Adaptive Net Indicators

Overview

The way a human would typically do pattern matching on financial data is as follows: He or she would scan the price stream (close) looking for distinct movements. Then he or she would examine the pattern formed by the changes in close for some number of bars (let's say 11) just prior to the distinct movements. If this human is pretty good at pattern recognition, he or she might even examine the changes in the open, high, and low of the preceding 11 bars as well as the change in close in those bars. The human is looking for what types of patterns in the prior 11 bars that foretell the distinct movement.

Once the human is satisfied that there is a high probability of the distinct movement following certain patterns, he or she can then watch for those patterns in the future. When the patterns appear, the human expects one of the distinct movements to follow and places orders appropriately.

The example above is the inspiration behind the enhancements to Adaptive Net Indicators (ANI) release 2.0. ANI always did pattern matching, but now in release 2.0 new functions not only include the current bar in the inputs but a number of lags of the input as well.

Technical Details

Adaptive Net Indicators are special versions of GRNN and PNN neural nets formulated to do pattern matching, both predicting and classifying. They retrain (quickly) on every new bar, so they are never more than 1 bar behind. You can set the contribution factors yourself so that the net uses your specification of how important the inputs are, instead of the other way around. Of course, you could also let the genetic optimizer find them if you own the NeuroShell Trader Professional or the DayTrader Professional. As a matter of fact, you can also let the GA find the optimal number of bars ahead to predict and the optimal training set size!! If that's not enough exclamation points, how about this: you can even let the GA find the best thing to predict by optimizing the parameters (including whether open, high, low, or close) of your output indicator!!! Adaptive Net Indicators is a package with unprecedented flexibility and capabilities!!!!

Many of you have expressed a desire to have confidence factors for your nets. The classification series of Adaptive Net Indicators will provide you with confidence. In addition, Adaptive Net Indicators will make no prediction at all if they feel they have no basis on which to do so.

Our Adaptive Net Indicators do pattern matching by comparing each new pattern encountered with a number of immediately previous known patterns. They do not use weights like most neural nets. The Net "output" (i.e., the output of the indicator) is derived from the outputs of the immediately previous patterns. It is most heavily influenced by the most closely matching of the known patterns, and so the output of any new pattern encountered will be much like similar known patterns. You the user can set the number of immediately previous patterns which the Net compares to the new pattern.

The Net indicator has "inputs" in which the pattern is stored, just like other indicators and neural nets. If a Net has 3 inputs, you can feed in the current values of the RSI, a CCI, and the Momentum to form the pattern, for example. Or you could feed the Net today's close, yesterday's close (lag 1 of close), and the close the day before (lag 2 of close). You can feed Net outputs into a Net, just as you could do with any other indicator. With some of our indicators, you can also input many "lags" of the primary inputs as well.

Nets also have another type of input, called the "Actual" value. This is where you show the Net what you want the outputs to be like. In other words, you "train" the Nets to produce values like the Actual value whenever the corresponding inputs are closely matched. You want the Net to predict for you the Actual value X bars in advance. The output of the Net is the prediction of the Actual value X bars ahead. You get to pick what X is for each Net. This is just like our neural nets in the NeuroShell Trader. Furthermore, you can even optimize the value of X in a Trading Strategy.

There is another big difference between our Adaptive Nets and the neural nets in the NeuroShell Trader. The contribution factors for each input are also inputs. That's right, instead of the net telling you the contribution of each input variable, you get to tell the Net what the contribution should be. The higher the contribution, the more heavily the Net will weigh that input when it does pattern matching. Of course, if you'd rather have the contribution factors figured out for you, the NeuroShell Trader Professional can optimize them.

There are two kinds of Adaptive Nets, depending on the type of output they produce. There are "Prediction Nets" whose outputs are predicted values (like price change, percent change in price, predicted indicator values, etc.). There are also "Classifier Nets" whose output is a probability of the pattern being of one type or another. Types might be "Buy" and "Hold", for example. Other types can be "Good" and "Bad", or "Up" and "Down", etc.

The Classifier Nets don't actually read or produce the strings like "Buy" and "Hold". You use positive numbers in the Actual for one category like "Buy", and zero or negative numbers for the other category like "Sell". The predicted output will be a number between -1 (strong probability of sell) and 1 (strong probability of buy). Numbers close to zero could be considered "Hold" (YOU would decide how close to zero a prediction should be to be considered a "Hold")

Example

One of the interesting things you can do with Adaptive Net Indicators is build adaptive moving averages. You can adjust how tight or how loose the adaptive moving average is. The chart below shows an adaptive net indicator configured as an adaptive moving average.

