Fat Tails ... and how to get 'em
suggested by a Motley Fool discussion
Stare carefully at these charts

>Yeah, so?
The first chart gives the daily returns for GE stock over the past eight years or so.

The middle gives a set of daily returns obtained by randomly selecting from the actual daily GE returns (over the same 8 years).

The last gives a set of daily returns obtained by randomly selecting from a Normal distribution of returns that has the same Mean and Standard Deviation as the GE returns (over those 8 years).

>Yeah, so?
Don't you see the difference?

>No! They're all funny looking!
The actual returns have wild variations in returns, reaching to over 8%, both positive and negative.

>But so does the middle chart!
Exactly! Aaah, but that last chart doesn't. Indeed, the returns selected from the Normal distribution are ... uh, tranquil in comparison.

>Tranquil? Is that a technical term?
The point is that actual returns can vary wildly. If you replace actual returns by some mathematical proxy ...

>Like a normal distribution of returns?
Yes. If you replace actual returns you don't get that wild variation.


>Yeah, so?
So, to get a feeling for a possible future evolution of GE prices (if you're doing Monte Carlo or predicting or estimating whether your portfolio will run dry), my suggestion is to forget about assuming a Normal or Lognormal or Micky Mouse distribution. Instead:

Select returns, at random, from actual historical returns

>Haven't you already said that?
Yes, several times ... like here.

>And "Micky Mouse"? Is that a technical term?
Go back to sleep.
>zzzZZZ

You can play with a spreadsheet.
Just RIGHT-click here and Save Target to download a .ZIPd Excel spreadsheet.