AU2020103525A4 - IUML- Stock Prices Predictor: Intelligent Stock Prices Predictor Using Machine Learning - Google Patents

Abstract

Our Invention "IUML- Stock Prices Predictor" is a suite of predictions is defined to model the financial data commonly used to calculate technical indicators one or more periods in the future. The invented technology using machine learning and neural networks are trained to make these predictions and the predictions are then integrated with the standard technical indicator calculations to produce predictive technical indicators which are superior because they lead more and lag less. The invented technology is also apparatus and method for a stock investment method with intelligent agents is described and illustrated. The invented technology is a stock prediction system that through experience learns to make money based on short-term stock predictions and due to inherent flexibility continues to be profitable in virtually all market environments. 21 C o mm od It y Stock Foreign Exchanges Exchang Exhage Predictive inancal Preditive us nIG •:I LC IGA LUTAIGA NOMTO RCSIGSSE FTEIVNIN

Description

C o mm od It y Stock Foreign Exchanges Exchang Exhage

Predictive inancal Preditive

us

nIG LC IGA LUTAIGA NOMTO RCSIGSSE •:I FTEIVNIN

IUML- Stock Prices Predictor: Intelligent Stock Prices Predictor Using Machine Learning

FIELD OF THE INVENTION

Our invention "IUML- Stock Prices Predictor" is related to intelligent Stock Prices Predictor Using Machine Learning and also to technical analysis of financial markets. More particularly, the invention relates to methods and systems for calculating predictive technical indicators prediction using computer programs.

BACKGROUND OF THE INVENTION

Historically, the methods that have been used by traders to analyze the financial markets in an effort to identify and forecast the direction of price trends have been divided into two distinct approaches: fundamental analysis and technical analysis. Fundamental analysis focuses on underlying macro- and/or micro-economic factors such as Gross National Product, central bank policies, rates of inflation, unemployment rates, market share, earnings, profitability and supply/demand. The premise behind technical analysis is that all the factors that affect a specific market at any given point in time are already built into that market's price, even if these factors are based on fundamentals or mass psychology. Technically oriented traders concentrate on using various technical studies, indicators, and market-forecasting theories to analyze market behavior.

Traders are people who buy and sell financial instruments that are publicly traded on exchanges. Trading software applications subscribe to data from the exchanges and present it to traders, usually in the form of charts and watch lists. Traders and trading applications have come up with a variety of calculations that can be performed on electronic exchange data. Some of the more common technical indicators include trend indicators, momentum indicators, and volatility indicators. Many technical indicators, such as moving averages, attempt to filter out short-term variation in price so the underlying trend can be observed. A side effect of averaging past prices is that the indicator tends to lag behind the market. This causes the trader to respond late to market changes, resulting in lost profit opportunity and risk of increased losses.

Various analytical and predictive techniques have been devised for purposes of predicting. Some techniques may operate on simple concepts but may use variables or parameters that must be characterized or selected by a human user or operator in order to arrive at an analysis or prediction. For example, the common measure of a "moving average" of a stock's price is a simple calculation but the start and end of the time period used to calculate the moving average may vary. Although traditional techniques have proven to be useful for prediction and analysis of stock prices, as the number and complexity of techniques grows it is often difficult for a human user of the techniques to effectively use the techniques and to combine or correlate the various results provided by the techniques.

PRIOR ART SEARCH

US5444819A *1992-06-081995-08-22Mitsubishi Denki Kabushiki Kaisha Economic phenomenon predicting and analyzing system using neural network US6269353B1*1997-11-262001-07-311shwar K. Sethi System for constructing decision tree classifiers using structure-driven induction US6735580B1 *1999-08-262004-05-llWestport Financial Llc. Artificial neural network based universal time series.

OBJECTIVES OF THE INVENTION

1) The objective of the invention is to a suite of predictions is defined to model the financial data commonly used to calculate technical indicators one or more periods in the future. 2) The other objective of the invention is to a technology using machine learning and neural networks are trained to make these predictions and the predictions are then integrated with the standard technical indicator calculations to produce predictive technical indicators which are superior because they lead more and lag less. 3) The other objective of the invention is to a apparatus and method for a stock investment method with intelligent agents is described and illustrated. 4) The other objective of the invention is to a the invented technology is a stock prediction system that through experience learns to make money based on short term stock predictions and due to inherent flexibility continues to be profitable in virtually all market environments.

