Question on the appropriate way to calculate the CDF of a volatile measure

If I track the the volatility of a stock on a sample period and at every point in time where I record the volatility I would like to also record the CDF of the volatility. Should I take the entire volatility sample and compute its CDF or should I compute and update the CDF at every point in time?

In MATLAB I did the following [dens, xi]=ksdensity(volatility,volatility,'function','CDF') and I get the CDF at every volatility.

Is this the correct way to do it? What are the alternatives and the pros and cons?

-
 What does it mean to measure the volatility of a stock at a particular point in time? Isn't volatility of necessity a measurement over a period of time? By which I mean, at one single point in time, a stock has one value, how can you say if it is volatile or not. – Peter Ellis Jul 21 '12 at 21:45 Say that I measure volatility from Jan.00 to the end of Dec.00, I then have one value for volatility. Assume that on Jan 1st the following year the volatility spurs and I then compute volatility from Jan.00 to Jan1st 01....I do so for every day until the end of 2001. I would like to know how the CDF evolves every time I compute the new volatility. – CharlesM Jul 22 '12 at 21:18