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

[EquisMetaStock Group] Re: Life is Too Short



PureBytes Links

Trading Reference Links

--- In equismetastock@xxxxxxxxxxxxxxx, pumrysh <no_reply@xxxx> wrote:
> kut2k2,
> 
> Just visited Jurik's site. Appears there have been a number of 
> changes lately. What I remembered was primarily the JMA.
> 
> As to your comment "Very simple tools seldom work consistently in an 
> ever-changing market", I disagree. Simple tools will work 
> consistently and over a wide variety of market conditions if you 
> know what you are looking at.

Let's say your simple tool is RSI. What's the best lookback period to 
use? You need to do post-optimization of your RSI trading strategy 
over your chosen market's historical record for at least 100 trades. 
If your system is good, you can trust the "optimal" parameter values 
for about 50 trades before you have to reoptimize to get "fresh" 
parameter values. Suddenly your very simple tool isn't looking so 
consistent any more.

The financial markets are very complex entities. Expecting to conquer 
them with very simple tools is optimistic, to put it mildly.

> You also mention "the whole point of 
> adaptive indicators is to account at least partially for market 
> changes". My question is this, how are you going to account for 
those 
> changes? Typically AMA's use simple tools such as a RSI or a Stoch 
to 
> adjust to market conditions. Some will tie to volatility or even 
> volume but in the end we're still talking about simple tools as the 
> basis for adapting.

I don't use RSI or Lane's "stochastics". There are better volatility 
indicators.

> As far as Kaufman's AMA, I agree.

The core of Kaufman's AMA is very intelligent. But I see trader after 
trader just copying the code without ever questioning whether it is 
even efficient as written. Just one example: Kaufman limits his AMA 
between 2 days and 30 days for reasons he never reveals. If it isn't 
clear why it should be there, it doesn't belong in a trading system. 
That's my motto. ;-)

> As to your idea for an AMA, it certainly sounds like a good idea but 
> again what do you plan to tie the changing market conditions to?

I follow trend. Trend is good, non-trend is bad. That's my primary 
change indicator.

> I tend to look at market conditions in terms of timeframes... 
> basically three.

Some of us have no choice but to stick to one timeframe. But I do look 
for systems that are robust over a range of timeframes.

> I also look at several indicators not just one.

Agreed.

> I want to know how prices are reacting to volume.
> Am I seeing a gain in volume and momentum?
> In my case a truly adaptive indicator would need to adjust 
> to more than one condition. The best way that I know of to 
accomplish 
> this is by use of a binary. Based on the binary score it becomes 
> relatively simple to use an adaptable indicator. If my binary score 
> is small then my adaptive is quick, if the binary score is high then 
> my adaptive would be slow. I really don't need to worry about the 20 
> variable limit. There are always ways to overcome Metastocks 
> limitations.

I did find a way to somewhat overcome the lack of dynamic time 
periods. It's a major PITA but very helpful for highlighting the flaws 
in my test indicators.

> 
> As far as DLL's, they can indeed be used to auto adjust lookback 
> periods. The ones that I have used will allow the indicator to use 
> different lookback periods based on the varying parameters that I 
> desire. This is not a canned AMA. I can use the DLL on any 
indicator.

Sorry, I misunderstood. I ask you the same question you asked me: what 
does your DLL tie the changing market conditions to?

> Finally, you mentioned that you have an AMA that's better than 
> anything in the public domain but you need to make it dynamic in 
> lookback in order for it to reach maximum power. I don't understand. 
> How can it be adaptive and not dynamic? Can you explain this?
> 
> Preston

A typical AMA is basically an EMA with a variable smoothing factor 
(equivalent to a variable lookback period for an SMA). The irony is 
that this simulation of a variable lookback period still depends on 
specification of a fixed lookback period for the AMA. Once again we 
have to determine the "best" lookback for an indicator, even though 
the indicator in this case is itself an auto-adjusting version of a 
simpler indicator. By properly auto-adjusting the lookback of an AMA, 
we can get a super-adaptive MA that will give real-time optimal 
filtering of a time series. The key word, and the catch, is 
"properly". :-)

Good trading,
kut2k2

> --- In equismetastock@xxxxxxxxxxxxxxx, kut2k2 <no_reply@xxxx> wrote:
> > --- In equismetastock@xxxxxxxxxxxxxxx, pumrysh <no_reply@xxxx> 
> wrote:
> > 
> > Very simple tools seldom work consistently in an ever-changing 
> market. 
> >  The whole point of adaptive indicators is to account at least 
> > partially for market changes. That's their potential for enhancing 
> a 
> > trading system.
> > 
> > A lot of people completely misunderstand what the power of, say, 
an 
> > adaptive moving average (AMA) is. If someone uses it as a 
crossover 
> > basis for raw price, that is exactly the wrong usage. But if you 
> > replace a fastEMA/slowEMA crossover system with an AMA/slowEMA 
> > crossover system, you almost certainly will have a faster system 
> with 
> > no tangible increase in whipsaws, as a rule. It all depends on the 
> > quality of the AMA (big hint: Kaufman's is great as a learning 
> tool, 
> > lousy as an actual trading tool).
> > 
> > What I am currently trying to achieve is a super-adaptive MA: an 
> AMA 
> > with a self-adjusting lookback period. That's why MS's lack of 
> dynamic 
> > time periods, or even a looping option, is such a major 
limitation.
> > 
> > I thought your "adaptive DLL" was a special tool for 
auto-adjusting 
> > lookbacks but it sounds like it's just a canned AMA. I already 
have 
> an 
> > AMA that's better than anything in the public domain (I guess that 
> > makes it "proprietary" <g>) but I need to make it dynamic in 
> lookback 
> > for it to reach maximum power. Thanks for your response.
> > 
> > Good trading,
> > kut2k2


------------------------ Yahoo! Groups Sponsor ---------------------~-->
Rent DVDs from home.
Over 14,500 titles. Free Shipping
& No Late Fees. Try Netflix for FREE!
http://us.click.yahoo.com/BVVfoB/hP.FAA/uetFAA/BefplB/TM
---------------------------------------------------------------------~->

To unsubscribe from this group, send an email to:
equismetastock-unsubscribe@xxxxxxxxxxxxxxx

 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/