Hi, i m taking the course of Statistical Arbitrage. I have downloaded the spreadsheet to calculate Y, Y^, Ye and Yd but the way they make it in the video is different as to what the spreadsheet really have.
In the video they calculate one (1) t-statistics value depending on the timeframe: 90, 120, overall duration. In the downloadable spreadsheet, they seem to calculate a rolling average of values and indicate that it's cointegrated for certain periods of times. Don't really understand the full idea:
do the t-statistics value change all the time or is it better to calculate it for different timeframes?
check the images i attached in google drive
2 are from the downloadable section
1 is a screenshot from the video
Hello Jose,
It's great to know you're progressing on the course.
You have an important question that leads us to the following explanation.
Relationships between economic variables are not stable throughout time. They actually pass through market regimes and each market regime makes the variables' relationships change.
Cointegration between 2 or more variables doesn't happen all the time. They also respond to market regimes.
So, whenever you find a cointegrating equation for a specific data span, you might not encounter the same for the next period. Consequently, you have the obligation to check again for cointegration to ensure the hedge ratio you found in your in-sample data is maintained for out-of-sample data.
That's why the excel sheet prepares you to have the correct procedure on how to trade mean reversion.
When you are going to trade, you specify the in-sample span you will use to check for cointegration. Then you apply the ADF test and then you say the following:
If I find cointegration, then I proceed to invest,
If I don't find cointegration, I won't trade.
With this procedure, you will be sure you will always invest your money in periods in which there is cointegration. Remember, relationship between variables changes all the time! So, it's better to ensure first cointegration and invest accordingly than to relay on a single estimation and believe cointegration will sustain forever between the traded instruments.
I hope this helps,
Thanks and regards,
José Carlos
thanks for your help Jose. Fast response, clear and well explained… 5 stars for you mate
Hello Jose,
Good to know you understood everything.
Whenever you have questions, we'll be here to help you for sure!
Regards
Jose Carlos