TradingView
Rashad
18 feb 2015 01:34

Moving CO-covariance (covariance on covariance) 

Bitcoin / U.S. dollarBitstamp

Descrizione

This is Covariance on Covariance. It shows you how much a given covariance period has deviated from it mean over another defined period. Because it is a time series, It can allow you to spot changes in how covariance changes. You can apply trend lines, Fibonacci retracements, etc. This is also volume weighting covariance.

This is not a directional indicator nor is moving covariance. This is used for forecasting volatility. This must be used in conjunction with moving covariance.
Commenti
Rashad
I would like to apologize, when naming this indicator I mixed up my terminology. This is coefficent of variation which shows the % a price deviates from its mean. Also known as unitized risk. In this case unitized risk of coefficent of variation
Rashad
In this indicator I forgot to multiply the result by 100. 0.0928 would be 9.28% etc. A small adjustment to the code will alleviate this. Just multiply cocovariance by 100, but if you like to read it as decimal points then this shouldn't matter to you.
miquael
interesting, so based on this info ... we can expect a downward motion soon?
miquael
or maybe just more very gradual incline up ?
Rashad
This isn't a directional indicator. It is for forecasting volatilty of volatilty. In this sense you can think of it as a 2nd rate derivative of volatility. By this, you should expect volatility to deviate 9.28% this week from the current expected 6.08% deviation over the next 30 days. Likewise the probability of this deviation forecast falls by (1/(Period length - number of days)). From a maximum of 68.26% probability derived from the empirical rule about the percent of data points 1 stdev from their mean. You can also apply trendlines and fib retracements in order to arrive at targets for both vol and vol on vol.
Rashad
Apply this to the VIX for some fun!
Rashad
Essentially the next 7 days could see (rough calculation without calculating decay from passing time of estimate) 0%-5.64%-12.74% + or - deviation in the next seven days. I can assume that the majority of data point will be within 5.64% to 12.74% + or - the current price, and could reasonably estimate that in the next seven days to expect a 5.64% change in either direction (+ or -)
Altro