Configurations

There are a total of 18 Adaptive Nets, nine Predictor Nets and nine Classifier Nets. There are 9 of each because each of the nine takes a different number of inputs as follows:

• Predict2 - Prediction Net which takes 2 inputs
• Predict3 - Prediction Net which takes 3 inputs
• Predict4 - Prediction Net which takes 4 inputs
• Predict5 - Prediction Net which takes 5 inputs
• Predict6 - Prediction Net which takes 6 inputs
• LagPredict1 - Prediction Net which takes many inputs: 1 primary input and any number of "lags" of that primary input.
• LagPredict2 - Prediction Net which takes many inputs: 2 primary inputs and any number of "lags" of those primary inputs.
• LagPredict3 - Prediction Net which takes many inputs: 3 primary inputs and any number of "lags" of those primary inputs.
• LagPredict4 - Prediction Net which takes many inputs: 4 primary inputs and any number of "lags" of those primary inputs.
• Classify2 - Classifier Net which takes 2 inputs
• Classify3 - Classifier Net which takes 3 inputs
• Classify4 - Classifier Net which takes 4 inputs
• Classify5 - Classifier Net which takes 5 inputs
• Classify6 - Classifier Net which takes 6 inputs
• LagClassify1 - Classifier Net which takes many inputs: 1 primary input and any number of "lags" of that primary input.
• LagClassify2 - Classifier Net which takes many inputs: 2 primary inputs and any number of "lags" of those primary inputs.
• LagClassify3 - Classifier Net which takes many inputs: 3 primary inputs and any number of "lags" of those primary inputs.
• LagClassify4 - Classifier Net which takes many inputs: 4 primary inputs and any number of "lags" of those primary inputs.

For more details, please view the product manual for this add-on.

Adaptive TurboProp2

Many of our customers have asked for a version of our famous TurboProp2 that retrains itself as often as every bar. Well here it is, formulated as an indicator! Just insert it into your chart or trading strategy just like any other indicator. Never has a neural net been easier to use, yet so powerful!

Overview

Adaptive TurboProp2 (AT2) is the same basic neural network that is used in the NeuroShell Trader, Trader Professional, and NeuroShell DayTrader Professional Prediction Wizard. However, AT2 can retrain itself as frequently as every new bar or after a number of bars which you can specify. The training set is always the most recent bars. AT2 does not make predictions on the training set bars, and its predictions are always "out of sample" (exception: when you specify that you are predicting zero bars ahead). Therefore, there are no walk forward periods to worry about, since the net is essentially walking forward each day (or each bar in the case of the NeuroShell DayTrader Pro).

An AT2 net may be used as an indicator, as an input to a regular TurboProp2 net in the Prediction Wizard, or it may be used directly in a Trading Strategy. The output is a indicator, meaning that the prediction is shown on the current bar, even if it is a prediction for 10 bars from now.

AT2 doesn't require the optimizer in the Professional versions of the Trader because the parameters are so easy to set. However, when used with the NeuroShell Trader Professional or NeuroShell DayTrader Professional, you can optimize the following parameters:

• 1. trainbars - the training set size (number of recent bars on which to build the model), up to a maximum of 10,000.
• 2. aheadbars - the lookahead period (the number of bars into the future the prediction is to be made), i.e., the number of bars ahead that the actual is predicted. Be careful not to allow 0 in the range, because then you are predicting the current bar, not the future.
• 3. actual - the parameters of any indicator used as an output.
• 4. hiddens - the number of hidden neurons to use when making a network (Turboprop2 in the prediction wizard automatically decides how many to use, but AT2 lets you or the optimizer decide). You can use up to a maximum of 20 hidden neurons, more than enough for this adaptive version.
• 5. retrain - this parameter specifies in bars how often the net is trained. If retrain = 1, a new net is created for each new bar encountered. If retrain = 10, the net is retrained every 10 bars.
• 6. input1, input2, etc. - these parameters are the network inputs. You can optimize all parameters of indicators used as these inputs.

Available Configurations

AT2 nets allow 2 to 14 inputs. The names of the indicators are correspondingly Tprop2, Tprop3, …, Tprop14.

Adaptive TurboProp2 requires release 3.2 or better of the NeuroShell Trader, NeuroShell Trader Professional, or NeuroShell DayTrader Professional.

For more details, please view the product manual for this add-on.

Advanced Indicator Set 1

The NeuroShell Trader® Professional has around 800 technical indicators. However, that didn't stop our users from wanting more! They sent us requests for more indicators, some of which were simple to build, and some of which required a great deal of research. So we built another add-on called the Advanced Indicator Set 1 with these requested indicators.

These are not part of the NeuroShell Trader®, NeuroShell Trader® Professional, or NeuroShell® DayTrader Professional, but you can buy them as an optional add-on if you want them. They require release 3.0 or above of the NeuroShell Trader®, NeuroShell Trader® Professional, or NeuroShell® DayTrader Professional.

