RE: [bestcharts] Re: BC versions 4.32/4.321

Thu, 20 Jan 2005 14:10:26 -0800

I believe what you are referring to is what is often called simulated trading, where you step through a file of quotes, simulating trading.
 
I can not see how you can hope to apply any stationary process and have any expectation of maximizing performance. Almost all processes are improved with a dynamic feedback and response system.
 
I don't think Jawed's suggestion would have as its goal or output, a system that would be a stationary solution to finding good entry/exit points.
 
Whether you test backward or forward on historical quotes, you can not predict future markets. Any conclusion you reach would only work similarly in a market that repeated past activity, and since you have no way of knowing if it is going to, you can only apply the system and observe the results. If it works, you have been rewarded by the coincidence of repetition, not any quality of the system.
 
Of course, this does not stop many of us from trying, but puts us close in activity to those who study the racing guides, hoping to pick a winner.
 

Rik Rasmussen
Best Charts Yahoo Group Moderator
http://groups.yahoo.com/group/bestcharts/

 
-----Original Message-----
From: Howard Larson [mailto:[EMAIL PROTECTED]
Sent: Thursday, January 20, 2005 2:23 PM
To: [email protected]
Subject: RE: [bestcharts] Re: BC versions 4.32/4.321

By direct support I mean an option to automatically download and split a single file at a specified point; then do back testing on the oldest half and forward testing on the later half.  I have seen Web ads for packages that highlight back testing but I don't know if they support BT in this way.
 
By "stationary parameters"  I am referring to the parameters of the stochastic process that might be used to describe the stock's price action, not to the parameter set stored in B-C.  I don't think any TA package offered makes any claims concerning this.
 
My own experience tells me it is easier to deal with uncertainty if the process involved is stationary.  What does stationarity mean?  By one definition, it means the measured result is independent measurement interval.
 
I, also, have observed this effect (also described by Jawed) when using multi-AIO.  But, as I have stated in earlier posts, it is uncertain how long this stationarity effect will persist.  This uncertainty notwithstanding, I believe Jawad's technique should sort this issue out on a case by case basis.
 
All the forgoing is academic and certainly does not construe an endorsement. It still holds, that the proof of the pudding is in the eating.
 
hl

Rik Rasmussen <[EMAIL PROTECTED]> wrote:
Howard,
 
When you say "B-C doesn't support that directly, so one must manually download historical quotes from Yahoo", how is that different from any other software that back tests?
 
Back testing inherently requires that historical quotes be downloaded at some time from some source.
 
While I agree that the ability to pick a set of "stationary parameters"  is "a desirable trait indeed", I do not anticipate that such a possibility will ever exist, and hence, I expend no effort in its pursuit.
 
The popular indicators such as MACD do have popular parameters that many consider "standard," However, even though they have an often successful ability to indicate good entry and exit points, they also often fail. Best Charts defaults to these parameters on the many indicators it includes.
 
Best Charts, unlike some software offered often at very high prices, makes no lofty, unrealistic claims.
 

Rik Rasmussen
Best Charts Yahoo Group Moderator
http://groups.yahoo.com/group/bestcharts/

-----Original Message-----
From: Howard Larson [mailto:[EMAIL PROTECTED]
Sent: Thursday, January 20, 2005 12:11 PM
To: [email protected]
Subject: RE: [bestcharts] Re: BC versions 4.32/4.321

This sounds like classic back testing/forward testing using historical quotes.  B-C doesn't support that directly, so one must manually download historical quotes from Yahoo; back testing and optimizing with a file having a date in the past (e.g., Aug 10) as the last quote, then forward testing with a file having a later date (e.g., the current date) as the last quote.
 
Consistent results for variable BT periods indicates to me that stock's price arises from a process with stationary or nearly stationary parameters; a desirable trait indeed.
 
Very good.
 
hl

Do you Yahoo!?
Yahoo! Search presents - Jib Jab's 'Second Term'
Yahoo! Groups Links

Reply via email to