ET News & Sponsor Info
General Topics
Technical Topics
Brokerage Firms
Tools of the Trade
Trading for a Living
Community Lounge
Site Support

# Calculating volatility measure from price movement alone

Discussion in 'Strategy Development' started by logic_man, May 20, 2011.

1. ### logic_man

Let's say I want to create a measure of an instrument's volatility, is there a way to do that solely based on price movement of that specific instrument? Something like price fluctuations per some defined period of time?

2. ### shortie

google for "volatility estimator"

3. ### taowave

Not sure what you are asking for,but it sure sounds like "Historical Volatility" which is readily available..Its a pretty simple formula,just google it..

You could also lookat ATR,but I think Historical Vol is what you are after..

4. ### logic_man

I could see how that seems to be what I'm asking for. The real application I'd like to do with this is to compare the volatility of the ES during the overnight session with the volatility during the day session, as a filter for trading triggers.

5. ### logic_man

Thanks, I will do some digging. It looks like I'll need tick-level data.

7. ### emg

Only HFT traders have such system. HFT recruit and employed top notch math and computer science grad.

8. ### ddude

No need for a quant when you have R 8^)
In R you could do something like:

# load a nice quant library...
require(quantmod)

# get the daily data from yahoo for a year back...
symbolData <- getSymbols('aapl', auto.assign=FALSE)

# this is the close-to-close volatility...
vClose <- volatility(symbolData, calc="close")

# display the end of the time series of volatility...
tail(vClose)

# this would be the Garman Klass kind of volatility...
vGK <- volatility(symbolData, calc="garman")

# well you get the idea...
vParkinson <- volatility(symbolData, calc="parkinson")

# wow nice library yea... and there is a Yang Zhang too...
vRS <- volatility(symbolData, calc="rogers")

Cordially,

-Digital Dude-

I don't think this has to be complex, simply get the % change in the time periods and get the standard deviation of it as a proportion of the stock value. Try it, it gives a very commonsense result. Other versions might include skew but are not necessarily better for this.

It never ceases to amaze how traders (baffle with BS) turn a simple issue into a complex one.