Click here for a brief overview of the indicators in Advanced Indicator Set 1, which are all listed below:

TECHNICAL INDICATORS

• Aroon Down
• Aroon Up
• Aroon Oscillator
• Chaikin Accumulation/Distribution Oscillator
• Hodrick-Prescott Filter (untradable)
• Hodrick-Prescott Window
• Keltner Channels
• True Range
• Vidya
• Volume Adjusted Moving Average (VAMA) (untradable)
• ZigZag (untradable)

CHAOS AND FRACTAL INDICATORS

• Choppiness Index
• Polarized Fractal Efficiency
• Hurst Exponent
• Hurst Significance
• Fractal Dimension

STATISTICAL INDICATORS

• Polynomial Regression
• Polynomial Predict

ASTRONOMICAL INDICATORS

• Moon's Illuminated Fraction
• New Moon
• Moon's First Quarter
• Full Moon
• Moon's Last Quarter

DATE INDICATORS

• First Monday of the Month Flag
• First business day of the Month Flag
• Last business day of the Month Flag
• Number of days since High occurred
• Number of days since Low occurred
• Week of the Month
• Week of the Year

MATH INDICATORS

• Ceiling
• Cluster
• Constant Value
• Floor
• Integer
• Percent
• Round

For more details, please view the product manual for this add-on.

Advanced Indicator Set 2

The NeuroShell Trader® Professional has around 800 technical indicators. However, that didn't stop our users from wanting more! They sent us requests for more indicators, some of which were simple to build, and some of which required a great deal of research. So we built another add-on called the Advanced Indicator Set 1 with these requested indicators.

These are not part of the NeuroShell Trader®, NeuroShell Trader® Professional, or NeuroShell® DayTrader Professional, but you can buy them as an optional add-on if you want them. They require release 3.0 or above of the NeuroShell Trader®, NeuroShell Trader® Professional, or NeuroShell® DayTrader Professional.

Click here for a brief overview of the indicators in Advanced Indicator Set 2, which are all listed below:

J. Welles Wilder’s Indicators

• Accumulation Swing Index
• Directional Movement Index
• MinusDI
• MinusDM
• Parabolic SAR
• Parabolic SAR value
• PlusDI
• PlusDM
• Swing Index

Redundant Haar Wavelets

• Redundant Haar Wavelet Coefficient
• Redundant Haar Wavelet Smoothed Curve

Miscellaneous Indicators

• Center Of Gravity Index
• Center Of Gravity Oscillator
• Median Price
• Median Value
• Random Walk Index of Highs
• Random Walk Index of Lows
• RAVI
• Retrace Buy
• Retrace Sell
• Stochastic RSI Oscillator
• Variable Length Moving Average
• Volatility Breakout High
• Volatility Breakout Low

Marc Chaikin’s Indicators

• Accumulation/Distribution Oscillator
• Chaikin Money Flow Persistency
• Chaikin Volume Accumulation Percent

Flag Indicators

• Flag2
• Flag3
• Flag4
• Flag5
• Flag1 toggle with reset condition
• Flag2 toggle with reset condition
• Flag3 toggle with reset condition
• Flag4 toggle with reset condition
• Flag2 with reset condition
• Flag3 with reset condition
• Flag4 with reset condition
• Flag5 with reset condition

Bar/Condition Count Indicators

• Bar Count Since Condition
• Bar Count Of Time Series
• Condition Count

For more details, please view the product manual for this add-on.

Advanced Indicator Set 3

Advanced Indicator Set 3 is an eclectic set of indicators that expand NeuroShell Trader's ability to find a value between two separate conditions or since the occurrence of a single condition. The Find and Count indicators are designed to answer traders’ questions about data such as what is the max value of the close in the time between a Bollinger Band High Breakout and a Bollinger Band Low Breakout or count the number of times the percent change in close was greater than x percent from February 1 to May 31.

In addition, Set 3 adds the power to remember or toggle values off and on based on conditions. The Remember and Toggle indicators may be used to create complex Predictions and Trading Strategies. These functions complement the NeuroShell Trader's emphasis on building indicators with a wizard rather than requiring the user to be a programmer. The result is a significant speed increase in building and testing trading rules.

Advanced Indicator Set 3 also includes indicators specifically designed for pair trading. The PairEntry and PairExit set lets you build traditional market neutral rules that simultaneously buy and sell a pair of instruments that follow one another and frequently cross prices. When those pairs are hard to find, the ProxyEntry and ProxyExit indicators expand possibilities for pair trading by using the NeuroShell Trader’s neural nets to create cointegrated pairs.

Rounding out this diverse mix are two indicators that compute the correct sun time and distance to the moon.

Overview of Indicators

The following descriptions of the Advanced Indicator Set 3 indicators are abbreviated. There are more comprehensive descriptions in the help file.

Conditional Indicators

The task oriented indicators in Set 3 such as Find, Count, Remember, and Toggle require the use of conditional indicators. The NeuroShell Trader includes many conditional indicators such as rules, Boolean, relational, crossovers, candlesticks, etc. Advanced Indicator Set 3 expands those choices with the following indicators:

FollowedBy - returns a 1 for true when a Condition1 is followed by a Condition2. Other times it returns a zero.

