Ever seen the wonders that data can do?
Recently I was studying linear regression and prediction intervals. I decided to make a regression model for Apollo hospital share price (NSE: APOLLOHOSP) based on NIFTY 50 stock price. Guess what, the results were spectacular!
I took the daily closing prices from 1 Jan 2022 to 1 Aug 2022.
The model was statistically significant after all the t-tests and f-tests were used to determine if there will be any difference in results if we computed the value of Apollo hospital based on NIFTY 50 and if we compute the value independently. And Woah, R square (Explaining how much variance in Apollo can be explained by NIFTY) came to be 62%.
Then, based on my future estimate of NIFTY 50, I moved on to calculate the prediction interval at a 95% significance level - which in Stat's terms means that the model should be correct 95% of the time.
But then, It is based on just a single price of NIFTY-50.
Could we expand it based on sensitivity analysis at different data points of NIFTY-50? Why not?
Could we also calculate VaR (Value at risk- used to calculate the minimum loss in a given day) to accompany our analysis, Why not?
But let's discuss these approaches on a different note! Until then!
Provide your opinions on- What do you think, historical data could be used for future analysis?
Disclaimer: This is not investment advice.