Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Strategies & Market Trends : Befriend the Trend Trading -- Ignore unavailable to you. Want to Upgrade?

To: jesso who wrote (8196)	6/10/2000 5:00:00 PM
From: Ken Adams	Read Replies (1) \| Respond to of 39683

jesso... ...your method of just looking at recent charts to decide which market(s) to trade the MAX system on is as useful and less of a hassle than trying to do serious backtesting. You summed up my feelings exactly. I really couldn't care less how well a system did 5 years ago or even 5 weeks ago. I only want a chart that is throwing out some tradeable moves in the very recent past. A couple of weeks, maybe 3. Beyond that, I lose interest. Ken

To: jesso who wrote (8196)	6/10/2000 5:55:00 PM
From: Dan Duchardt	Read Replies (1) \| Respond to of 39683

jeso, While the future will never repeat the past exactly, if you determine these and other statistics for a system after backtesting the past 10 years of data which generated thousands of trades, there's a good chance the results will be somewhat similar in the future as long as the market you're testing hasn't changed dramatically. I think it's fair to say that in any 10 year period of time the market is likely to go through several dramatic changes. Testing that far back, and looking only at accumulated performance will likely wash out times of great opportunity that may have existed in a much smaller time frame. As a case in point, the BTTT-MAX applied to a few well chosen stocks proved to be a very effective system for a period of a few weeks (maybe longer; before I came across this thread). Yet several recent messages can be characterized as being of the "Hey, this isn't working so good anymore" variety. I don't think we need to look back years to see a dramatic change in market conditions. Daily charts of the Dow, Nasdaq, and S&P 500 all the tell the same story. The market entered a fairly tight consolidation phase this last week after several weeks of oscillating between prominent highs and lows separated by several days. This leads me to jump to a hypothesis that can only be tested by looking at relatively short sample periods, but all within the last several months. I wonder, for example how well (or really not so well) BTTT-MAX did in the flat market of September 1999 (poorly would be my guess), compared to comparable time periods in the subsequent era from the low in October until now. Do 2-week or 1-month periods from October 1999 through March 2000 yield comparable results to one another, and how do they compare to the recent good times? In other words, I'm wondering if we can just accept this past week as an improbable flat period in an otherwise oscillating market that might very well return next week and bring back the good old BTTT-MAX performance with it. If so, the lesson might be that when the daily and weekly ranges compress as they have, put the strategy on hold until the market resumes it's old ways and stop wasting money on the whipsaws. That will probably cause one to miss the first one or two good moves when and if they return, but at the rates of return BTTT-MAX has provided, the little bit of lost opportunity might be well worth the savings. Dan

To: jesso who wrote (8196)	6/10/2000 7:42:00 PM
From: booters	Read Replies (2) \| Respond to of 39683

I have a general question about backtesting please. Some 20 years ago I started in this business developing trading systems for the futures market. In those days if you did that you did it from scratch. Wrote your own code and all, no canned programs. Alot of the data we used was also collected inhouse. Because of all this one was very familiar with exactly what the program was doing and exactly what the data looked like ( we had our own programs to look at and clean the data). I re-entered the fray a short while back and pulled out some of my old stuff and went to work again, thinking it would be much easier now because of all the historical data you can get easily. After some research into who was widely used and respected for providing data I ordered some (Futures data). After a few days I realized something was not right and looked very closely at how they compiled it and rolled from one contract to the next. The process was horribly flawed and after pressing them they finally admitted it. Bottom line is you can not use the data to properly roll a position from one contract to the next and this is a major provider of historical data. There are many people out there backtesting programs and getting flawed results without knowing it. I do not mention this to scare anybody and I realize futures data and stock data (most of you use stock I think) are very different but it does lead to the question I have. If I am correct (my experience is with futures) when a stock splits say 2 for 1, then the past historical data is divided in 2 so the transition is smooth. This being the case then any fixed number used in the formula for testing that was not aware of the adjustment would return false data. Example, buy if price goes 1/2 point above previous days high. So finally the question. Do the canned programs many of you use take this into account? boots