SUMMARY OF THE INVENTION

This invention relates to the development of methods, systems, and devices for developing technical indicators based on the combination of both historical and predicted data for a market. Historical data for a market can be obtained from available sources such as an exchange where the market trades. Predicted data can be obtained using a predictive server that uses intermarket analysis data to train neural networks to predict financial time series data. The invention can include software running on a back-end server (a predictive server) and a trading application (e.g., an application that runs on a trader's PC), which operate within an information processing system that includes market exchanges and financial data providers.

In one aspect of the invention, historical time series data can be obtained and analyzed, predicted future times series data can be obtained using intermarket analysis performed using neural networks, conventional technical indicator information can be obtained, and an algorithm can be used to integrate the predicted data with the conventional technical indicator information to arrive at a predicted technical indicator. The use of both historical data and predicted data results in a technical indicator that can lead more (lag less) than the conventional technical indicator on which it is based.

Intermarket analysis searches for relationships between markets that can be used to obtain useful information about what the prices of the markets will do. An intermarket whose price activity leads that of a market of interest is especially useful for financial forecasting. In the invention, intermarkets can be selected on a market-by-market basis prior to the training of the neural networks. The selected intermarkets can then be used during the training process to provide additional inputs to the neural networks, which improve the accuracy of the predictions. The intermarkets can be screened or selected on the basis of correlation analysis. The correlation can be a statistical calculation that measures the degree and type of relationship between two series of numbers, e.g., a positive correlation indicates the two series move together, a negative correlation indicates they move inversely, and zero correlation indicates the two series vary independently of each other.

Accordingly, the invention features a computer-implemented method of calculating a technical indicator of a market. The method includes the steps of: (a) obtaining historical data relating to the market from a server that includes a database of historical time series data for the market; (b) obtaining predicted future data relating to the market, e.g., using a predictive server that uses intermarket analysis data to train a neural network to predict financial time series data for the market, or other suitable methods such as trend line analysis, fundamental analysis, and other market-forecasting theories; (c) using both the historical data and the predicted future data to calculate the predictive technical indicator; and (d) outputting the calculated predictive technical indicator.

The information processing system can be divided into a predictive server and a predictive trading application. The predictive server can include a market manager component, an historical data manager, a neural network trainer, a trading application builder, an intermarket analysis component, a technical analysis component, a predictive technical analysis component, and a predictive server database. The calculated predictive technical indicator can be communicated to a trading application via a communications network. The trading application can include a watch lists component, a charts component, a grids component, a reports component, an intermarket analysis component, a technical analysis component, a predictive technical analysis component, and a predicative trader database.

The invention features a system for calculating a predictive technical indicator of a market. The system can include: (a) a first server that includes a database of historical time series data for the market; (b) a second server that uses intermarket analysis data to train a neural network to generate predicted financial time series data for the market or other suitable forecasting methods; and (c) a computer in communication with the first and second servers, the computer including computer-usable program code that uses both historical time series data for the market and the predicted financial time series data for the market to calculate the predictive technical indicator.

The second server can include a market manager component, an historical data manager, a neural network trainer, a trading application builder, an intermarket analysis component, a technical analysis component, a predictive technical analysis component, and a predictive server database. The system can further include a trading application, and the calculated predictive technical indicator can be communicated to the trading application via a communications network. Alternatively, the system can further include a trading application on a trader's personal computer which calculates the predictive technical indicator using data communicated to it from a geographically remote (e.g., more than 1 km away) server. In the system, the trading application can include a watch lists component, a charts component, a grids component, a reports component, an intermarket analysis component, a technical analysis component, a predictive technical analysis component, and a predicative trader database.

Also within the invention is a computer program product that can include a computer usable medium including computer-usable program code that, when executed by a computer, calculates a technical indicator for a market. The computer-usable medium can include: computer-usable program code that uses both historical time series data for the market and predicted financial time series data for the market obtained using a neural network that can be trained using intermarket analysis data to calculate the predictive technical indicator.

Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although methods and devices similar or equivalent to those described herein can be used in the practice or testing of the present invention, suitable methods and devices are described below. All publications, patent applications, patents and other references mentioned herein are incorporated by reference in their entirety. In the case of conflict, the present specification, including definitions will control.

Embodiments of the invention are described in conjunction with systems, clients, servers, methods, and machine-readable media of varying scope. In addition to the aspects of the present invention described in this summary, further aspects of the invention will become apparent by reference to the drawings and by reading the detailed description that follows. An apparatus and method for a stock investment method with intelligent agents is described and illustrated. In one embodiment, the invention is a stock predicting system that through experience learns to make money based on short-term stock predictions and due to inherent flexibility continues to be profitable in virtually all market environments.

