Statistical Volatility - A New Model Required - Aussie Stock Forums

# Thread: Statistical Volatility - A New Model Required

1. ## Statistical Volatility - A New Model Required

Most of what I learned about trading options was learned trading on the US exchanges. Not very many weeks ago, I decided to migrate my account back to the Australian Stock Exchange. The reasons for this were exclusively lifestyle based; I didn't want to work Friday nights anymore because I was missing too many good piss-ups. If I was going to be asleep half of Saturday, it is going to be for recovery, rather than because of nocturnal money making habits.

Anyway that's beside the point.

In the United states the standard measure of statistical volatility (SV) is a simple 20 or 30 day, annualised percentage volatility equation. This is expressed in Amibroker or Metastock formula lanuage like this:

(StDev(log(C/Ref(C,-1)),20) * sqrt(252))*100

You could plot this on a graph, along with the implied volatility (IV), and see very plainly that IV tended to reflect the 20-30 SV. I have underlined "tended to reflect", because SV and IV will never always match each other. SV is measured over past data, whereas IV is an estimate of future volatility. Both SV and IV oscillate in a quite broad range in those markets.

However upon applying my templates to the Aussie market, something was wrong, something was different. IV tended to stay in a much tighter range compared to the US. But the standard 20 day SV charted a much more extreme course than IV. IV was not reflecting this measure of SV at all.

It was clear to me that the Australian market makers were using a different measure, a much more sedate measure than the Americans. Simply lengthening the look-back period of the standard SV equation wasn't doing it, much to lagging and innacurate when current volatilities were changing. I played with exponential moving averages of SV, but that didn't really work either. Then I had this idea; how about an exponentially weighted SV equation? My idea was a longer term SV measure but weighted with the most recent SV, much like an exponential moving average.

This seemed to work! After playing with different parameters, I got an SV measure that IV seemed to be reflecting.

So herewith I tender my Exponential Statistical Volatility (In Amibroker AFL):

long = (StDev(log(C/Ref(C,-1)),100) * sqrt(252))*100;short = (StDev(log(C/Ref(C,-1)),10) * sqrt(252))*100;
x1 = short - Ref(long,-1);
x2 = (x1* (2 / (10+1) ) ) + Ref(long,-1);
Graph0 = x2:

As you can see the IV tends to reflect this particular measure of IV and my reasoning is that the Aussie MM's must be using something similar. I think this is important because one of the best ways of trading options, is by making volatility forecasts based on SV/IV history. So we need to be playing in the same ballpark as the MM's. There will be no perfect measure of SV because of what SV is; merely a statistic based on past information. But I think this is actually a better way of looking at SV anyway.

Thoughts?

2. ## Re: Statistical Volatility - A New Model Required

Wayne that is interesting - never thought that our MM's might be using a different formula I just had another look at your original OSH SV/IV charts where IV was lower than SV, but looking at your new chart at approx the same period of time, IV was actually a bit higher than SV. IV has continued to fall a little since that time, so I expect the paper trade is not doing too well atm, however, based on the newer charts you would not have taken the trade.

Looking seriously at getting Amibroker (did take the trial some time ago) - just a bit daunted at learning the programming language as well as learning the software. Perhaps you could post a BHP chart with your new formula and see how it compares with the one you posted for me a few days ago?

3. ## Re: Statistical Volatility - A New Model Required

Margaret,

With the OSH trade, one could still reasonably expect an increase in both IV and SV, based on the cycles evident in the chart. We shall see however.

BHP with new formula\/:

4. ## Re: Statistical Volatility - A New Model Required

Thanks for posting the chart, Wayne - your new formula is certainly a lot more realistic and vastly different to the first chart - well done to work it out!

5. ## Re: Statistical Volatility - A New Model Required

Hi,
Could someone convert the Exponential Statistical Volatility into metastock code please?
Denis

Originally Posted by wayneL
Most of what I learned about trading options was learned trading on the US exchanges. Not very many weeks ago, I decided to migrate my account back to the Australian Stock Exchange. The reasons for this were exclusively lifestyle based; I didn't want to work Friday nights anymore because I was missing too many good piss-ups. If I was going to be asleep half of Saturday, it is going to be for recovery, rather than because of nocturnal money making habits.

Anyway that's beside the point.

In the United states the standard measure of statistical volatility (SV) is a simple 20 or 30 day, annualised percentage volatility equation. This is expressed in Amibroker or Metastock formula lanuage like this:

(StDev(log(C/Ref(C,-1)),20) * sqrt(252))*100

You could plot this on a graph, along with the implied volatility (IV), and see very plainly that IV tended to reflect the 20-30 SV. I have underlined "tended to reflect", because SV and IV will never always match each other. SV is measured over past data, whereas IV is an estimate of future volatility. Both SV and IV oscillate in a quite broad range in those markets.

