|
PureBytes Links
Trading Reference Links
|
I've havent done any long term backtesting so I can only quote others.
O'Shaughnessy in "What Works on Wall Street" points out that you should
include a lot of time to include both bull and bear markets. He went back
to 1950 to develop his methodology. Joe Granville in his first book
"Strategy of Daily Stock Market Timing...", mentioned a stock that
periodically (I think it was about 6 years) would go up smartly and then
decline back to the deck and all the alleged gurus would start touting the
stock. Joe Granville had looked at a long term chart and he saw this
peculiar behavior of this stock and predicted otherwise; he called these
periodic upturns and downturns gravestones.
While we cant easily go back a very long time with many high tech stocks,
we should be careful about systems that seem to work in a rising market and
only have 5 to 10 years look back.
Lionel Issen
At 05:27 PM 2/18/98 -0600, Richard Estes wrote:
>Support and resistance points can be in place for years. In testing a
>system, I agree with Rick Mortellra that 5 years is a minimum of data. You
>want to see how your system reacts under the different cycles of the market
>and the stock. Remember you are testing how well a system stands up over
>time, not the stock.
>
>Someone that does not test a system well, will end up loosing because they
>have looked at an optimized view.
>
>I would look at the average win vs average loss, drawdowns, time in trade.
>over win/loss ratio.
>
>Richard Estes
>http://www.intop.net/~restes/
>
>-----Original Message-----
>From: Harley Meyer <meyer@xxxxxxxxxxx>
>To: rickyp@xxxxxxxxxx <rickyp@xxxxxxxxxx>
>Cc: metastock-list@xxxxxxxxxxxxx <metastock-list@xxxxxxxxxxxxx>
>Date: Wednesday, February 18, 1998 4:25 PM
>Subject: Re: Whats Hot??????????
>
>
>>As far as back testing, I feel the last year is most relevant. Anything
>>beyond that can be less meaningful, since you are trying to evaluate a
>>specific type of price pattern with your trading system. You also need
>>to be aware of price patterns that just don't work for your system. I
>>would also add that the biggest plus is a good win / loss ratio.
>>
>>Harley
>>
>>
>>
>
>Attachment Converted: "d:\eudora\attach\Richard T. Estes Jr.1.vcf"
>
|