TimeIs - specifies a time as a condition for another indicator.

DateIs - specifies a date as a condition.

Peak - specifies the highest value within a certain number of bars.

Valley - looks for the lowest value within a certain number of bars.

BarNumber - returns a 1 when the specified bar number appears in a chart. The first bar in the chart is numbered 1, the second bar is numbered 2, etc.

Find Indicators

Find - calculates values such as count, point change, percent point change, max, min, sum, average, standard deviation, variance, median, regression slope, number of advancing bars, number of declining bars, max advance, and max decline based on two conditions occurring in a data stream.

FindInclusive – does the same as the Find indicator but includes the bars on which the conditions are true.

FindSince and FindSinceInclusive - find similar values from the time Condition1 occurs until the end of the chart.

Count Indicators

Count - calculates a running total of the number of times a specified condition occurs in between a start condition and an end condition.

CountInclusive – same as the Count indicator but includes the bars on which the conditions are true.

CountSince and CountSinceInclusive - these indicators begin their running total with the occurrence of Condition1, but Condition2 is replaced by the current bar.

Remember

Remember - outputs a time series value based on a condition you specify. For example, the Remember indicator would output the value of the high (the time series) when the 9 period moving average crosses above the 13 period moving average (the condition). The output value would remain the same until the condition is met once again, at which time the current high would become the output.

Remember2, Remember3, and Remember4 - output a different value depending upon which of several conditions is activated. The Remember indicators may be used as variable inputs to other indicators, Predictions or Trading Strategies. The conditions are true/false indicators or any indicator that produces a value of either 0 or a non-zero value.

Toggle

Toggle - signifies the presence or absence of conditions you specify. The Toggle indicators may also be used to remove certain data when training a neural net, e.g., only train a net on bars between 1 p.m. and 4 p.m. The Toggle indicators output 0, 1, or a missing value based on the settings for the What parameter.

ToggleInclusive – same as Toggle but includes but includes the bars on which the conditions are true.

ToggleSince and ToggleSinceInclusive - these indicators begin their running total with the occurrence of Condition1, but Condition2 is replaced by the current bar.

Pair Trading

PairEntry - compares the value of the close, which is different for each of the two chart pages, to the value of Price1 (the first stock) and Price2 (the second stock) and determines which stock goes long and which stock goes short. The values of Price1 and Price2 are also used to compute the spread between stocks in order to determine an entry point for the trade.

PairExit - determines the exit conditions for trades entered with the PairEntry indicator. The PairExit indicator includes the StopSpread parameter that can be set as an emergency exit for both stocks.

ProxyEntry – similar to PairEntry but uses a neural net prediction to create proxy pairs by using one stock to predict the price of another. Uses the spread between the actual and predicted prices to pair trade the predicted stock and the one used as an input to the net.

ProxyExit - determines the exit conditions for trades entered with the ProxyEntry indicator. The ProxyExit indicator includes the StopSpread parameter that can be set as an emergency exit for both stocks.

Sun and Moon

EquationofTime - returns the difference between solar time (as measured by a sundial) and mean clock time.

MoonDistance - outputs the distance between the center of the Earth and the center of the Moon expressed in kilometers.

For more details, please view the product manual for this add-on.

Cluster Indicators

The Cluster Indicators, like some of the other add-ons from Ward Systems Group, are based on how the human brain might classify the market. However, this NeuroShell Trader® add-on isn't even a neural network (although neural net enthusiasts will see this as a Kohonen Self Organizing Map, trained by genetic optimization!)

Think about this. Suppose you have two good inputs. Now suppose you made a scatter plot of a bunch of recent bars worth of these inputs. (For the math minded, all recent bars are plotted with one input on the Y axis and the other on the X axis.) Now suppose the inputs are good, as defined by the fact that their patterns foretell an uptrend or a downtrend in the price. Then wouldn't the bars that precede an uptrend look somewhat similar, at least after they were "normalized"? Wouldn't those that precede a downtrend look similar too? In other words, wouldn't the buy bars cluster together and the sell bars cluster together?

The graph displays long entry and short entry clusters for a Cluster Indicator based on the MACD and CCI indicators as inputs.

Our new Cluster Indicators tell you how far the current bar is from the cluster center of the recent "buy" bars or "sell" bars. You can use a Cluster indicator as a buy rule, buying when the current bar is "close" to the cluster center of recent "buy" bars. Sell when the current bar is "close" to the "sell" cluster center.

The genetic optimizer in the Trader Pro or DayTrader Pro finds the cluster centers that optimize the profit.

