A larger dataset would help in general as it would decrease the standard error of your estimates. But, the aim shouldn't be to "engineer" a statistically significant alpha. Ultimately, you'd like to know whether alpha was zero or not. An alpha estimate not statistically different than zero is also useful information...
We used 13-week t-bills in this tutorial. You can download the data from Yahoo! Finance (search for ^IRX). Using monthly t-bills would also make sense here as we're working with monthly returns. But, in general, the Jensen's alpha estimate wouldn't change much if you used a 1-month t-bill vs a 13-week one.
Thank you for this content, it is very informative and easy to follow. Just one item I am struggling with though, please could you advise where the T Bill data is sourced from and how it is formulated to give the percentages shown?
Sorry for the late response! We've downloaded the t-bill data from Yahoo! Finance. It's the 13-week treasury bill (symbol: ^IRX). If you go to the "historical data" tab, you can download the data as a spreadsheet file. What you need is the "Close" (or "Adj Close") column. And, unlike stocks, the data here is "return" (per annum) rather than "prices". Hope this helps!
![Lp. Сердце Вселенной #60 РОЖДЕНИЕ ЛОЛОЛОШКИ [Финал] • Майнкрафт](http://i.ytimg.com/vi/YoR0pAV9FVQ/mqdefault.jpg)
What could you do to make it more significantly significant? Just have a larger dataset spreading further back in time?
A larger dataset would help in general as it would decrease the standard error of your estimates. But, the aim shouldn't be to "engineer" a statistically significant alpha. Ultimately, you'd like to know whether alpha was zero or not. An alpha estimate not statistically different than zero is also useful information...
what maturity for treasury bills are you using and how do you decide?
We used 13-week t-bills in this tutorial. You can download the data from Yahoo! Finance (search for ^IRX). Using monthly t-bills would also make sense here as we're working with monthly returns. But, in general, the Jensen's alpha estimate wouldn't change much if you used a 1-month t-bill vs a 13-week one.
Thank you for this content, it is very informative and easy to follow. Just one item I am struggling with though, please could you advise where the T Bill data is sourced from and how it is formulated to give the percentages shown?
Sorry for the late response! We've downloaded the t-bill data from Yahoo! Finance. It's the 13-week treasury bill (symbol: ^IRX). If you go to the "historical data" tab, you can download the data as a spreadsheet file. What you need is the "Close" (or "Adj Close") column. And, unlike stocks, the data here is "return" (per annum) rather than "prices". Hope this helps!
When should you use artithmetic mean and when geometric mean in this context?