CORPFIN 2503 – Business Data Analytics
2021 S2, Workshop 6: Applications of multinomial logit models
£ius
1 Intraday stock market data: ANZ
Let’s download intraday stock market data (1 min) for ANZ for the last 3 days using
the following procedure:
1. Search for ANZ.ax
2. Click on Price & Charts
3. Select Price history
4. Change necessary settings, and
5. Click Excel icon on the right hand side.
By the way, there are more ways how to achieve the same result in Eikon.
2 Predict the next change in ANZ stock price (in
the lab and at home)
Suppose you are intra-day stock trader. You trade the same stock many times a
day. You would like to check whether it is possible to predict the direction of ANZ
stock price for the next minute. Suggested steps:
1. using the data downloaded in Task 1, create a *.csv �le with the following
variables:
� date
1
� time
� volume
� net;
2. import this data set into SAS
3. generate a variable (change_ind) that equals:
� �No change� if net = 0
� �Decrease� if net < 0 � �Increase� if net > 0;
4. create 2 dummy variables:
� = 1 if lagged net < 0, and 0 otherwise � = 1 if lagged net > 0, and 0 otherwise;
5. create a new variable that equals to the natural logarithm of the lagged value
of volume (lag_ln_volume)
6. estimate multinomial logit regression model where:
� the dependent variable is change_ind
� the independent variables are lag_ln_volume and the 2 dummy vari-
ables that re�ect the lagged signs of net
� in the analysis ignore the �rst and last 15 minutes of trading time; in
other words, estimate the regression for those observations that satisfy
the following condition:
� time>=”9:45:00.000″t & time<="15:15:00.000"t 7. save the predicted probabilities 8. (at home) manually compute the predicted probabilities for the �rst observa- tion 9. (at home) discuss the accuracy of the model by comparing the predicted values of the dependent variable with its actual values. 2 Intraday stock market data: ANZ Predict the next change in ANZ stock price (in the lab and at home)