Tips on modeling?

[johnchristopher]

Hi everybody!

I was wondering if anyone could give me some helping pointers with something.

I'm trying to find the most suitable model for a single asset's stock returns, using Matlab and historical data.

However when I use normplot on the data, one of the tails is rather fat, and I can see that It's not completely normal-distributet.

As you might have figured out, Im kind of new at this

So i was wondering if anyone had some tips or pointer on methods on how to handle this.
It's my first time modeling, so I'm kind of scared of going in the complete wrong direction when solving this.

And by the way,
thanks for a GREAT site!!

 

[Ross]

Are you saying that you're seeing stock returns that aren't normally distributed? That's the correct result. Tails of stock returns are usually fatter than a normal distribution.

 

[Sanket Patel]

As Ross pointed out, equity returns are anything but normal - they have excess kurtosis. You'll get a better fit using t-distribution.

 

[johnchristopher]

Hi Sankel Patel and Ross,

Thanks for the quick respons.
As i mentioned I'm new at this, and although I've heard before that the tails usually are fatter, I don't really know how to model it.

I.e. what should one do/think of when modeling returns?

t-distribution sounds interesting. Is there someway in Matlab to quickly check how well this would fit, something similar to normplot, probplot maybe?

Once again, thanks for the quick respons.