The Cluster indicators also make dynamite inputs to a neural network. (For the neural net enthusiasts, this creates a paradigm much like the popular classic Radial Basis Function neural nets, except in the Trader Pro, your clusters don't have to all be based on the same inputs, and in fact can be optimized! There was also a paradigm called Counter-propagation that was similar to feeding our Cluster indicators into a neural net.)

Now expand the concept to as many as 16 inputs. Those can't be plotted because you'd need a 16 dimensional scatter plot. But you can still measure how close the current bar is to a 16 dimensional cluster center! There are 16 Cluster Indicators.

Let's carry the concept a little further. Suppose some of the inputs are less effective than others in determining whether there is a buy or sell situation. You wouldn't want to consider these inputs as much when determining clusters and "closeness". No problem, we have 8 Cluster "contribution" indicators, each of which has contribution factors which reduce or increase the impact of an indicator. You can set these contribution factors yourself or you can let the genetic optimizer do it for you. Either way, you now have a more powerful clustering indicator.

Cluster Indicators require release 3.2 or better of the NeuroShell Trader Professional or NeuroShell DayTrader Professional.

For more details, please view the product manual for this add-on.

Fuzzy Pattern Recognizer

How many times have you wished you could just let your software scan a price series and let you know when it has found a particular pattern that you like? You have in your mind a special pattern of moves that you believe could precede a strong change in the market, but you'd like the software to just monitor the incoming bars looking for your pattern. Well now you can do just that with the Fuzzy Pattern Recognizer add-on for the NeuroShell Trader series! And you can do it using fuzzy logic!

Fuzzy logic was invented by Lofti Zadeh, and written about extensively in books by Bart Kosko. Fuzzy logic has been used successfully in machines and software around the world, even in "smart" household appliances. Now you can add this smart control to your trading, and it is easy to use because we've done the hard work for you!

The Fuzzy Pattern Recognizer is a fuzzy logic engine which lets you describe your pattern in "fuzzy" rules (approximate rules of thumb). It is really a series of indicators that show you, on a scale of 0 to 1, how closely the current price series matches your pattern. Turn on an alert in the DayTrader Professional and the Fuzzy Pattern Recognizer will scan your incoming bars looking for all of your favorite patterns. The alert will sound when the match is above .6, .8, or any value you select.

Fuzzy rules are rules that are general in nature, not exactly specified. We describe price or indicator curves with the following fuzzy (verb) rules:

• rises sharply
• rises
• remains steady
• drops
• drops sharply

Here are some example rules written in English which can be specified in the Fuzzy Pattern Recognizer add-on:

• The close rises sharply, then stays steady, then rises sharply again.
• The RSI rises, then drops, then rises, then rises sharply.
• The high remains steady, then drops, then rises, then remains steady, then drops, then drops sharply, then remains
• steady.
• The close drops sharply.

Here are some additional highlights of the Fuzzy Pattern Recognizer:

• If you don't have any favorite patterns, you can let the genetic optimizer find them for you if you own the NeuroShell Trader Professional, or DayTrader Professional.
• You can specify that you want to look for two or more patterns, and then use the "Fuzzy OR" indicator to specify that any of your patterns are acceptable.
• You can specify that you want to look for two or more patterns, and then use the "Fuzzy AND" indicator to specify that all of your patterns must be present.
• "Fuzzy OR" and "Fuzzy AND" indicators can be combined for complex searches.
• As you can see from the example rules, patterns can be found not only in a price stream, but in most indicators such as stochastics, moving averages, etc. as well.
• You specify the number of bars in which a fuzzy verb such as "rises" applies. In other words, you determine that "rises" means over a duration of, say 10 minutes or ten days.
• You can also set the maximum expected variation, i.e., what you consider the maximum "sharp rise" to be. In that way, you can specify fuzzy rules, and then later apply those same rules to less or more volatile issues without recoding the indicator. All you have to do is reset the maximum expected variation.
• Although one use for Fuzzy Pattern Recognizer indicators is in Trading Strategy rules (conditions), you can also feed them into neural nets or other indicators.

There are 8 "fuzzy (verb) rule" indicators in the add-on, Fuzzy1, Fuzzy2, ... , Fuzzy8. Fuzzy1 handles "1 segment" rules such as "close rises" or "open drops sharply". Fuzzy2 handles "2 segment" rules such as "close rises, then drops sharply". Fuzzy3 is for "3 segment" rules such as "open rises, then drops, then remains steady", etc.

Example of the Fuzzy2 indicator output (bar chart at the bottom).
The search rule is "Close rises, then drops", Segment Size=8 bars, and Max Change=7.

There are 3 FuzzyOR rules and 3 FuzzyAND rules: FuzzyOR2, FuzzyOR3, FuzzyOR4, FuzzyAND2, FuzzyAND3, FuzzyAND4. As previously mentioned, these combine rules. For example, suppose you are looking for either a pattern that rises sharply then drops sharply, or a pattern that rises sharply, then remains steady, and then drops sharply. Then you would use a FuzzyOR2 indicator which takes as arguments a Fuzzy2 indicator and a Fuzzy3 indicator, each specifying their respective rules.

