[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [amibroker] Measure a stocks' volatility with AB ?



PureBytes Links

Trading Reference Links


all cheering aside....
 
Of course the Black Scholes model is basedon 
Gaussian distributions, thus the sqrt(t), not t^1/4 as your apparent 
typo indicates, extrapolates time variation.  If you only want tosee 
variation in stock volatility - forget all this stuff and simply plot Bollinger 
Bands and move on; perhaps even better look at the MA(ATR()) - certainly easier 
for most to consume and digest.
 
OTH, if you are working options and using the 
BS model, which actually does a remarkable job of modeling option pricing, it is 
very useful to use the numbers that the price makers use.  Comparing the 
20-day historical volatility to the implied volatility for options is enough to 
convince one that the BS is useful, albeit modified by the floor action - the 
like reason for Anthony's observation that options drop over the weekend - there 
is more risk for a weekend "event" than an overnight "event", hence the price 
makers build in risk premium.
 
 
I certainly know which I use and how and why I use 
it; hopefully, people using other's answers test them against a known or 
desired result to determine if they meet their needs, remember that "sometimes 
you DO get what you paid for."   :)
 
Cheers,
 
Richard
 
btw, Waz - I like you thought about collecting 
indicator values at buy/sell points. I have been muddling with that 
idea for sometime but never crystallized is as nicely as you did.  One 
problem is that one likely needs to define a universe, and there is always the 
risk of tuning the results.  Of course, that pitfall exists from any 
inverse modeling system whenever you extend the domains.  Still a great 
idea to refine understanding of which parameters are important.  

<BLOCKQUOTE 
>
----- Original Message ----- 
<DIV 
>From: 
Listes 
trading 
To: <A title=amibroker@xxxxxxxxxx 
href="">amibroker@xxxxxxxxxxxxxxx 
Sent: Monday, April 29, 2002 7:04 
AM
Subject: RE: [amibroker] Measure a 
stocks' volatility with AB ?

<FONT face=Verdana color=#800000 
size=2>The use of 260 or 365 for computing annualised volatility supposes that 
the stock follow a gaussian random walk, that is daily returns are 
scaling like time power 0.5 (t^1/2). In fact, the statistical laws followed by 
stocks are quite more complicated and the ones proposed by scientists (Levy 
stable laws, GARCH models, fractal brownian motion,...) are very difficult to 
use (most of them do not have analytical solutions). The problem of the 
gaussian random walk is that it undervalues the risk and that is a major 
problem for option valuation.
<FONT face=Verdana color=#800000 
size=2>This is to say that using 260 or 365 days to estimate volatilityis not 
a major issue. Imagine a stock scaling with a t^0.7 scaling law instead of 
0.5, the error on the estimation of the volatility is quite 
dramatic!
<FONT face=Verdana color=#800000 
size=2>Just stick to one of it, 260 or 365, to be able to compare stocks, 
keeping in mind that the underlying model is not very 
accurate.
<FONT face=Verdana color=#800000 
size=2> 
<FONT face=Verdana color=#800000 
size=2>Waz

