March 2006 Newsletter

In this issue:

I. NeuroShell Trader release 5.0 status

II. Converting to release 5.0

III. IBFeed / IBFeed Professional Released!

V. Release 5.0 features – Optimized Paper trading

VI. One way to stop this newsletter

*******************************************************

I. NeuroShell Trader release 5.0 status

We did indeed start shipping release 5.0 on March 15 as planned, and we are grateful for the overwhelmingly positive response to this release and even the update fees. We can count on one hand the number of people who expressed disapproval of these fees, which we tried very hard to keep as low as possible.

Although we have started shipping, we still have a long way to go to get them all out. If you haven’t gotten yours yet, don’t start worrying until at least our next newsletter, so please don’t call and ask about yours until mid April! We continue to ship them in the order they were sent to us.

Certainly there will be some bugs in release 5.0 that neither we nor our beta testers found. As you report them, and we reproduce and fix them, we will be preparing release 5.1. That release will be free of course to all who updated to release 5.0.

To learn all about Trader 5.0 and place your update order, please visit www.ward.net/trader50.asp

*******************************************************

II. Converting to release 5.0

If you aren’t worried about saving charts and data that you created in release 4.0, then you needn’t worry about “converting”. Just deactivate release 4.x, and then install release 5.0, activating with the new serial number and password we will send you.

If you have charts and data you want to move over to release 5.0, then you can read about it on www.ward.net. Look under “NeuroShell Trader Release News, Upgrade Information and Beta Tests” at the article called “Updating and Migrating from 4.x to 5.x”.

Basically, there are four methods. The easiest one, the fourth method, is a conversion program that we haven’t finished writing yet, but when we do you will be able to download it there. You can either wait for that, or try one of the other methods, which involve varying amounts of computer savvy.

Obviously we will help you if you get in a big jam, but you can imagine how busy we’ll get if everyone gets in a jam all around the same time. So we urge you to carefully read about the methods, and evaluate what needs to be done in light of your own abilities. If you are somebody who is unsure about copying data files – you are definitely one of the ones who should wait for the conversion program!

*******************************************************

III. IBFeed / IBFeed Professional Released!

Bowfort Technologies has just released a new data feed for Interactive Brokers. You can access up to 6 months of historical data and it works with NeuroShell 4 and NeuroShell 5. The data is free if you have an IB trading account, with the addition of a monthly fee for the datafeed interface from Bowfort Technologies (IBFeed). Another advantage is that you can analyze the same data you’re trading (removes the disparity between separate brokers and data providers). IBFeed can also be combined with the automated trading feature in NeuroShell 5.

Plus, order IBFeed by the 31st March, mention you saw this offer in the March newsletter and get your 2nd month free!

*******************************************************

IV. Editorial Comment by Steve Ward, CEO

1. I hope that all of you carefully study the new 5.0 examples. I did most of them myself, and put a couple of weeks’ effort into them, but that isn’t the reason I hope you will take them seriously. The real reason is that they methodically take you through the features of the new system that I think are important. They start out easy and then slowly build up your knowledge about how the re-tooled wizards work, especially the much different Prediction Wizard. They contain some pretty simple trading strategies and predictions, many of which I have used in my own trading. They show you how to make good systems without using all the features. For example, it may not immediately be obvious to you how the new Chart Page Calculations indicators will help you make portfolio models, including hedged ones, but the examples should clarify that. Some people have asked that we update the AI College with release 5.0 which we may sometime, but frankly, if you can learn the examples you won’t need further training, in my opinion.

2. It’s time to upgrade your computer. There are some great deals out there. I just got a dual core 2.8 gig desktop processor with a gig of RAM, 80 gig hard drive, 19 inch flat panel monitor and some other stuff, and it cost around $660 including shipping and sales tax. I got it from Dell, who is always advertising good deals, but I’m sure their competitors have great offers too. It has made a big difference in how well NeuroShell works, and with dual core you could be running two copies at once on the same machine, one trading and the other optimizing models. Or you could be running NeuroShell or the AI Trilogy on one processor, and doing something else on the other. The only downside of buying a cheap new computer these days – they are cheap because the companies get paid for pre-installing lots of trial versions of software on them, and free stuff you probably will not want. It took me a couple of hours to update the OS, find out what all this trial junk was, and get it off. Maybe you will want it, but if you’re like me you like a clean machine.

3. Live from the forum. Some recent articles on the forum were talking about using free data. We mentioned there and I’ll repeat that later versions of NeuroShell Trader read the daily stock data after you’ve downloaded from sites like Yahoo. So if you are only interested in a few stocks, that might be a great way to save a few bucks each month. Yahoo’s published restriction seems to only prohibit redistribution.

*******************************************************

V. Release 5.0 features – Optimized Paper trading

Most of us like to optimize in an earlier period and evaluate on a later period. Using the evaluation of new data (called out-of-sample data by statisticians), what you will wind up doing is repeatedly optimizing models until you find the one that works the best on the new data, i.e. the one that shows up best during the evaluation (out-of-sample period). This is sometimes called “data snooping” because your evaluation data is not really out-of-sample anymore. Nevertheless it is still the most effective method to arrive at a model which has the best likelihood of holding up as you trade with it into the future.

Therefore, we have automated this process of building models, then evaluating them, keeping the one that works best during evaluation. The automated method is invoked when you check the box called “Save optimization which performs best on later paper trading” in the Dates tab.

In the chart of our new Example 9 we have done just that, creating what we call in NeuroShell a paper trading period. It shows up in orange on the chart. We optimized on data before 2005, the goal being to find the best profit before 2005. However we kept the combination which worked best starting 2005, not the combination which worked best before 2005. In other words, each time the genetic algorithm found a new combination of variables, the profit was tested both for the period before 2005 and the period starting 2005. NeuroShell remembered the best profit starting 2005 and left us with that combination, even though it may not have been best before 2005.

The paper trading feature is designed to prevent optimization from overfitting, and finding a solution which works the best in the future.

Of course if you data snoop, statisticians will say that you still haven’t properly evaluated your model with real out of sample data. Of course statisticians usually assume normal distributions, and a number of other factors not present in market trading. However, if you want to build your model with paper trading and still satisfy the statistician in you, you can select both of the boxes:

“Save optimization which performs best on later paper trading”
“Start trading before last chart date” – that will need to be turned on – (Tools->Options->…).

That will enable saving the model which works best on paper trading, while still giving you a real out-of-sample period after the paper trading period. The disadvantage is that you have an old model, at least as old as the out-of-sample trading period. Given that the market is frequently changing, and we suspect the number of statisticians who got rich in the market is quite small, we suggest that you consider using the paper trading feature without the added out-of-sample period once you have a model type you like.

*******************************************************

VI. One way to stop this newsletter from coming

It is really easy. Just change your email address and don’t tell us.

*******************************************************

Was this article helpful?

Related Articles