In this issue:
I. Discount on ChaosHunter coming to an end
II. Editorial from Steve Ward, CEO
III. Holiday closure
IV. Summer hours
V. One way to stop this newsletter
***********************************************************
I. Discount on ChaosHunter ends soon.
When we announced our new ChaosHunter in early April we made it possible for existing users of NeuroShell Trader or the AI Trilogy to purchase it at a $200 discount. That discount has been in effect now for over five months and many of you have taken advantage of it. The discount will end, however, at the end of this month. If you are interested in the ChaosHunter now is the time to get it before the discount ends September 1. Also, the additional $100 discount if you are one of our VIPs (those who have in the past spent more then $4000 on Ward Systems Group software and services) will terminate at the same time.
The ChaosHunter is model building software developed for both scientific and trading applications. The primary design criterion for the software was to enable users to see the exact formulas that evolve. If neural nets are in the equations, they are simple combinations of neurons that can be implemented in most programming languages or trading scripts, e.g., TradeStation’s proprietary Easy Language(R). Of course there is a direct way to load ChaosHunter models into NeuroShell Trader/DayTrader Professional. We also provide an OCX to run these models in your programs, and an Excel add-in.
For scientific users, ChaosHunter is much like the NeuroShell Predictor, allowing you to build a formula to predict continuous variables like speed, concentration, elapsed time, sales, cost, etc. In the same vein, traders could predict change in price.
Traders can also build models with buy/sell trading signals that are not predicated on first predicting price or change in price. The signals are evolved based on a profit fitness function, and a number of fairly easy to program technical indicators are included.
You can learn more about ChaosHunter on www.chaoshunter.com, where you can read the entire electronic manual. There is a 30 day money back guarantee.
***********************************************************
II. Editorial from Steve Ward, CEO
1. ChaosHunter 1.1. There were a handful of minor bugs in the original ChaosHunter 1.0 which have been fixed in ChaosHunter 1.1. If you have the original version 1.0 (look at the Help -> About menu) you can receive a free update to 1.1. Just send your emailed request to
su*****@wa*********.com
(for faster service, include your serial number). We are now in the process of building release 2.0 with features some of you have requested.
2. Seminar? Marge and I are contemplating another NeuroShell Trader Professional seminar maybe in January or February. We like warm places in the winter. Last fall we did a seminar in Tucson at the request of West Coast users, but most attendees were not from the West Coast! So we’re thinking Melbourne Florida, again. Anyway, give us some feedback so we can gauge the possible attendance to help us with our decision. Here is info on previous seminars: http://www.neuroshell.com/seminar.asp
3. NeuroShell Trader Professional Objective functions. One of the most frustrating things to me about optimizing models in NeuroShell Trader Professional is the tendency of models to learn to make most of their money in long bull or bear markets, to the detriment of smaller opportunities other times. For example, if you are building US stock daily models, you will find that they concentrate on the 90s bull market and 2000 to 2002 bear market, because that is when the most money is to be made.
One way to avoid the problem is to start model building in 2003 when the markets looked a little more like they do so far in 2008. Using the paper trading feature helps even things out, too. However, there is another often overlooked way to reduce the overriding influence of specific historical periods. That is to change your choice of optimization objective functions. The most recommended one – profit times equity curve correlation – will help as it tries to build rising equity curves without big dips.
However, there are a couple of objective functions that involve dividing by the standard deviation of profit. The standard deviation will be smaller when profits have less variance. Dividing by a smaller standard deviation will produce larger values of the objective function, meaning the resulting strategy will make more consistent profits over time, albeit smaller profits. Try those objective functions – you might like what you find.
***********************************************************
III. Labor Day Holiday closure
Ward Systems Group will be closed Monday, September 1, to celebrate the US Labor Day Holiday.
***********************************************************
IV. Summer hours
Ward Systems Group may on many days close early at 4 to 4:30 pm Eastern Daylight time in order that our employees can enjoy more time outdoors this summer. Please try to conduct your business with us earlier in the day until we resume normal hours this fall. Because of summer vacations, our staffing may at times be lighter as well.
***********************************************************
V. One way to stop this newsletter
It is really easy. Just change your email address and don’t tell us.
***********************************************************