December 11, 2017
IN THIS ISSUE
HOW TO JUDGE A WINNING PREDICTION
VOTES NEEDED FOR 2018 READERS’ CHOICE AWARD
HOW TO JUDGE A WINNING PREDICTION
Neural Nets Are Powerful Tools
Sometimes NeuroShell Trader users ignore the prediction feature of the program because they don’t understand it and prefer to use trading rules that they understand. However, when you don’t know the rules that model a security, neural nets are the best gadget in the toolkit. Neural nets can capture more patterns from your indicators than your brain can process. Neural nets don’t get tired or bored so you miss a trading signal.
In fact neural nets are so good at learning patterns in the training set, a good selection of inputs will show trading signals hitting all of the peaks and valleys. Before you pick out your yacht or plan to spend your time permanently in a sunny location, you might want to learn some techniques for judging a net that is likely to be successful beyond the training data.
The predictions in this chart use indicators from the Price Momentum category. The training objective is to Maximize Return on Account * Equity Curve Correlation in order to make profits while minimizing drawdown. The model optimized indicator parameters.
The chart above displays trading signals for both the training (gray) and out-of-sample period (green).
There are several different factors that suggest this prediction will continue to be successful.
1. The trading signals appear at peaks and valleys in both the training (optimization) period and (most importantly) during the out-of-sample trading period.
2. The model captures both short term and long term moves.
3. The out-of-sample equity is not based on one or two “lucky” trades.
4. The equity curve for both data sets does not display any major drawdowns.
A few final thoughts. Very few models last forever because market conditions change. Neural nets have to be retrained either by using the “Retrain” button in the Prediction wizard or using using a Trader add-on such as Adaptive TurboProp2 which allows you to specify how often to retrain the net.
If there is a significant shift in the market, you might have to find new indicators or even switch to a different security.
Remember, your overall objective is to make money, not trade your favorite stock.
Click here to download the chart.
______________________________________________________________________________________________________________________
VOTES NEEDED FOR 2018 READERS’ CHOICE AWARD
We know your previous votes have put us over the top in the race to be named the Best Artificial Intelligence software in the Technical Analysis of STOCKS & COMMODITIES (TASC) magazine’s Readers’ Choice Awards. We’re asking for your support for the 2018 race.
The TASC readers are very conscious of the award winners and your vote drives new sales, which helps fund the technical support we offer for FREE. It’s a win/win arrangement for everyone.
How to Vote
1. On the page that is displayed, enter your subscriber ID from your magazine address label (look at their example to see which numbers on the label are the subscriber ID).
After you are logged in, look for the Artificial Intelligence Software (Expert, Neural) category on page 3. Be sure to mark your ballot for NeuroShell Trader Professional (Ward Systems Group Inc.) as number 1.
If you have also used GeneHunter, please vote for this software as number 2 in the same category.
2. Do the same for the Standalone Analytical Software, $1000 and more category. Look for NeuroShell Trader Professional on page 2.
3. In the Software Plug-Ins section on page 3, vote for MESA for NeuroShell Trader (MESA Software).
Note that if you switch pages, sometimes your votes are lost so be sure that all of your votes are marked before you cast your final ballot.
We’re all busy so we ask that you vote now so you won’t forget later. Voting closes December 31, 2017.
The winners will be announced in the 2018 Bonus issue of STOCKS & COMMODITIES Bonus issue.
A big thank you from Ward Systems Group for voting!