PhD Tested

Strategy Testing

A Conversation with John Bollinger, Financial News Network Technical Analyst and Lou Mendelsohn, President Market Technologies



MarketWrap - September 5, 1986

John Bollinger:
This is MarketWrap for Friday the 5th of September. This is John Bollinger sitting in for Bill Griffith. Now we continue our look at computer software for market trading with a look at a black box versus the full disclosure approach offered by Lou Mendelsohn, President of Market Technologies. Welcome Louis.

Louis Mendelsohn:
Thank you, John.

John Bollinger:
Why don’t you start by telling us what Market Technologies is?

Louis Mendelsohn:
We market a software product called ProfitTaker Futures Trading software.

John Bollinger:
Okay, what is ProfitTaker? When you say Futures trading software, what does it do?

Louis Mendelsohn:
ProfitTaker affords the individual trader the opportunity to generate specific buy/sell signals based upon the unemotional, quantitative approach to trading whereby he can develop trading models that suit his own trading style and risk propensity and which can be configured for each individual commodity.

John Bollinger:
So, this is essentially an analysis package which you run on a specific commodity. Do you use historical data?

Louis Mendelsohn:
Yes, the package actually was the first to introduce the capability back in 1983 for historical testing of historical data and what is actually occurring is a model is developed through the use of historical testing and then that model is applied in real time trading to generate the daily update and the buy/sell signals for trading in real time.

John Bollinger:
Now when you say a model is developed I know that one of your key ideas is the contrast between the black box approach and the full disclosure approach. When you say a model is developed, what do you mean by that? Is it fully disclosed?

Louis Mendelsohn:
ProfitTaker’s models are fully disclosed. What that involves is a complete disclosure of what the technical indicators are that are built into the software, and more importantly a disclosure of what the actual logical relations are among the various indicators to one another so that when a specific buy/sell signal is generated the trader has full knowledge and understanding as to what made that signal get generated. It allows him to really have the confidence that is necessary to act on the signals and to be able to rely upon them confidently.

John Bollinger:
OK, now in contrast what is what you call the black box approach?

Louis Mendelsohn:
Well, the black box approach is a situation where the technical indicators may be partially described or not described at all. And the logic itself is not disclosed to the user. So that the user is more or less acting kind of like on blind faith.

John Bollinger:
So, in other words you have a program you give it some data and run it and it says buy or sell, but it won’t tell you why it said buy or sell.

Louis Mendelsohn:
That’s right.

John Bollinger:
And your system tells you completely.

Louis Mendelsohn:
Exactly.

John Bollinger:
What is the real advantage or disadvantages. Suppose if you had two systems that gave essentially similar signals; one was fully disclosed, one was a black box. What is the trade off there?

Louis Mendelsohn:
Well, the trade off is very simply. It’s not something that is always understood by traders when they are first getting involved in these kinds of packages but really what it gets down to is confidence. In the black box approach, or in what’s more recently been called the gray box approach to software where some of the indicators are disclosed, some are retained as proprietary, the trader is acting on those buy/sell signals without a real, full understanding as to what is making a signal occur. As a consequence, when markets get choppy or if the price activity starts to get volatile or there is an abrupt retracement, there is a tendency to become gun shy. There is a situation where you want to tend to override the signals and you lose confidence in the software and that really undermines the discipline that should be achieved through the use of such packages. With fully disclosed software, where the logic as well as the technical indicators are completely disclosed to the user, the user can really draw upon his own experience as a trader and be able to confidently act on the signals and to stick with them through thick and thin.

John Bollinger:
Your program, your approach is commonly what is called an optimizer; explain.

Louis Mendelsohn:
Well, ProfitTaker as I said initiated this feature in software, in futures trading software several years ago. The basic approach is to look at each individual commodity and custom tailor the specific models to the particular commodity that you are interested in trading. You are also able to customize the models to suit your own trading style so that depending upon the account capitalization, the risk propensity that the trader has, what his style is, he can really fine tune trading models to each individual commodity and basically realize maximum profitability from that.

John Bollinger:
OK, one of the things that has bothered a large number of traders for a long time is you finally work out a trading system and you have all the pieces and it only works for a very short amount of time and it starts to degrade very quickly thereafter. Does your system address that?

Louis Mendelsohn:
Yes, that was one of the key issues that I addressed when I first worked on the development of ProfitTaker. The issue really gets down to how many variables or how many technical indicators are in the software. There is a tendency on the part of software developers to think that more is better and the more indicators you put in or the more entry and exit rules you build into the software some how that is going to be more productive.

It is in so far as the historical simulations that are generated. They look very profitable. They look really excellent in terms of the kind of performance that you would expect that they would have in real time. Unfortunately, there is no follow through. The real answer is you want to keep the system as simple as possible. You want to use as few indicators as possible to basically get a tone of the market in the development of the model and that way the likelihood of the model decaying or degenerating is going to be lessened.

John Bollinger:
OK, thank you very much. It has been a pleasure having you.

Louis Mendelsohn:
Thank you.

John Bollinger:
We have been talking with Lou Mendelsohn.

 

 

FREE Support
& Training

VantagePoint Awards
& Software Reviews

Free Recent
Forecasts & Live Demo

86% Challenge
The Results Are In