BRIEF DESCRIPTION OF THE DIAGRAM

FIG. 1: is block diagram illustrating an information processing system of the invention.

FIG. 2: is block diagram illustrating the architecture of a predictive server.

FIG. 3: is block diagram illustrating the components of a trading application for use in the invention.

FIG. 4: is flow chart showing a method of generating a predictive technical indicator.

FIG. 5: is a chart displaying the price of gold over three months using daily bars, a 5-day simple moving average, and a 10-day simple moving average.

FIG. 6: illustrates relationships between an embodiment of an application and various other modules, data stores, and interfaces, such as may be embodied in a medium or in media. (Prior art)

FIG. 7: illustrates an embodiment of an application utilizing intelligent agents. (Prior Art)

DESCRIPTION OF THE INVENTION

This invention provides methods, systems, and devices for developing predictive technical indicators based on the combination of both historical and predicted data for a market, and can include software running on a back-end server (the predictive server) and a trading application (e.g., an application that runs on a trader's PC), which operate within an information processing system that includes market exchanges financial data. In a preferred embodiment, the methods, systems, and devices are used in conjunction with the methods and systems described in U.S. patent application Ser. No. 12/632,186 entitled "Intermarket Analysis" filed on Dec. 7, 2009 by inventor Louis B. Mendelsohn. The below described preferred embodiments illustrate adaptation of these methods, systems, and devices. Nonetheless, from the description of these embodiments, other aspects of the invention can be made and/or practiced based on the description provided below.

Various aspects of the invention may be embodied as a system, method, or computer program product (e.g., embodied in one or more computer readable media having computer readable program code embodied thereon), and might be in the form of hardware, software, or a combination of software and hardware. Computer readable media may be a computer readable signal medium (e.g., an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing) or a computer readable storage medium (e.g., an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing).

Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing. Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object-oriented programming language such as Java, Smalltalk, C++, and procedural programming languages such as C. The program code may execute entirely on a user's computer, entirely on the remote computer or server, or partly on a user's computer and partly on a remote computer or server. A remote computer may communicate with a user's computer through any type of communications network, e.g., a local area network, a wide area network or the Internet.

In the figures, blocks of the flowchart illustrations and block diagrams might be implemented by computer program instructions, which may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine such that the instructions execute to implement the functions/acts specified in the blocks. Computer program instructions may be stored in a computer readable medium that can direct a computer, other programmable data processing apparatus, or other devices to produce an article of manufacture.

In some alternative implementations, the functions noted in the block may occur out of the order noted in the Figures. For example, two blocks shown in succession may be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. Each block of the block diagrams and/or flowchart illustrations might be implemented by special purpose hardware-based systems that perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

Embodiments of the invention are described in conjunction with systems, clients, servers, methods, and machine-readable media of varying scope. In addition to the aspects of the present invention described in this summary, further aspects of the invention will become apparent by reference to the drawings and by reading the detailed description that follows.

An apparatus and method for a stock investment method with intelligent agents is described and illustrated. In one embodiment, the invention is a stock predicting system that through experience learns to make money based on short-term stock predictions and due to inherent flexibility continues to be profitable in virtually all market environments. In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the invention. It will be apparent, however, to one skilled in the art that the invention can be practiced without these specific details. In other instances, structures and devices are shown in block diagram form in order to avoid obscuring the invention.

The reference in the specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment nor are separate alternative embodiments mutually exclusive of other embodiments. In the following detailed description of embodiments of the invention, reference is made to the accompanying drawings in which like references indicate similar elements, and in which is shown by way of illustration specific embodiments in which the invention may be practiced.

These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention, and it is to be understood that other embodiments may be utilized and that logical, mechanical, electrical, functional, and other changes may be made without departing from the scope of the present invention. The flowing detailed description is, therefore, not to be taken in a limiting sense, and the scope of the present invention is defined only by the appended claims.

In one embodiment, the system is the implementation of a Technical Analysis approach to the stock market that is based on and exploits the following assumptions. Some of these assumptions are rather non-traditional and may even turn out to be false, but due to the flexibility of our overall architecture and interactions even bad choices can turn out to be good. Stock prices are not a "random-walk" and past price-volume trading behavior provides enough information (if processed carefully) for future price behavior to be predicted at a level of statistical and profitable significance. Given proper normalization and canonization of past data, all securities in all time frames exhibit behavior that is useful in helping to predict a future price move at a given time.

