You could do any one of the following (though none are perfect):
1) Collect the intraday data and calculate the standard deviation of period returns yourself. This raises the issue of the frequency of period observation and how you would weight a volatility index. The standard deviation is obviously volatile itself so any index should be front weighted.
If you are trying to project volatility for binary betting take a look at some GARCH or EWMA models (Google them for more info). Forget GARCH(1,1) but there are some neat developments since.
2) Download the prices of binary options / bets from any of the vendors and extract the implied volatility according to their prices. This has the benefit of being empirical and market driven so is probably the best method of projecting volatility.
As the spreads are not insignificant you would be advised to look at the more liquid ones on a betting exchange and take the mean of the bid/ask. You can impute the implied vols by iterating them out of the pricing formula for the binary (in a Black Scholes framework).
3) If you want a quick approximation, based on market rather than historic data, you could take the next expiry's option prices, calculate the implied vol (use the Newton Raphson method seeded with the Manaster & Koehler start value to guarantee convergence) and then calculate a period implied vol as sigma*sqrt t.
Where sigma is the IV and t is the proportion of a year that you want (e.g, 1 hour would be 1/(24*252) = 0.000165344 assuming a 252 day count.
Personally I would probably opt for a combination of 1 and 2 if I wanted accuracy but 3 if I wanted a quick, market based answer.
Hope this helps,
NQR |