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

Re: Mechanical Daytrading



PureBytes Links

Trading Reference Links

At 11:37 AM 3/22/99 , you wrote:
>Over the last year or so, there have been many great posts to this 
>list on the subject of daytrading the S&P.  I think I speak for 
>many others on this list when I express my appreciation to all 
>the daytraders who have shared their thoughts here.  
>Let me bring up one point I am curious about with the hopes it might 
>start an interesting thread:  are there any successful S&P 
>daytraders who use a 100% mechanical system?  If I recall correctly,
>all of those who have described their techniques have indicated that
>they use some level of discretionary judgement.  Some, like Tom Cathey
>in his post of today, describe themselves as largely discretionary.
>Is the successful mechanical S&P daytrader just an oxymoron (like
>congressional ethics)? 

To get things started and since I missed most of the past year stuff, 
I'll make a reply. 

What tends to be termed as discretionary is usually just something that is 
hard to code up in TS. If you use the right tools, a lot of discretionary
stuff can be "algorithmised". Or it is data series that is hard to manage
or come by say like to news items.

If you look at what Tom is saying, he uses a technical indication of panic,
breakthru a level that results in a stops getting run, a sharp reversal, etc.
Then he has S/R levels gotten from various time frames. 
The first item probably could be fitted into the 64K code limit.
The 2nd item is probably not going to. Most would give up or find manual
validation easier.

Why is a chair a chair? You recognise it to be. Could you give a 
general description of it? Yes, but you know, some chairs have 4 legs,
some three, some 1 and some with no visible legs. So if you define it 
in terms of legs, you gonna be in trouble because in reality, you use
legs as a rule of thumb to identifying a chair, there are still other 
considerations like some means of support to the floor and some kind
of back rest.

The way most systematizing is done is by means of identifying what I 
defined to myself as the "lowest common factor" approach. To define
a factor in a way to achieve the lowest possible failure rate in
identification 
of the factor. Again this is bias of wanting to get many data points in that
more means greater statistical significance. Type 1 (assuming null is what 
you want) error bias and uses stops to manage the type 1 error. 

Discretionary traders are likely to those who find these errors proceeding 
from their ability to elucidate the factors they used unacceptable. System 
traders are likely those who find the errors proceeding from their defintions 
of the factors they use acceptable. It may be that system traders are or have 
found better ways to define the factors they used relative to discretionary
traders.

The element of discretionary trading that can't be model now or in the future
is probably be ESP.

The protocols used by Van Tharp in strategy identification is helpful as most
decision making processes are in the realm of the cognitive unconscious.

I had to take time off from trading for more than a year. I came early this
year
and I still hadn't realised the SP had been halved though I was aware of
the Emini. 
The stuff I worked on back then still apply today and I still need a
machine to 
crunch the numbers. Some system traders tell me after a while, they don't need
the machine, they can see the situation in a chart with maybe some technicals
to frame/scale the chart. They've gone full circle.

Efforts on improving on a system tends to be trying to find a parameter to
get more 
historical profits. Rather it should be attemping to define even better
existing definitions 
of market logic/axioms. A rough example of an axiom, fastest way to get a
rally is 
squeeze the weak shorts.