FuzzyOR and FuzzyAND indicators may take other FuzzyOR and FuzzyAND indicators as arguments, as long as the lowest level indicators are "rule" indicators such as Fuzzy2.

There are also 8 FuzzyGA indicators, which allow the genetic optimizer to tune the fuzzy logic engine better.

We believe that Fuzzy Pattern Recognizer indicators open a whole new world of possibilities in trading with artificial intelligence. We think it will be our most popular add-on yet!

The Fuzzy Pattern Recognizer add-on requires release 3.2 or better of the NeuroShell Trader, NeuroShell Trader Professional, or NeuroShell DayTrader Professional. If you do not own one of the Pro versions, you can still use the Fuzzy Pattern Recognizer add-on to find patterns, but you will not be able to optimize the Fuzzy indicators.

For more details, please view the product manual for this add-on.

Fuzzy Sets

You've asked for it, and now it is here: the Fuzzy Sets add-on for NeuroShell Trader Professional and DayTrader Professional.

Fuzzy logic was invented by L.A. Zadeh and further popularized by Bart Kosko. Fuzzy logic is not some theoretical idea that has yet to catch on. By 1990 the Japanese had over 100 real fuzzy control applications and products. The city of Sendai in Japan has controlled its subway with fuzzy logic since 1988. General Motors’ highly successful Saturn applies fuzzy logic for automatic transmission shift control. Duke University engineers have shown that intentionally imprecise rules of thinking can help hotel computers sell the right room to the right customer at the right time, thus boosting income.

Ward Systems Group introduced what we believe was the first commercial trading software to allow users to describe price movements with fuzzy logic statements (the Fuzzy Pattern Recognizer add-on). A user could essentially give NeuroShell instructions such as:

"Find patterns where the price rose, then dropped, then dropped sharply, then rose sharply."

The new Fuzzy Sets add-on is somewhat different in that it allows users to describe a combination of values of traditional indicators with fuzzy logic. With Fuzzy Sets the user can instruct NeuroShell with functions equivalent to fuzzy statements like:

"Buy when the Stochastic %K indicator is very high, and the Commodity Channel Index is high, and the spread between two moving averages is low."

Contrast that with the typical trading utilization of an indicator is to make decisions based upon rules like whether that indicator (let’s say the Stochastic %K) is above or below some threshold. For example:

Buy when the STOCH%K > 70 and sell when the STOCH%K < 40

Those 2 rules above are sometimes called "crisp" rules, because they "break off" suddenly. In the last rule above, you are buying when the STOCH%K is 70 but not when it is 69. A human might still give consideration to buying when the STOCH%K reached 69, or even at a level of 65, although he would not consider it as strong as a buy as when STOCH%K = 70. In fact, a human would probably consider STOCH%K=80 to be a much stronger buy sign than STOCH%K=70.

Fuzzy rules are a little different than crisp rules in that they do not break off suddenly. Let's consider the following example, where we will use the STOCH%K again. Suppose we have decided that STOCH%K can be considered to be in one of 5 categories or "fuzzy sets":

• very low
• low
• medium
• high
• very high

Then we could create "fuzzy" rules such as one of these two:

Buy when STOCH%K is high and sell when STOCH%K is very low. Buy when STOCH%K is very high and sell when STOCH%K is medium.

Actually the Fuzzy Sets add-on does not really use fuzzy words like low, very high, etc. For one thing, instead of just 5 fuzzy sets (very low, low, medium, high, and very high) you can have 10 or 20 or even more if you like. So instead of words, we use 1 for the lowest, 2 for the next lowest, etc. on up to the highest range.

Of course, most people will have no idea what fuzzy set you should be looking for in order to buy or sell, so that's where the NeuroShell optimizer comes in. The optimizer will decide the total number of fuzzy sets, and which set you should have for a long entry, long exit, short entry, and short exit.

Our Fuzzy Sets indicators can be used in Trading Strategies as well as neural nets. We have some that do a Fuzzy-AND of up to 10 fuzzy indicators, and other that do a Fuzzy-OR of up to 10 fuzzy indicators.

For more details, please view the product manual for this add-on.

Neural Indicators

Neural Indicators (NI) are several types of neural networks packaged as technical indicators. They will probably appeal to those who like to experiment with neural nets, because there are several types, all with different features. They are based somewhat on the "backpropagation" nets that we feature in our NeuroShell 2 classic network package, but there are two major differences. First, they are not trained with the old backprop neural network. Second, they are not trained to predict anything, only to produce accurate buy and sells. Here are the salient features:

Outputs

NI provide outputs from -1 to 1. The general interpretation of these outputs will be some sort of "binary" classification, like "buy" (>0) or "don't buy" (< or = 0). Usually, we simply insert them in a trading strategy as a rule (condition) to buy and sell.

