Where We're Always Looking to the Future


 


Product Tour

Download

Register

Buy

Add-Ons

Tips and Tricks

Online Tutorials

User Reviews

Links

About Us

Current Issues

FAQ

Home



Understanding and Utilizing
ANNI’s Prediction System and Trade System Outputs

The purpose of this document is to allow users to better understand how to use ANNI’s main system outputs. By understanding how they are meant to be used, you can greatly increase the value of ANNI in your investment strategies. Taking a few minutes to read through this will be well worth your time and effort.

One of the first misconceptions about ANNI’s outputs is that they are meant to be taken literally. This is incorrect...they are more or less indicators for what a particular stock may do in the near future and should be treated as such. ANNI’s AI has the ability to find repeatable patterns in a stock very well but the key here is ‘repeatable patterns’. When ANNI produces an output it is assuming that all current conditions will continue to remain the same therefore the accuracy of the predictions would be fairly high if this was the case. Unfortunately as we all know in the stock market...nothing remains constant for long. Many factors can influence a stock that will change its current conditions either by a little or by a lot. Those users that are familiar with the efficient market theory realize that most of these factors are taken into account almost immediately as they become known. With ANNI training on historical data it cannot take into account current events that are just beginning to unfold. It can however begin to take into account the effect of these events once the change is recorded into the historical data file. The greater the change, the longer it will take for ANNI to adjust since sometimes huge events can take days to unfold (this can be seen when new data is downloaded and an abrupt change in the confidence level occurs before new training is performed - a red flag to use caution on ANNI’s predictions). The smaller the change the less impact it has for ANNI’s predictions obviously.

What this all means is that before proceeding with any investment move based on ANNI’s prediction systems, a user should verify what current conditions may affect the security of interest (i.e. look at current and recent news both for the security of interest as well as the industry and the market in general - ANNI provides online shortcuts to perform this operation very easily including pulling up current news on entire Portfolio Groups at a time in addition to the Quick Analysis tool showing current news at the bottom of the reports...so please take advantage of these features and use them wisely). If current conditions match with ANNI’s outputs an investor can proceed knowing their risk has been reduced as far as possible. If current conditions do not match with ANNI’s outputs an investor should reconsider their actions or proceed with caution.

In simple terms what if ANNI’s outputs show an Up move with a Buy signal yet the company of interest recently reported that the last 3 years of earnings would have to be restated because the company was improperly backdating stock options given to its officers (something that recently has happened with several companies), would you still follow ANNI’s recommendation and Buy into the company? Of course not! Having something like this come out of a company cannot be predicted by ANNI therefore its outputs will have no way of forewarning the investor (which in this case the results of this type of news will be bad and a possible Short rather then a Buy may be in order). ANNI will be able to take into account the effects the news has on the stock price as it moves from current events to historical data, but until then an investor still needs to take a few steps before hand to protect themselves as best as possible.

The second misconception is how ANNI is producing its outputs. Both the Prediction System and the Trade System utilize a population of solutions rather then a single solution (this is the Genetic Algorithm part). With the Prediction System the outputs are a weighted average of the best nets which allow it to produce more robust outputs and to be more stable when changes begin to take place (this is because one or more nets may already have identified the new patterns that are emerging and thus are taken into consideration when the weighted average is calculated). It is by no means perfect though which is why it is meant to be an indicator of the possible direction of the security rather then it being exactly what the security will do (i.e. check current news and fundamentals for confirmation). The Trade System is a little different in respects that it does not use a weighted average of the best trade signals but rather one single ‘best’ solution to current conditions. This is because the outputs of the Prediction System vs. the Trade System are significantly different and a weighted average of the Trade System could/would end up producing sub-optimal results rather then improved results (because some trade signals are conflicting whereas the Prediction Systems nets generally produce similar outputs to some degree). This means every time new data is introduced or further training of the Prediction System is done a new Trade Signal may be selected that the system feels is better suited for current conditions.

So what exactly does that mean and what affect does that have on the Trade Systems Outputs? When dealing with a constantly changing market and a Trade System whose hopes are to produce the best trades possible a conflict comes up. As many traders who have attempted to produce their own trades systems can attest to...optimizing a particular trade system for a particular security only works for so long if at all (since it is generally optimized via back testing which sometimes is no longer relevant to current conditions). The more the trader tweaks the system to produce better and better results, the quicker that trade system falls out when the security/market changes. Because ANNI’s Trade System works with up to 250 unique solutions simultaneously we have a way to beat this problem head on. This means the solutions can be optimized for various particular conditions and interchanged as the security/market changes. This provides a user with an optimal trade system that is fully dynamic to handling/reacting as quickly as possible to a change in the security by simply switching the trade signal to one that is better suited to current conditions.

This method has caused some confusion though when a user observes the Trade Systems output suddenly change from a Hold to a Wait without ever showing a Sell. What has occurred here is a change in the ‘best’ signal where the prior signal was currently in a trade and the new signal is not. This is not a catastrophic problem though and is actually the best way to deal with changing conditions since what has occurred is the Trade System identifying a change that required a different signal to deal with it. Before the change the other signal was indeed the best one to be used...but now with the change the newly selected signal is better. This is how the Trade System quickly reacts to changing conditions. What this means is that a user needs to properly interpret the signals when this occurs. It is actually fairly easy to do as long as you keep these rules in mind when observing the Trade Systems outputs. Action signals (i.e. Buy, Sell, Short and Cover) can be taken for face value and acted upon accordingly. Hold and Wait signals depend on your current position. If you are currently out of the security and a Hold is suddenly generated this means the same as a cautious Buy (or Short if the Hold is on a Short position). The reason why it is a cautious signal is because the Hold could be from an action that took place over a week ago and that same opportunity has diminished (i.e. risk has increased, reward has decreased). The same holds true for the Wait. If you are currently in a position with the security and suddenly a Wait signal is generated this means the same as a cautious Sell (or Cover if you are currently in a Short position). The cautious part means check out what is occurring with the security, see if any current news could be affecting the PPS (price per share). Maybe place some alerts on the stock to help notify you if indeed something does go sour assuming you cannot find anything at the moment to confirm or deny the signal.

As you can see, a little bit of footwork and research along with ANNI’s outputs can go a long way to improving your trading strategies. By properly matching up ANNI’s outputs to current conditions you can get confirmation of the outputs and react accordingly. Otherwise if you simply use ANNI’s outputs alone you can quickly get yourself into a lot of trouble since ANNI is no crystal ball...it is simply a very complex indicator of what possibly might happen.
 



Home | Download | Register | Buy | Current Issues | Contact Us | Join Associates | Help


Copyright © 2008 Neural Science