Predicting the Future Stock Prices of ConAgra

This report analyzes past data to predict future changes in the stock prices of ConAgra. It includes ANOVA, regression analysis, and coefficient determination. The dataset contains data on 378 trading days over a two-year time period from 4th April 2016 to 26th March 2018. The report concludes that only 15.9% of the variability in the stock prices can be explained by the independent variables.

Running Head: PREDICTING THE FUTURE STOCK PRICES
Predicting the Future Stock Prices of ConAgra
Name of the Student
Name of the University
Author Note

1PREDICTING THE FUTURE STOCK PRICES
Table of Contents
1.0 Executive Summary.............................................................................................................2
2.0 Description of the Data........................................................................................................2
3.0 Variance Inflation Factor.....................................................................................................3
4.0 Residual Analysis.................................................................................................................3
5.0 Analysis of Variance (ANOVA)..........................................................................................4
6.0 Coefficient of Determination (R2)........................................................................................5
7.0 Testing of Hypothesis..........................................................................................................6
8.0 Coefficients..........................................................................................................................6
9.0 Prediction of Share Prices....................................................................................................7
10.0 Conclusion..........................................................................................................................8
Appendix....................................................................................................................................9

2PREDICTING THE FUTURE STOCK PRICES
1.0 Executive Summary
The main aim of this report is to analyse and evaluate the past data and assess whether it can
be used suitably to predict the future changes in the prices of the stocks of the company
ConAgra. The stock market data is on various products from April 2016 to March 2018. The
company deals mainly with packaged food services and is one of the largest packaged food
services across the globe.
The daily data will be used to conduct the analysis. The relationship between the changes in
the prices of Euro with the help of a multiple regression model. The multiple regression
model will include the variance inflation factor (VIF) and the analysis of variance (ANOVA)
along with the value of the adjusted R-square.
2.0 Description of the Data
The dataset contains data on 378 trading days over a two-year time period from 4th
April 2016 to 26th March 2018. There are 6 measurements of inputs from selected days which
are considered as independent variables and one output variable which is considered as the
dependent variable. The dependent variable in this paper is the future prices of ConAgra. The
daily changes in prices of different assets are measured in each of the columns of the dataset.
These financial assets include:
 Gold
 Aluminium
 The Baltic Dry Index
 Canada Intermediate Oil
 The index of Stock Prices of Standard and Poor 500 (S&P500)
 The future prices of ConAgra

3PREDICTING THE FUTURE STOCK PRICES
In the dataset, it can be observed that most of the variables that have been used for the
prediction of the future changes in the stock prices are interaction variables. The original data
has been transformed to this dataset where the changes in the prices vary between 0 and 1.
The changes in the prices that have occurred originally has been ranked and sorted after that.
After the sorting, the percentage changes are divided by 378. Thus, 0 indicates the highest
decrease in the price, 1 indicates the highest increase in the price and the median change in
the price is indicated by 0.5.
3.0 Variance Inflation Factor
Before analysing a data, it is important to test the existence of multi-collinearity to the
dataset. Multi-collinearity indicates the existence of high correlation between two or more
variables. It is difficult to fit a model to the existing data in the presence of multi-collinearity.
In order to test for multi-colinearity, the variance inflation factor (VIF) test is conducted.
The PHStat add-in has been used in Excel to do the analysis. The VIF values obtained
as a result of the analysis for each of the independent variables are less than 5. Thus, it can be
said that the independent variables are uncorrelated or very little correlation. Thus, the
independent variables are also independent of each other and there is thus no need to delete
any variables from the study.
4.0 Residual Analysis
Residual analysis gives an idea of the presence or absence of outliers to the data. This is
estimated with the help of the normal probability plot. The normal probability plot given in
figure 1 shows that the distribution of the data is normal. Thus, calculations can be done on
this data and the results will be valid as the regression and ANOVA analysis follows the
assumption of normality. The histogram given in figure 2 also shows that the residual is also

