Monday through Thursday
9 AM to 5:00 PM
Friday
9 AM to 4 PM
All times are EDT
|
|
|
Dear Ward Systems Group Product Owner:
Here's the latest from Ward Systems Group.
|
|
Using the Power User Features to Identify a Paper Trading Period
by Marge Sherald, CEO
|
In last month's newsletter, we talked about using the paper trading period to mitigate the effects of over optimizing a model. This month we're going to talk about using the Trader Power User's ability to batch process saved Trading Strategy templates and walk-forward optimization to identify a paper trading period of interest.
|
 | | This chart uses two different Trading Strategy Templates with different paper trading periods. |
|
In the chart above, we used the same trading rules, but in the upper Trading Strategy we chose a two month paper trading period. In the lower Trading Strategy, we chose a one month paper trading period. The profit in the trading period (out-of-sample) was the same: $88.00. Which one would I choose to use in the future? I would use the first model with the two month paper trading period because it was able to capture more up and down price movement than the Trading Strategy with only one month of paper trading.
So how do you implement using templates to decide on the correct size of the data set for paper trading? Simply create a Trading Strategy with your rule set, set the paper trading size and other Trading Strategy parameters and then save it as a template. (There's a button on the results screen that allows you to save the template.) Next copy the Trading Strategy on the chart (right click on the Trading Strategy displayed on the chart and select copy) paste it on the chart and the right click once again and select modify. Leave everything the same except change the amount of time for the paper trading period on the dates tab under Modify Trading Strategy Parameters. Save this template as well as any other variations you want to test.
If you have NeuroShell Trader Power User or NeuroShell DayTrader Power User, you can open a new chart with multiple chart pages, and when you insert a new Trading Strategy, you can select several different Trading Strategy templates that you had previously saved. The Trading will then start to optimize all of the Trading Strategy templates for all of the chart pages. When optimization is finished, the results will be displayed, and you can check the ones that you want to have graphed on the chart. As a final test, look for the Trading Strategies that produce trading signals on the peaks and valleys that occur during the paper trading and trading period to decide which models to use in the future.
|
Walkforward Testing
If you're using the Trading Strategy template method described above, you will see the results for one instance of optimization, paper trading, and trading data in each Trading Strategy. However, if you have enough historical data, you can view the results for multiple instances of those data sets by using the walkforward feature in the Trading Strategy Wizard. (Walkforward optimization is a feature of the Power User versions only.) With walkforward optimization, the Trader allows you to evaluate the paper trading and out-of-sample performance of a Trading Strategy that is reoptimized regularly on newer data.
First create your Trading Strategy rules, either in the Trading Strategy Wizard itself or by loading templates as described above, and then when you reach the Modify Trading Strategy Parameters button, go to the Advanced Tab and enter the number of walkforward tests you wish to run. The number of tests will depend on the amount of data in your chart and the sizes of the optimization, paper trading, and trading data sets. If you select two walkforwards with optimization, paper trading, and trading data sets, you'll see results for walkforward 2 (beginning with the oldest data) for optimization, paper trading, and trading data sets; followed by results for walkforward 1 for the optimization, paper trading, and trading data sets; and finally walkforward 0 which displays results for the optimization and paper trading sets only. If you decide to go forward with trading this model, it will be the model with trading rules and parameters found during walkforward 0 that will be used.
Just as we did with Trading Strategy templates, you can copy a Trading Strategy that includes walkforward tests and change the amount of data included in the paper trading period to compare results. You can also save Trading Strategies with walkforward tests as templates.
To learn more about walkforward testing, watch the video that you can access from the Trader's Help menu. Look in the Power User section at the bottom of the video list.
If you own a Power User version, you can download a copy of these charts from www.ward.net in the examples section.
If you haven't upgraded to the Power User versions of NeuroShell Trader, check out the sale prices below!
|
|
Using ChaosHunter Neural Nets in NeuroShell Trader
by Dr. Andrei Deviatov, ChaosHunter Developer
|
One of our customers wanted to implement a ChoasHunter model directly in NeuroShell Trader without calling the ChaosHunter formula with a .DLL indicator. The problem was the ChaosHunter formula included a neural net, and he wanted to know which of the neural network add-ons to use to get the same results.
The N2, N3 and N4 neural nets in ChaosHunter are so simplistic that there are really no counterparts for them in NeuroShell Trader. The closest (and yet very distant) would be Ward nets in the Neural Indicators add-on. However, N2, N3 and N4 can be reproduced using the Indicator Wizard in NeuroShell Trader. For example, n2 = tanh (a*b+c*d), where tanh is hyperbolic tangent and a, b, c, and d are arbitrary inputs (really anything - constants, prices, or other indicators). Tanh is not found as a built-in indicator in NeuroShell Trader, but again can be easily reproduced using the exponent indicator since by definition:
tanh(x)=(1-exp(-2*x)) / (1+exp(-2*x))
Correspondingly:
n3=tanh(a*b + c*d + e*f) n4=tanh(a*b + c*d + e*f + g*h)
|
 |
|
Tech Support for Old Releases
Ends January 1, 2015
|
Beginning January 1, 2015, our free technical support will only be available for the following releases of our software:
NeuroShell Trader (all versions): Release 6.4 or higher
ChaosHunter: Release 4
AI Trilogy: Release Release 3
NeuroShell Predictor: Release 3
NeuroShell Classifier: Release 3
NeuroShell Run-Time Server: Release 3
GeneHunter: Release 4
NeuroShell 2: Release 4.2
If you need to update to the current release, we're offering a 20% discount on all updates.
|
 |
NeuroShell Trader Power User Sale
Save 20% on upgrades to either the NeuroShell Power User or the DayTrader Power User.
Add the following to your trading toolbox:
Multiple Frequencies on the same chart
Position Sizing
Pyramiding/Scaling
Walkforward Optimization
Multiple Template Optimization and Backtesting (batch processing)
ChaosHunter
Save 20% on ChaosHunter
Sale ends Tuesday, September 30, 2014
Have questions? Not sure which
release of the software you own? Give us a call at 301.662.7950, send E-mail to sales@wardsystems.com or SKYPE us at wardsystems.
|
|
|
Fax: 301-663-9920
Skype (sales only) Wardsystems
|
|
|
|
|
Copyright © 2014 All Rights Reserved.
Trademarks: NeuroShell, NeuroShell Trader, ChaosHunter, GeneHunter, Ward Systems Group and the old lady logo are all registered trademarks of Ward Systems Group, Inc. AI Trilogy is also a trademark of Ward Systems Group, Inc.
|
|
|
|