<FONT face=Tahoma 
size=2>-----Original Message-----From: Anthony Faragasso 
[mailto:ajf1111@xxxx]Sent: lundi 29 avril 2002 
04:24To: amibroker@xxxxxxxxxxxxxxxSubject: Re: 
[amibroker] Measure a stocks' volatility with AB 
?Richard,Since I also deal alot with 
Options, Here is my reason for usingsqrt(365).Should you use 365 
( the number of calendar days ) or 256 ( the numberof trading days), I 
believe there is no absolutely correct answer. I use365 mainly for 
consistency. One of the key functions in an option'sprice is time (t). 
The time component is expressed in terms offractions of a year and is 
universally thought of as Sqrt(calendardays/365) as opposed to 
Sqrt(trading days / 256).This 365 day calendar effect can be best 
demonstrated by looking at theprice behavior of an option from Friday 
afternoon to Monday morning. Allother things being equal, the option's 
price will drop more from Fridayafternoon to Monday morning than it 
would from Thursday afternoon toFriday Morning.  That's because you 
have about 2 3/4 days of time decayof the weekend instead of the usual 
3/4 day of decay during the week.Just to be consistent with the way 
options behave in the real world,  Iuse 365, the number of calendar 
days. If you use 256, the options pricedecline would be smooth every 
trading day, no matter how many calendardays transpired between trading 
periods.There is an even more troubling problem when you use the 
number oftrading days, and that has to do with an assets expected 
rate-of-returnand risk over the course of a year. The formula for 
volatility is theannualized daily standard deviation. Recall that you 
calculatevolatility by multiplying the standard deviation figure bythe 
squareroot of the number of periods in a year. If you use 256( now) 
becausethat is the number of trading days in a year, that means you'll 
have touse some other number when the number of trading days per year 
ismodified.  That means adding trading days would increase 
volatility,eliminating trading days would decrease volatility. Trouble 
is, changingthe available trading periods has already happened several 
times, andthe real world results refute this notion.Prior to 
1952 the stock market traded on Saturdays. That meant thatthere were 
more than 300 trading days per year, instead of the current256. When the 
NYSE made the change in 1952, did market risk shrink ? No.Did market 
returns shrink? No. Would diminished returns even be alogical 
expectation? No. Because volatility is a measure of risk andreward, 
however, that is what should have happened based on a reductionin the 
number of available trading days. Not only are those 
conclusionsillogical, they're not supported by actual 
events.Another example is the proliferation of overseas and 24-hour 
trading.Based on the increasing number of trading periods, using 256 
dayspreviously would require that you use a larger number now for 
certainassets ( like stock index futures traded on Globex). If you used 
avolatility calculation model that used trading days in its 
calculation,you'd need to increase the number of trading days to the 
appropriateamount.  With the standard deviation component constant, 
volatilitywould have to increase due to the increased number of trading 
periods.Some might argue that volatility would remain constant,even 
with theincreasing and decreasing trading periods. If that is true and 
you usetrading days to annualize volatility, increasing the number of 
tradingdays would imply a smaller daily standard deviation figure. But 
doesincreasing your sample size ( which you are effectively doing 
byincreasing the number of trading days) imply a reduction in the 
standarddeviation? Of course not.It doesn't even make intuitive 
sense, and is not supported by empiricaldata.  Taken a step 
further, assuming that volatility remains constantwhile increasing the 
number of available trading days means thatinvestors' reactions to news 
and events should result in smaller pricemoves. For example, lets say 
that a stock goes from $10 to $12. Asmaller standard deviation of the 
price change means that the pricechange must be smaller. So the constant 
volatility premise means thatinvestors are currently valuing the stock 
at an artificially high priceof 12 instead of correctly valuing it at a 
lower price, simply becausethe markets are currently closed on 
weekends.  That is wrong. The stockis priced at 12 because the 
information about the stock that iscurrently available warrants a price 
of 12. Saying that volatiltiy wouldremain constant when the number of 
trading days changes implies that themarket is currently irrationaland 
inefficient due to the absence ofweekend trading.Since I believe 
neither 256 nor 365 is perfect , and because I wish tobe consistentwith 
the way options behave in the real world ( that is toemulate the time 
deterioration over the weekend) I use 365 days in myoption models and my 
volatility calculations. :-]]Best 
wishesAnthonyRichard Alford wrote:> Anthony, 
I suspect you should use sqrt(260), the common number of> bars/year, 
instead of sqrt(365).  The result will agree with published> 
results, for instance on McMillan's site. I took the liberty of> 
attaching the indicator I use if you have the nerve to open it with> 
all the virus running around.> Cheers, Richard ----- Original Message 
----->>      From: Anthony 
Faragasso>      To: 
amibroker@xxxxxxxxxxxxxxx>      Sent: 
Sunday, April 28, 2002 6:49 PM>      
Subject: Re: [amibroker] Measure a stocks' volatility 
with>      AB 
?>       Hello, 
Derek,>>      I don't know if this is 
what your are looking for, but this>      is 
what I>      use for 
volatility:>>      pds=20;//Set your 
time period>      Graph0 = 
StDev(log(C/Ref(C,-1)),pds)*sqrt(365)*100;>>      
Anthony>>>>      
dereklebrun wrote:>>      >  
Hi is there any technical analysis way to measure 
a>      stock's 
price>      > volatility in AB 
?>      > If yes, how 
?>      
>>      > 
Thanks,>      > 
Derek>      
>>      
>>      > Your use of Yahoo! Groups is 
subject to the Yahoo! Terms>      of 
Service.>>>      Your useof 
Yahoo! Groups is subject to the Yahoo! Terms 
of>      Service.>>> 
Your use of Yahoo! Groups is subject to the Yahoo! Terms of 
Service.Your use of Yahoo! Groups is subject to the <A 
href="">Yahoo! Terms of Service. 
Your use of Yahoo! Groups is subject to the <A 
href="">Yahoo! Terms of Service.