|
PureBytes Links
Trading Reference Links
|
In a message dated 8/12/98 6:09:26 AM, rkurzon@xxxxxxxxxxxxx wrote:
<<Why don't some of you that are experienced in successful system
development outline the process that you use to develop a consistently
profitable system. I'm not looking for any secrets rather the general
process you use. >>
I like to start with the entry idea combined with a simple exit based on time
in the trade. Exit after 5 bars for example. I want to know that I have an
entry that is better than a coin flip so I am looking for winning percentage
better than 55%. Most entries consist of a setup and a trigger.
Once I find an entry that gets me started in the right direction most of the
time I then start working on the exits. I like to have a fixed dollar stop
so I always know what my worst case is likely to be. Then I add a trailing
stop that moves with the market and keeps losses to less than the worst case
most of the time. Next I add some kind of profit taking exit that will lock
in a minimum profit after some profit target has been reached. I'm still
trying to let the trade run if it wants to but I'm not giving it much room at
this point.
This procedure is for trend following systems and would have to be modified
somewhat for counter trend trading or very short term trading. In these
systems I would be more likely to just take a minimum profit without giving
any back or trying to let the profit run.
Keep the entries as simple as possible. The exits may have to be fairly
complex. If there are many variables the system is going to look good
historically but its not going to work in real time. Watch out for data bias
(like S&Ps and bonds have strong upward bias which means short side systems
that may be very good in the future will not show good historical results).
I find that I need multiple systems so I can cope with up, down and sideways
markets over various time frames. I tend to believe that the shorter the time
frame the more random the price action so tend to favor longer term systems.
Shorter term systems would be better in theory because of more samples faster
and possibility of smoother equity curve but randomness and costs make it
difficult for me to design systems that trade this way. Hope this info helps.
Chuck
traderclub.com
|