Probabilities

NI give probabilities of the outputs they produce, because if they are providing a sell, for example, the closer to 1 the output is, the stronger the probability of sell. The closer to -1, the stronger the probability of "don't' sell".

Unsupervised

NI are "unsupervised" neural networks, meaning that they do not need to be trained by showing them the correct answers, like most neural networks (i.e., "supervised" neural networks). You do not teach them by providing any kind of actual output which they learn to reproduce, as with most neural nets.

Evolutionary

NI "learn" how to give their outputs based upon evolutionary pressure. The genetic optimizer in the NeuroShell Trader Professional or NeuroShell DayTrader Professional "evolves" NI that give better and better outputs. Survival of the fittest controls the evolutionary process as usual, where fitness is determined by how much money the NI make, or how good they work as inputs to other nets or indicators.

Architectures

Although anyone can use NI, neural network aficionados will love them because there are several highly technical neural network "architectures" from which you can choose.

Warning: You do not have to understand how these architectures work in order to build profitable trading rules. You can insert Neural Indicators into your trading rules and let the optimizer find the appropriate settings. Neural Indicators are highly technical from a neural network standpoint, and although they can be used by novices, they can only be fully understood internally by those who have a background in how neural networks function.

Ward Nets. This architecture has two different "activation functions" in the hidden neurons. These are called "Ward Nets" since Ward Systems Group invented them many years ago (they first appeared in our classic product NeuroShell 2). The genetic optimizer will find out how to pick the activation functions for you. Ward Nets include Ward2, Ward3, Ward4, Ward5, and Ward6, depending upon the number of inputs.

Jump Connections. This architecture has connections directly from inputs to outputs as well as the usual hidden neuron connections. This architecture also features connections from one hidden neuron to the next, like Turboprop 2 has. Jump Nets include Jump2, Jump3, Jump4, Jump5, and Jump6, depending upon the number of inputs.

Recurrent Nets. This architecture analyzes not only the current bar of information to produce its output, but it also reviews a condensed summary of the most recent bars as well. More recent bars receive more weighting than older ones. Recurrent Nets include Recur2, Recur3, Recur4, Recur5, and Recur6.

Sparse Nets. These are nets which are not fully connected between the input and hidden neural layers. This means that more inputs can be fed to them without increasing the number of weights drastically. The fact that less information can be stored in sparse weighting connections is compensated for by the fact that less weights allow better optimization. Sparse Nets include Sparse8, Sparse10, and Sparse12.

Conditional Versions

We have implemented modified versions of all of the above nets. The conditional version of each of these architectures produces a value of either true or false rather than a value between -1 and 1. Therefore the conditional version of any of the architectures can be used in a Trading Strategy without the necessity of inserting it into an A>B indicator.

Generalization

These nets generalize very well, meaning they do not have a strong tendency to "overfit" or "curvefit" like backpropagation neural nets do.

Neural Indicators require the NeuroShell Trader Professional or NeuroShell DayTrader Professional, release 3.2 or better.

For more details, please view the product manual for this add-on.

Pattern Matcher

The Pattern Matcher add-on grew out of many requests by users who wanted to know if patterns in their data streams had occurred before in history, and if so, what type of activity followed those historical patterns. Some users also wanted to identify a pattern in history and be alerted when a similar pattern occurs in the future. The Pattern Matcher add-on meets both needs by providing indicators that match patterns and revealing the subsequent activity.

Pattern Matcher indicators work on any data stream in the NeuroShell Trader, Trader Professional, or DayTrader Professional. The Pattern Matcher can find patterns in raw price data streams or indicators made from raw data streams. It is often advantageous to match patterns in indicators applied to prices rather than to the prices themselves, as you will see in a subsequent section of this document.

The Pattern Matcher is unlike any of our other add-ons in that it is not designed for use in either a neural network or a trading strategy to make buy/sell decisions. It is intended that the user treat the Pattern Matcher as an intellectual tool to provide input to one's own biological neural networks for discretionary trading, or as confirmation of other indications you may be obtaining in other ways.

So what does this mean? It means that the Pattern Matcher is designed to provide answers to the questions that have plagued technical analysts since long before computers were born:

• Has the current pattern that we see in the most recent bars occurred in the past?
• If so how many times, and how closely do those past patterns match the current one?
• If there were a number of closely matching patterns in the past, what type of market moves did they precede on average?

You will be making your own subjective decisions about whether or how to trade based on the information the Pattern Matcher is showing you. These subjective judgments will be based on your own study and experience watching similar patterns, and based upon your decisions about what data streams in which to watch for patterns. Price streams (like open, high, low, close, and volume) may not be the proper data streams to be monitoring for patterns. Indicator data streams, in our experience, seem to provide a more reproducible pattern source.

