|
PureBytes Links
Trading Reference Links
|
I totally agree with the thoughts expresses in the post you referenced. I
have tried making systems out of the tradition indicators without much
success. They look good as indicators on charts but that is about it. They
all lag price by so far that they are almost useless for trading. And the
fact that so many people use them makes them useless anyway. Any remaining
advantage would be "arbitraged away". In fact, as Neal Hughes noted, it may
be more advantageous to fade such signals (e.g., do the opposite)
I think our brain sees what it wants to see when looking at charts. If you
look at enough indicators, you can eventually find several to confirm what
you intended to do anyway. Someone once said that indicators are designed
to sell books to people who think they can buy the "Holy Grail" for $175.
But that does not mean that you cannot program trading systems that work
well with TradeStation. You can program almost anything (subject to the 64K
limit), and good systems don't have to be complicated.
All trading systems that work depend upon some characteristic of a market
that is repeatable enough of the time to make a profit, considering trading
costs. It either has to be some fundamental factor, such as momentum, (the
tendency of price to keep going in the same direction for a while), or some
psychological factor, (such as prices bouncing off of support/resistance,
Fibonacci, or trend lines because people think they will and make it
self-fulfilling).
Some phenomenon, such as Elliott Waves and the position of the planets are
simply in the eyes of the beholder (IMHO).
Most multi-bar chart patterns (head & shoulders, cup & handle, flags,
etc.), result from psychological factors. They are tradable if you can
recognize them visually but computers are lousy at such things.
Most bar patterns are a manifestation of some underlying price movement
that occurs on a smaller time scale than that you are observing. For
example, the "hammer" candlestick pattern on a daily chart results from a
sharp "V" bottom on an intraday chart. An inside bar on a daily chart is
congestion on an intraday chart.
The TradeLab post you quoted mentioned a lot of statistical functions.
Statistics is valuable since you are basically playing the probabilities.
Statisticians have figured out dozens of tests for whether something you
think is happening is really happening. It is very useful in testing
systems. (I now wish I had paid better attention in my statistics classes.)
Backtesting is very useful in exposing the logic of a trading system to all
kinds of price patterns that you hadn't planned for when designing the
system.
I often use an analogy to describe this. You can build a computer program
to fly an airliner, the autopilot. It does a pretty good job and is
programmed to handle most, but not all, of the situations it will
encounter. You test the autopilot by letting it fly the plane though all
kinds of conditions. Then you analyze how it reacted to each condition.
Every noise or shudder is analyzed to death.
The fellow who has the system that works well on two stocks seems a little
like saying that my autopilot works well when flying over Nebraska on a
sunny day. There is a possibility that he has discovered some
characteristic that is a result of the behavior of the specialist on the
NYSE who trades the stock. I know a person who has figured out how IBM
trades and takes advantage of it.
But most people don't bother to even look at why each winning trade worked
and why each losing trade failed to see if their logic is working properly.
They just tune the parameters until the overall profit is OK. Never mind
that it all occurred on one big trade. Or that the equity curve gyrated up
and down and that their test data just happened to stop on a peak of the
equity curve.
Unfortunately, it is so easy to do an optimization run in TradeStation that
people can try hundreds of combinations quickly. If it were harder, they
would have to design their tests more carefully and would probably spend
more time analyzing the results.
In summary, all the indicators are probably there because all the trading
books talk about them and they look pretty good on charts. No doubt, lots
of people just buy TradeStation or SuperCharts to look at charts. But when
I design a trading system, I fall back to more fundamental statistical
properties of the price patterns. And they do not have to be very complex,
And they can easily be programmed in TradeStation. (But probably not in
SuperCharts without the PowerEditor.)
Bob Fulks
At 10:19 AM -0500 7/5/98, Robert W Cummings wrote:
>
>I put this partial post from another trading list I'm on. I decide to put
>this here to see what this list thinks of these remarks. Seems to me that
>there people out there that somehow believe they can take the gamble out of
>trading through more sophisticated trading methods. Maybe there are people
>out there taking our money trading against our indicators we use in TS. Has
>technical trading already moved to a higher level than TS can offer and
>when our indicators work and make us money are we really just considered
>lucky by some and doomed long term.
>
>Robert
>
>>From another list a partial post;
>
>Simple gambler functions like RSI, stochastic (which, incidentally, comes
>from the Greek "stokhazesthai," meaning "to guess at"), and the like will
>be difficult to find until we and various users write them. However, there
>will be no shortage whatsoever of truly valuable functions from many
>sources .
>
>No one makes money over the long term using popular indicators, except by
>chance. They work in back-testing only with optimization that matches them
>to what happened in the past. They work in the future only in fantasies.
>They are great for gambling. They are great for professionals trading
>against those who use them. They are worthless to anyone trading with them
>for income.
>
>As those who don't already know learn the truth of what I just said through
>the loss of hard-earned money, they will be ready to advance to
>professional tools that actually work. When they are, TradeLab and VB will
>support almost any advanced market analysis tool imaginable.<
|