However upon applying my templates to the Aussie market, something was wrong, something was different. IV tended to stay in a much tighter range compared to the US. But the standard 20 day SV charted a much more extreme course than IV. IV was not reflecting this measure of SV at all.

It was clear to me that the Australian market makers were using a different measure, a much more sedate measure than the Americans. Simply lengthening the look-back period of the standard SV equation wasn't doing it, much to lagging and innacurate when current volatilities were changing. I played with exponential moving averages of SV, but that didn't really work either. Then I had this idea; how about an exponentially weighted SV equation? My idea was a longer term SV measure but weighted with the most recent SV, much like an exponential moving average.

This seemed to work! After playing with different parameters, I got an SV measure that IV seemed to be reflecting.

So herewith I tender my Exponential Statistical Volatility (In Amibroker AFL):

long = (StDev(log(C/Ref(C,-1)),100) * sqrt(252))*100;short = (StDev(log(C/Ref(C,-1)),10) * sqrt(252))*100;
x1 = short - Ref(long,-1);
x2 = (x1* (2 / (10+1) ) ) + Ref(long,-1);
Graph0 = x2:

As you can see the IV tends to reflect this particular measure of IV and my reasoning is that the Aussie MM's must be using something similar. I think this is important because one of the best ways of trading options, is by making volatility forecasts based on SV/IV history. So we need to be playing in the same ballpark as the MM's. There will be no perfect measure of SV because of what SV is; merely a statistic based on past information. But I think this is actually a better way of looking at SV anyway.

Thoughts?

6. ## Re: Statistical Volatility - A New Model Required

Too easy

Code:
```long:= (StDev(log(C/Ref(C,-1)),100) * sqrt(252))*100;
short:= (StDev(log(C/Ref(C,-1)),10) * sqrt(252))*100;
x1:= short - Ref(long,-1);
x2:= (x1* (2 / (10+1) ) ) + Ref(long,-1);
x2```
Cheers

7. ## Re: Statistical Volatility - A New Model Required

hi,
could u tell me what program u guys are using to get IV?
could i get IV through either etrade or commsec?

thanks.

8. ## Re: Statistical Volatility - A New Model Required

Originally Posted by wayneL
So herewith I tender my Exponential Statistical Volatility (In Amibroker AFL):

long = (StDev(log(C/Ref(C,-1)),100) * sqrt(252))*100;short = (StDev(log(C/Ref(C,-1)),10) * sqrt(252))*100;
x1 = short - Ref(long,-1);
x2 = (x1* (2 / (10+1) ) ) + Ref(long,-1);
Graph0 = x2:
Hi Wayne,

Having another try at Amibroker over the weekend as I would really like to look more closely at the Aussie IV/SV charts. However, when I paste your code into the formula editor, it says "syntax error 30" when I click on the verify syntax button.

I'm going through the tutorials now hoping that the mud will clear, but wondered it you can help point me in the right direction? Usually pick up software fairly quickly, but this one is proving more difficult!

Cheers,
Margaret.

9. ## Re: Statistical Volatility - A New Model Required

Originally Posted by sails
Hi Wayne,

Having another try at Amibroker over the weekend as I would really like to look more closely at the Aussie IV/SV charts. However, when I paste your code into the formula editor, it says "syntax error 30" when I click on the verify syntax button.

I'm going through the tutorials now hoping that the mud will clear, but wondered it you can help point me in the right direction? Usually pick up software fairly quickly, but this one is proving more difficult!

Cheers,
Margaret.
Margaret,

I have written some info here

Getting IV into AB

Good luck

10. ## Re: Statistical Volatility - A New Model Required

Originally Posted by LOVE_OPTION
hi,
could u tell me what program u guys are using to get IV?
could i get IV through either etrade or commsec?

thanks.
LO

I am subscribing to the data via www.premiumdata.net it is contained in the "indicators" Add-on. It costs an few extra buckks a month.

Cheers

11. ## Re: Statistical Volatility - A New Model Required

Originally Posted by wayneL
Margaret,

With the OSH trade, one could still reasonably expect an increase in both IV and SV, based on the cycles evident in the chart. We shall see however.
Why don't you buy volatility when the red line is well above the black one and sell it to when it's the other way around?

cheers,
Chemist

12. ## Re: Statistical Volatility - A New Model Required

Originally Posted by chemist
Why don't you buy volatility when the red line is well above the black one and sell it to when it's the other way around?

cheers,
Chemist
Because buying and selling volatility ain't that straight forward hombre'.

13. ## Re: Statistical Volatility - A New Model Required

Originally Posted by wayneL
Because buying and selling volatility ain't that straight forward hombre'.

chemist

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•
Aussie Stock Forums