At Ward Systems, we have used the Pattern Matcher in the way just described. We have found that there is less value in trying to feed Pattern Matcher input into either neural nets or rules. We believe the tool is a powerful resource which will help you even the playing field, so to speak, as you compete with professional traders who have years of experience recognizing and reacting to patterns in the markets. They rely on their memories and intuition - you now have an analytic tool to bring to odds of success more in your favor.

Furthermore, you should not assume that past patterns are an indicator of future performance. The pattern matcher may very well be a powerful tool for the contrarian trader too. The markets are quite capable of reversing what usually happens after recurring patterns. If this were not true, it would be easy to predict and make money in the markets, and we all know that it isn't easy.

Overview of Indicators

The following descriptions of Pattern Matcher Indicators are abbreviated. There are more comprehensive descriptions in the help file.

Match - The Match indicator takes a very recent pattern on the chart and displays the probability that other patterns in history on the chart match that current pattern.

MatchPast - The MatchPast indicator takes a designated pattern on the chart and displays the probability that other patterns in both history and future on the chart match that designated pattern.

ProjectAvg - The Project Average indicator returns the average of all projections (leads) of the projection series for all pattern matches that are greater than the indicated projection threshold.

ProjectMin - The Project Average indicator returns the minimum of all projections (leads) of the projection series for all pattern matches that are greater than the indicated projection threshold.

ProjectMax - The Project Average indicator returns the maximum of all projections (leads) of the projection series for all pattern matches that are greater than the indicated projection threshold.

ProjectStndDev - The Project Standard Deviation indicator returns the standard deviation of all projections (leads) of the projection series for all pattern matches that are greater than the indicated projection threshold.

ProjectRange - The Project Range indicator returns the range of all projections (leads) of the projection series for all pattern matches that are greater than the indicated projection threshold.

ProjectUpperB - The Project Upper Band indicator returns the upper envelope of all projections (leads) of the projection series for all pattern matches that are greater than the indicated projection threshold.

ProjectLowerB - The Project Lower Band indicator returns the lower envelope of all projections (leads) of the projection series for all pattern matches that are greater than the indicated projection threshold.

ProjectCount - Given a projection threshold, the Project Count indicator returns the total number of matches between the current pattern and patterns in history.

ProjectPercent - Given a projection threshold, the Project Percent indicator returns the percentage of total number of matches between the current pattern and patterns in history.

For more details, please view the product manual for this add-on.

Turning Points

The Turning Points add-on for the NeuroShell Trader series is based on the concept of finding local peaks and valleys in a price series.

A turning point (TP) peak is defined as the point where the high is higher than or equal to any other highs in the neighborhood of the point. A turning point valley is defined as the point where the low is lower than or equal to any other lows in the neighborhood of the point.

The above definitions are illustrated by an example graph below.

The add-on contains indicators that do all of the following with respect to these turning points:

• Plot straight lines from one turning point to the next
• Plot support and resistance lines horizontally from the peaks and valleys
• Compute an oscillator which shows how the current close compares to a number of previous support or resistance levels
• Plot horizontal Fibonacci retracement lines calculated from peak to valley or valley to peak segments
• Compute statistical measures (mean, median, and standard deviation) of the price differences, time differences, and slope differences between turning points in a user-defined window
• Provide the probability that the current price level is at a new turning point based upon either of two of the computed statistical measures

We are especially fond of the probability indicators above, because many issues exhibit repeating cyclic tendencies which can be captured with this technique. Of course, news and other factors can affect normal cyclic tendencies, and not every trade generated will be correct. However, there is plenty of capability in this add-on to build trading strategies that even we haven't thought of yet.

Following is a complete list of the indicators in the Turning Points add-on:

• Turning Point Plot Indicator
• Tpplot
• Individual Turning Point Measure Indicators
• TPbars
• TPchange
• TPpercent
• TPslope
• Turning Point Probability Indicators
• Peak probability (% change)
• Valley probability (% change)
• Peak probability (bars)
• Valley probability (bars)
• Mean Indicators
• PVmeanbars
• VPmeanbars
• PVmeanchange
• VPmeanchange
• PVmeanpercent
• VPmeanpercent
• PVmeanslope
• VPmeanslope
• Standard Deviation Indicators
• PVsdbars
• VPsdbars
• PVsdchange
• VPsdchange
• PVsdpercent
• VPsdpercent
• PVsdslope
• Vpsdslope
• Median Indicators
• PVmedianbars
• VPmedianbars
• PVmedianchange
• VPmedianchange
• Pvmedianpercent
• VPmedianpercent
• PVmedianslope
• Vpmedianslope
• Support and ResistanceSupport Level
• Support Level
• Resistance Level
• Support Oscillator
• Resistance Oscillator
• Fibonacci Retracements

The Turning Points add-on requires the NeuroShell Trader, NeuroShell Trader Professional, or NeuroShell DayTrader Professional, release 3.8 or better.

For more details, please view the product manual for this add-on.