|
PureBytes Links
Trading Reference Links
|
This dicussion about systems has evolved into two separate concepts. 1)
What is a system? 2) Can you make money without a "pure" system? My
definition of a system is a an inter-related group of objectively
defined parts that solve objectively defined problem(s). If the human
oberver of the system by definition cannot be objectively defined, then
he cannot be part of a system. On the other hand, a human can utilize a
system and make money with it. Martin Graffman
Bob Fulks wrote
>
> At 11:35 PM -0400 7/25/99, The Omega Man wrote:
>
> >Bob, I think you're missing it... The idea here is to inform newer
> >traders of the way the game's really played. The game is not "build a
> >system and trade it mechanically". You and I agree that that is
> >impossible. The game is "build some tools to help yourself then use
> >your market experience to help you trade". Don't we agree that that
> >is really the way the game is played?
> >
> >And if we agree on this, what should our advice be to newer traders?
> >If every method is truly discretionary at root, then what should
> >newer traders be working to develop?
>
> There is this impression that the world is black and white - that you
> must either be a discretionary trader who stares at indicators and
> decides what to do, or you must be a system trader who blindly lets
> your computer trade your account mechanically, presumably while you
> are enjoying yourself on the golf course.
>
> In fact, there are all shades of gray.
>
> > At one extreme, there is the total discretionary trader who
> watches the screen all day and make trades.
>
> > You can also set up indicators with alerts to "wake you up" when
> the market seems to be getting close to a point where you might want
> to make a discretionary trade. This can eliminate the need to watch
> the screen all the time and lets you read a book while you wait.
>
> > You can also design a system to tell you when you want to get
> into and out of a trade, but then when you get the signal, you can
> watch the market for a few seconds or minutes and pick a good entry
> point. If you want to go long and the market is in a short-term down
> trend, you can often make a few extra bucks by waiting a few moments
> until it starts to turn and use a market order. These fills are often
> better than using limit or stop orders. Or, if the market is already
> moving up rapidly, you might decide to pass on this one since you
> would probably get a lousy fill anyway. Backtesting assumes you get
> filled. In fast markets, you probably will not be.
>
> > You can have a "pretty good" simple trading system that does
> fine under "normal" conditions. When you get a signal, you can look
> at the indicators and see if things look normal. If so, go with the
> trade. Of course, you miss some good ones this way but eventually you
> begin to learn which market conditions are good for your system and
> which ones are bad. As you begin to learn this you can then go back
> and add code to help make the decision more mechanically. Most things
> in life are improved by "successive refinement".
>
> > Handling bad data is a similar situation. Some systems are more
> sensitive to bad ticks than others so, at a minimum, you need to look
> at the price chart to see if your system was triggered by a bad tick,
> far away from current prices. This can be trickier than it sounds.
> Most systems use some smoothing so the effects of bad data can last
> quite a while.
>
> > At the other extreme is the totally automatic system that can
> handle all market conditions, all kinds of bad data, and has no bugs
> in the code. Programming such a system would probably be beyond the
> capabilities of most TradeStation users. It would probably require
> many pages of code whereas a 90% to 95% solution might require only a
> few lines of code.
>
> Many people count on stops to get them out of bad situations that
> their system code cannot handle. But even this can also be tricky.
> For example, consider the following simple system:
>
> Ave1 = SomeAverage(Close, Len1);
> Ave2 = SomeAverage(Close, Len2);
>
> if Ave1 > Ave2 then Buy;
> if Ave1 < Ave2 then Sell;
>
> You find this works sometimes but not in all cases. So you decide to
> use the Money Management stop in TradeStation to get you out of bad
> trades. But notice that once you are out, the conditions for entry
> are still true so TradeStation gets you right back in.
>
> But you do not understand this so you do an optimization run on the
> limit of the Money Management stop and select the best value. With
> this value you can easily enter and be stopped out many times on what
> was supposed to be a single trade. In this simple case, what is
> happening might be obvious but in a complex system, you might not
> understand what is happening. So your system is telling you to get in
> and you can easily see that this would be stupid. But as a good
> system trader, you take every trade until you go broke.
>
> I hate to see people being told that they must either be a
> discretionary trader or design a system that they can blindly trade
> mechanically, no matter what happens. There are a lots of "pretty
> good" systems that are somewhere in between, and designing these (or
> "improving" a system they buy) is well within the capabilities of
> most people.
>
> Bob Fulks
|