes and store it in your working directory.

Now that we have the stock data, let us begin analyzing it to predict Google’s stock value for the next 2 years.

It is a multistage process which can be shown as:

__Step 1: Importing the data__

Firstly, we need to import the CSV file into our working environment using the read.csv () function. I will be using the “closing price” values for stock prediction. Refer the following screenshot for the process.

** Fig 1: Reading GOOG stock prize data**

__Step 2: Creating a time series function__

Stock data represents a time series. As we can see from the data, the closing prices are listed in the 5^{th} column, and are ordered from the most recent ones to the previous prices. To create a time series in R, ts() function is used. It can be shown as:

** Fig2: Converting input into a time series**

Here,

Rev (goog[,5]) -> reversing the closing price values

Start = c (2013,8) -> starting date for the stock prices

Freq=12 -> 12 months in a year

In addition to the single time series, we can manipulate the data by taking the log of values and placing them both into an accessible data frame.

** Fig 3: saving the closing values and the log values in a separate file**

__Step 3: Plotting the stock values__

Now we can load the data frame and plot the GOOG stock prices to get an idea of any trends in the prices over 3 years.

** Fig 4: to obtain a plot of the stock prices for 3 years**

As you can see on your window, the following plot appears:

As you can analyze, the period 2014-2014.5 saw a substantial fall of the prices, which again improved post 2015.

__Step 4: Decomposition__

There is a function, stl() which decomposes the GOOGLE time series into seasonal, trend and remainder components. stl () finds the seasonal component through loess smoothing (or taking the mean if s.window=”periodic”). The seasonal component is removed from the data and the remainder is smoothed to find the trend. The residual component is calculated from seasonal plus trend fit residuals. Use the stl() function to decompose the closing prices in **google**, and plot the result.

** Fig 5: google time series decomposition**

__Step 5: Forecasting using stl()__

Let us load the forecast function with the library () package. It enables us to forecast different time series models and linear models with the forecast () function. Let us consider the decomposed time series, google.stl, and determine a prediction method as “arima”. Let us predict 2 years into the future (h=24) with a 95% confidence interval (level=95). Then, visualize the results through a plot function.

** Fig 6: ARIMA forecasting**

The resulting plot can be shown as:

** Fig 7: Forecasted stock prices for Google**

With this result, I would definitely buy Google stocks.

So folks this was about share price prediction and forecasting of time series data. Hopefully you gained a few insights into the procedure. Keep checking this space for further posts on analytics.

**References**:

- For a tutorial on stock market prediction using a package named “Quantmod”, click here.
- For another example, click here.
- I referred this link for understanding ARIMA model.