4PREDICTING THE FUTURE STOCK PRICES
following a bell shaped curve almost. Thus, it can be said that the residuals are normally
distributed.
-4 -3 -2 -1 0 1 2 3 4
0
0.2
0.4
0.6
0.8
1
1.2
Normal Probability Plot
Sample Percentile
Future_change
Figure 1: Normal Probability Plot showing non linearity
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 More
0
10
20
30
40
50
60
Histogram
bins
Frequency
Figure 2: Histogram of Residuals following almost a bell shape
5.0 Analysis of Variance (ANOVA)

5PREDICTING THE FUTURE STOCK PRICES
Table 1 given below gives the results of the analysis of variance (ANOVA). With the
help of ANOVA, the existence of the relationship between the independent and the dependent
variable can be established. It can be seen from table 1 that the Significance F value, also
known as the p-value has been obtained as 0.000, which is much less than 0.05. This
indicates that the null hypothesis is rejected. In case of ANOVA, the null hypothesis states
that there is no significant relationship between the independent and the dependent variables.
AS the null hypothesis is rejected, it can be concluded that there exists a linear relationship
between one of the five independent variables and the future share prices.
From the ANOVA table, the strength of the relationship cannot be established. The
relationship can be strong or weak. The coefficient of determination is thus required to
establish the strength of the relationship.
Table 1: ANOVA Table showing the Existence of Relationship
df SS MS F Significance F
Regression 6 5.2054 0.8676 11.6919 0.0000
Residual 371 27.5292 0.0742
Total 377 32.7346
6.0 Coefficient of Determination (R2)
Table 2 shows the value of R2 obtained as a result of the regression analysis. From the
table, it can be seen that the R2 value is 0.1590. Thus, the percentage of the future stock
prices of ConAgra Brands that can be explained by the independent variables is 15.9 percent.
Thus, it can be said that the model is not effective in explaining too much of the future stock
prices of ConAgra Brands. When the future stock prices of ConAgra Brands is unpredictable,
the value of R2 will become close to zero. Thus, it can be concluded from here that the future
stock prices of the company are not totally random.

6PREDICTING THE FUTURE STOCK PRICES
The strength of the relationship between the variables is yet to be assessed. This will
be discussed in the following section.
Table 2: Regression Statistics
Multiple R 0.3988
R Square 0.1590
Adjusted R Square 0.1454
Standard Error 0.2724
Observations 378
7.0 Testing of Hypothesis
From table 3, it can be seen that the p-value for all the input variables are less than
0.05 except for the variable Euro. Thus, it can be said Euro is insignificant in predicting the
stock prices for the company. This variable can be eliminated from the model. The rest of the
4 input variables have a relationship with the output variable.
Table 3: P-Value for the Independent Variables
Coefficie
nts
Standar
d Error t Stat P-value
Intercept 0.4983 0.0347 14.3506 0.0000
30yearBond_x_CanadaDollar_x_Gold_x_SP500 -0.7614 0.1885 -4.0404 0.0001
30yearBond_x_Sugar -0.3309 0.1126 -2.9393 0.0035
30yearBond_x_Consumer_x_Aluminium_x_Canada
Dollar 0.6884 0.1366 5.0378 0.0000
Baltic_x_Oil_x_CAG_x_Grains -0.6057 0.1230 -4.9239 0.0000
Year_x_Sugar_x_Oil 0.5551 0.1539 3.6070 0.0004
Euro 0.0520 0.0611 0.8507 0.3955
8.0 Coefficients
In table 3, the column named coefficients indicates the extent to which the input
variables or the independent variables affect the future stock prices of ConAgra Brands. By
ignoring the y-intercept, and investigating the other variables the effect of the variables on the
future prices can be assessed.

