|
PureBytes Links
Trading Reference Links
|
Thanks for the response. I was beginning to consider answering myself.
>From what you're saying, I would gather that you basically turn the process
around. You first find a group of indicators (I've read a few discussions
of the "best" indicators) that usually signal good trading opportunities.
Then you screen for stocks ( also with certain other characteristics, some
fundamental?) that are "ready" for a trade as per these indicators? Then
you apply reasonable stops?
Was this "system" derived from a statistical test of some kind from a large
group of indicators/stocks or by trial & error from various sources? Or
both? There might be two ways to approach this: an overall (impractical?)
statistical test of all stocks/indicators OR finding SOMETHING that works
and add this and that until you gradually see serious improvements.
In your experience with stocks, generally how much time passes before the
average stock changes characteristics? Or do you play the best candidates
for the greatest gain and move to the next and not worry too much about that?
Have you seriously backtested your "system" over various market conditions?
What was your general method of backtesting? Find something that works
and then improve it by adjusting rules, etc.?
My attempts at "continuous" type trading systems were first attempts (
involving a closed-end fund and indexes) and more or less "proved" for me
the reality of the 60-70%(?) "effectiveness" of most technical analysis (if
you're lucky).
I realize from limited experience, that 99%(?) of price series change
"characteristics" over time, which is the other side of the coin from the
60-70%(?) effectiveness of TA.
Along these lines, I recently sent for OmniTrader and expect to use it
after fundamental screening with Telescan or with "recommended" stocks from
various sources like Worden. This gives me ( I hope) the best realworld
chance of getting started trading stocks (I need all the help I can get, I
suspect). Does this plan sound reasonable to you? It was either this or
buy a good many trading systems for use with Metastock. I figured the top
down & guided approach of Omnitrader would allow me to see the bigger
picture and make the finding & testing of various trading methods easier.
Also allows some diversification in types of systems and methods.
I currently have sector/ closed-end systems that I assume have a good
probability of doing better than the market in any given year. I sense
that any attempt at a higher performance level means sweating the details
and going all out for the maximum probability of doing as well as possible
while controlling risk carefully. In the same vein, I sense that the
greater the performance expected, the less & less the probabilty of
actually having things happen that way. In other words, do it right or
don't do it.
Seems the process has 2 stages; find something, whether it's a higher
volatility fund, sector fund rotation, or individual stocks/futures that
have the potential to beat the market AND then find a way to harness this
potential with a valid trading system. The term "valid" seems to be the
issue and it has to do with the overall statistical probability of a
particular system performing as expected; this is the real issue?
Do I know whereof I speak?
At 10:40 PM 1/14/98 -0600, you wrote:
>Hi to all,
>
>I am responding to the following:
>
>-----Original Message-----
>From: Rick Roegner <oiis@xxxxxxxxxx>
>To: metastock-list@xxxxxxxxx <metastock-list@xxxxxxxxx>
>Date: Monday, January 12, 1998 12:58 PM
>Subject: Systems-indicators
>
>
>>I've noticed that whenever I design & optimize a trading system (or the
>>same for an indicator) in Metastock, that in various areas over the data
>>history, the system or indicator does not work so well.
>>
>>Has anyone devised a method to adjust a system or indicator over the price
>>history during these "weak" spots; let alone possibly into the future
>price
>>history?
>
>I agree that it is difficult to built an indicator/system with enough
>flexibility
>to perform well over a large number of periods for most/all stocks .
>
>I have adopted an alternative that works well for me.
>I use a system of a combination of indicators
>that is very predictive when the signal is given for a stock. However, it
>does
>not attempt to continuously predict price action of any particular stock.
>Rather, I have carefully (I hope) selected a group of stocks that have
>behaviors
>for which my indicators are effective. I only trade when my indicators
>signify a hit,
>and then only for a limited number of periods. The advantage to this
>method is
>that my indicators due not have to continuously predict the behavior of any
>one stock
>for a significant number of periods, and due not have to predict future
>price changes for all stocks.
>
>I believe one successful technique is to limit a system to a predetermined
>set of stocks for which the system is most effective and limit the duration
>of the trades
>to that time period for the system has proven successful. That is, the
>system is
>not "pushed" beyond that for which it has proven successful.
>
>
>
R. Roegner
|