|
PureBytes Links
Trading Reference Links
|
<<
Suj : Re: Look in the mirror
Date : 26/08/98 01:03:27
From: tmorge@xxxxxxxxxxxxxxx (Timothy Morge)
Folks were just passing out information. That doesn't call for personal
attacks
from the mad frog physics tester or the etched-silicone featured man. If you
have something to add, add it. If you want to take a swipe at someone because
your bored, go do it elsewhere, please.
>>
So, you do not understand to backtesting as I do (previous thread some weeks
ago), if I read between lines (easy!)
Well, why not...
Next step:
<<
Suj : Re: getting back on track....
Date : 26/08/98 04:20:11
From: tmorge@xxxxxxxxxxxxxxx (Timothy Morge)
<<
Thanks for some instructive-constructive thoughts. Can anyone give me a short
commentary that explains what this system and function[s] are doing or trying
to
do? [or if not, maybe someone can point to the original articles?
==============
Being code-stupid, it's hard for me visualize just what raw code is trying to
do].
==============
>>
So, you are not comfortable with programming, why not...
This explains why we cannot communicate.
We are living in different [Omega]worlds
The concerned code is below (from ZooKeeper )
> ------------------------------------------------------------
> {System: Linear Time}
> Inputs: TSFLen(12),BPlus(12),MaLen(10),Dsp(3);;
> Vars: TSF(0),LRV(0),Ma(0);
>
> TSF=TimeSeriesForecast(TsfLen,Bplus);
> LRV=LinearRegValue(C,Tsflen,0);
> Ma=Average(C,MaLen)[dsp];
>
> If LRV crosses above TSF then buy on close;
> If Close crosses below Ma then exitlong on close;
> If LRV crosses below TSF then sell on close;
> If Close crosses above Ma then exitshort on close;
> ------------------------------------------------------------
> {Indicator: Linear Time.....scaling is set to price}
> Inputs: TSFLen(12),BPlus(12);
>
> Value1=TimeSeriesForecast(TsfLen,Bplus);
> Value2=LinearRegValue(C,Tsflen,0);
>
> Plot1(Value1,"TSF");
> Plot2(Value2,"LRV");
> ------------------------------------------------------------
> {Indicator: Displace Ma Stp.....scaling is set to price}
> Inputs: MaLen(10),Dsp(3);
> Plot1(Average(C,MaLen)[dsp]);
>>
Next step:
The the answer to your question from ZooKeeper:
<<
Suj : Re: getting back on track....
Date : 26/08/98 04:45:35
From: rsos@xxxxxxxxxx (ZooKeeper)
No original articles. The indicators were originally generated on a night of
lackluster sleep.
The idea was to use the Linear regression lines and the
inverted version of it (TimeSeriesForecast)
==============================
to illuminated a change in
direction. Whenever the LRV crosses above the TSF, it's supposed to mean
that there's change in direction to the upside. And when the LRV crosses
below the TSf, it's supposed to head down. The problem is, this gets whacked
in trading ranges (as most trending stuff does). And then a displaced moving
average is used as a stop.
>>
Althought the system is a valid one (I have had the same idea , and tested it
years ago), the explanation is partially wrong:
============
ZooKeeper uses:
Value1=TimeSeriesForecast(TsfLen,Bplus);
Value2=LinearRegValue(C,Tsflen,0);
TSF and LRV are the very same code, with the difference of the projected bar
value.
LRV projects by mean of a linear regression to the current bar.
TSF in any case, its not an inverted version of LRV
TSF does the same, but Bplus bars in advance.
A quick look at the TSF code will show you the evidence:
*******************************************************************
Study : TimeSeriesForecast
Last Edit : 7/7/95
Provided By : Omega Research, Inc. (c) Copyright 1995
*******************************************************************
inputs: Length(numericSimple), BarsPlus(numericSimple);
TimeSeriesForecast = LinearRegValue(C, Length, BarsPlus);
So, you could write:
Value1=LinearRegValue(C,Tsflen,Bplus);
Value2=LinearRegValue(C,Tsflen,0);
And it's more understandable, even to EL code stupid people.
It becomes obvious that the system is a dual linear regression projection
crossover with the same lookback period (Tsflen), but not the same projection
horizon (Bplus bars and zero bar).
No need to ask after that.
FYI, building , testing trading systems that works historical and realtime ask
for a minimum knowledge of programming and logical sense like I show you as
an example.
If you cannot understand this , you cannot understand me, and your frog like
reference will not change anything to the fact that you will have some
difficulties to prove me wrong in any of my posts.
Posting verbose message on my "stupid backtesting theory" will never remove
the fact that if you do not succed in coding systems that works in real world
with TradeStation, you are missing the very best part of the software, instead
of battling for information on TS5 beta and supposed TS server performance and
the like that produces nothing excepted noise (to which I participate).
Regards,
Pierre Orphelin
www.froglegs.com/with_garlic.htm
|