7PREDICTING THE FUTURE STOCK PRICES
The largest positive coefficient (0.6884) can be seen for the interaction variable
“30yearBond_x_Consumer_x_Aluminium_x_CanadaDollar”. This interaction variable is
created by multiplying
 30-year treasury bond
 Number of consumers
 Prices of Aluminium
 Prices of Canada Dollar
There is a variation in the values with respect to different seasons. With the increase
in the prices of aluminium, there will be an increase in the number of consumers and the
prices of Canada Dollars. Thus, when the price of aluminium increases, the stock prices
increase for ConAgra Brands.
The largest negative coefficient (-0.7614) can be seen for the interaction variable
30yearBond_x_CanadaDollar_x_Gold_x_SP500. This interaction variable is created by
multiplying
 30-year treasury bond
 Standard and Poor 500 Prices
 Prices of Gold
 Prices of Canada Dollar
Thus, when the price of aluminium increases, the stock prices decrease for ConAgra
Brands. All the significant variables have a coefficient value which is not close to zero. Thus,
no significant needs to be deleted from the model. Only the insignificant variable “Euro” can
be removed from the model.
9.0 Prediction of Share Prices

8PREDICTING THE FUTURE STOCK PRICES
It can be seen from the table 4 given below that the actual values of the stock prices of
ConAgra Brands is within the 95 percent confidence interval limits. Thus, there are no errors
in the model developed.
Table 4: Confidence Interval for Actual and Predicted Prices of ConAgra
Confidence Level 95%
Date 3/21/2018 5/2/2018 5/31/2017 9/28/2016 9/11/2016
Predicted Y (Y hat) 0.92398 0.37832 0.52006 0.67611 0.61577
Actual Change 0.99599 0.01202 0.99800 1.00000 0.00000
For Average Predicted Y (Y hat)
Half Interval Width 0.03132 0.03061 0.01249 0.01481 0.01403
Confidence Interval Lower Limit 0.89266 0.34771 0.50756 0.66130 0.60174
Confidence Interval Upper Limit 0.95530 0.40892 0.53255 0.69092 0.62981
For Individual Response Y
Half Interval Width 0.07635 0.07523 0.03197 0.03940 0.03690
Confidence Interval Lower Limit 0.91965 -0.06320 0.96603 0.96060 -0.03690
Confidence Interval Upper Limit 1.07234 0.08725 1.02997 1.03940 0.03690
10.0 Conclusion
The future Stock prices of the company ConAgra is estimated from the given dataset.
The VIF values have shown that no correlation exists between the independent variables. The
normal distribution of the variables is established with the help of the normal probability plot
and the histogram. Thus, regression have been conducted. It has also been observed that only
15.9 percent of the variability in the stock prices can be explained by the independent
variables. The ANOVA table shows the existence of significant relationship between the
independent and the dependent variables. The predicted and the actual prices differ as the
value of R2 is less. If the R2 value can be increased, the difference between the actual and the
predicted values will be reduced.

9PREDICTING THE FUTURE STOCK PRICES
Appendix
Regression Analysis
30yearBond_x_CanadaDollar_x_Gold_x_SP500 and all
other X
Regression Statistics
Multiple R 0.5304
R Square 0.2813
Adjusted R Square 0.2717
Standard Error 0.0749
Observations 378
VIF 1.3914
Regression Analysis
30yearBond_x_Sugar and all
other X
Regression Statistics
Multiple R 0.4623
R Square 0.2137
Adjusted R Square 0.2031
Standard Error 0.1254
Observations 378
VIF 1.2718
Regression Analysis
30yearBond_x_Consumer_x_Aluminium_x_CanadaDollar and all
other X
Regression Statistics
Multiple R 0.3646
R Square 0.1329
Adjusted R Square 0.1213
Standard Error 0.1034
Observations 378
VIF 1.1533
Regression Analysis
Baltic_x_Oil_x_CAG_x_Grains and all
other X
Regression Statistics
Multiple R 0.2045
R Square 0.0418
Adjusted R Square 0.0289

10PREDICTING THE FUTURE STOCK PRICES
Standard Error 0.1148
Observations 378
VIF 1.0436
Regression Analysis
Year_x_Sugar_x_Oil and all
other X
Regression Statistics
Multiple R 0.6661
R Square 0.4438
Adjusted R Square 0.4363
Standard Error 0.0918
Observations 378
VIF 1.7978
Regression Analysis
Euro and all other X
Regression Statistics
Multiple R 0.6009
R Square 0.3611
Adjusted R Square 0.3525
Standard Error 0.2310
Observations 378
VIF 1.5651