For example, IBM's trading day tomorrow may resemble the MEX (Mexican) index 255 days ago, especially if a strong analogy can be established between their current and underlying technical environments. Despite these similarities, after normalization, each security or index may also exhibit characteristics and rhythms that are essentially their own "signature."

A market predicting system must be complex enough to model a large gamut of technical trading strategies at varying time frames in order to simulate the habits of populations of traders that follow (or appear to) follow these strategies. Given a security, certain predicting strategies will have proved to be more useful than others at predicting recent stock behavior. A stock predicting strategy can never be "very bad" since its very badness can be exploited by trading contrary to it. The only useless features and predictions are those that are essentially random. However, perversely, some "mal-features" may manage to change their success as soon as one tries to exploit them. Clearly, it is these mal features that must be ignored or avoided or exploited when properly recognized.

Combining these assumptions, a useful stock prediction can be developed as a function of:

a. The past price behavior of the stock,

b. Its past price behaviors, and the relationship to other securities in similar scenarios,

c. The relative successes of various features (trading strategies) at predicting correctly or incorrectly recent price behavior (weighing these successes or failures by the amount of win or loss as described in detail later). These features may come from traditional technical analysis books, general and chaos theory time-series analysis, and other human or computer designed features and "expertise modules". As long as mal-features and over fitting can be avoided, adding new features to the system should improve performance in the long run once the system becomes adept at using these features.

d. The rhythm of the successes and failures of individual features. Features themselves may be viewed as securities for which predictions (at a meta-level) become relevant,

e. The Metropolis simulated annealing strategy of "heating up" (to encourage innovation) a system that is doing poorly and "cooling" a system that doing well is also used. Specifically, the system adjusts the learning rate to be higher (hotter) or lower (cooler) by decreasing or increasing the historical time period covered by output signals used by the system to make a final prediction. In one embodiment these adjustments are made in the final combining neural network so if the system is doing well it effectively considers larger advisor histories than it does when it is doing poorly. This added randomness should keep systems out of ruts created by any mal-feature behavior.

Referring now to FIG. 1, in one aspect, an embodiment of the invention includes an information processing system 10 that features a predictive server 12 and a trading application 14. In the system 10, market data from exchanges such as commodity exchanges 16, stock exchanges 18, and the foreign exchange 20 is obtained from one or more financial data providers 22. The market data is transferred from one of the financial data provider 22 to the predictive server 12. The trading application 14 can get its market data from any of the data providers 22. The predictive server 12 processes the market data using neural networks and intermarket analysis to produce trained neural networks which are then provided to the trading application 14 where the predictive indicators are calculated.

The commodity exchanges 16 can be one or more of any exchanges where the buying and selling of commodities such as grain, cattle, and lumber is performed. Examples of commodity exchanges include the Brazilian Mercantile and Futures Exchange, the CME Group, the Chicago Climate Exchange, the Hedge Street Exchange, the Intercontinental Exchange, the Kansas City Board of Trade, the Memphis Cotton Exchange, the Mercado a Termino de Buenos Aires, the Minneapolis Grain Exchange, the New York Mercantile Exchange, the U.S. Futures Exchange, Bursa Malaysia, the Central Japan Commodity Exchange, the Dalian Commodity Exchange, the Dubai Mercantile Exchange, the Dubai Gold & Commodities Exchange, the Iranian oil bourse, the Kansai Commodities Exchange, the Mercantile Exchange Nepal Limited, the Multi Commodity Exchange, the National Multi-Commodity Exchange of India Ltd, the National Commodity Exchange Limited, Bhatinda Om & Oil Exchange Ltd., the Karachi, the National Commodity and Derivatives Exchange, the Shanghai Futures Exchange, the Shanghai Singapore Commodity Exchange, the Tokyo Commodity Exchange, the Tokyo Grain Exchange, the Zhengzhou Commodity Exchange, the Commodity Exchange Bratislava, the Climex, the NYSE Liffe, the European Climate Exchange, the London Metal Exchange, the European Energy Exchange, and the Australian Securities Exchange.

The stock exchanges 18 can be one or more of any exchanges where the buying and selling of stocks or other securities occurs. Examples of stock exchanges include the American Stock Exchange (AMEX), the Boston Stock Exchange, the Chicago Stock Exchange, the Cincinnati Stock Exchange, the NASDAQ, the New York Stock Exchange (NYSE), the Pacific Exchange, the Philadelphia Stock Exchange, the Toronto Stock Exchange (TSX), the Alberta Stock Exchange (ASE), the Canadian Venture Exchange

(CDNX), the Nasdaq Canada, the Bourse de Montreal, the Jamaica Stock Exchange, the Bolsa Mexicana de Valores (BMV), the Euronext, the Helsinki Stock Exchange HEX, the Paris Stock Exchange, the Frankfurt Stock Exchange, the Italy Stock Exchange, the Amsterdam Stock Exchange, the Oslo Stock Exchange, the Lisbon Stock Exchange, the Warsaw Stock Exchange, the Bucharest Stock Exchange (BVB), the Russia Stock Exchange, the Madrid Stock Exchange, the Stockholm Stock Exchange, the Swiss Stock Exchange, the London Stock Exchange (FTSE), the Tel Aviv Stock Exchange, the Tokyo Stock Exchange (TSE), the Nagoya Stock Exchange, the Nasdaq Japan Market (NJ), the Stock Exchange of Hong Kong (SEHK), the Taiwan Stock Exchange, the Thailand Stock Exchange, the Kuala Lumpur Stock Exchange, the Korea Stock Exchange, the Singapore Stock Exchange, the Buenos Aires Stock Exchange, the Sao Paulo Stock Exchange (BOVESPA), the Rio de Janeiro Stock Exchange, the Brazilian Mercantile and Futures Exchange (BM&F), the Maringa Mercantile and Futures Exchange, the Santiago Stock Exchange, the Australian Stock Exchange (ASX), the New Zealand Stock Exchange (NZSE), and the Johannesburg Stock Exchange.

The foreign exchange (FOREX) 20 is where over-the-counter currency trading takes place. There currently is no central clearing house for over-the-counter currency trading, but rather a network of banks, commercial companies, central banks, hedge funds, investment management firms, retail foreign exchange brokers, non-bank foreign exchange companies, money transfer/remittance companies, and other entities. The financial data provider 22 can be an organization that obtains and delivers information on markets to interested parties via a variety of data products. Examples of financial data providers include the Commodity Research Bureau, Standard & Poor's, MTS Reference Data, Exchange Data International, Reuters Datalink, Thomson Financial, Interactive Data Corporation, ICAP, and Bloomberg.

The predictive server 12 can be any device or system capable of obtaining data from the financial data provider 22, using neural networks and intermarket analysis to produce a trained neural network, and transmitting data output from the trained network to the predictive trading application 14. The predictive server 12 can accumulate historical data going back decades for thousands of markets supported by the system. It can then use such historical data to train and retrain the server's neural networks (e.g., which can contain over 1,000; 2,000; 5,000; 10,000; 20,000; 40,000; 60,000; 80,000; or 100,000 connections) to make a suite of indicators over a variety of time durations.

Referring now to FIG. 2, in one embodiment, the predictive server 12 can include at least four (e.g., 5, 6, 7, or 8) of: a market manager component 30, an historical data manager 32, a neural network trainer 34, a trading application builder 36, an intermarket analysis component 38, a technical analysis component 40, a predictive technical analysis component 42, and predictive server database 44. The market manager component 30 manages the collection of markets supported by the system 10 that are packaged into a variety of products in multiple languages. The historical data manager 32 accumulates (manually or automatically) a variety of financial data for every market in the system.

The neural network trainer 34 trains the suite of predictions for each market and tests them against new data, retraining as necessary or desired. The intermarket analysis component 38 searches for and quantifies intermarket relationships that can be used by the neural networks to increase the accuracy of the predictions. The technical analysis component 40 computes technical indicators that are used as inputs to the neural networks to enhance the training process. The predictive technical analysis component 42 integrates the neural network predictions with technical indicator calculations to produce predictive technical indicators for prior periods. Their accuracy can be measured by the predictive server. The predictive server database 44 stores all the market data required to support the system 10. The trading application builder 36 extracts the data necessary for a particular product, version and language of the trading application 14.

The neural network trainer 34 can include an artificial neural network, which is a system that uses a mathematical technique that models the neurons and synapses of a brain. Artificial neural networks have been effectively applied to pattern recognition and time series forecasting problems where the underlying relationships are poorly understood. The neural network can be organized into an input layer, one or more hidden layers, and an output layer. The neural network must be 'trained' on a set of data that includes both inputs and outputs. The training process adjusts the weights of the hidden layer neurons in a guided fashion until the inputs multiplied by the hidden weights is as close as possible to the outputs. Once trained, inputs for which the outputs are not known can be fed into the neural network. It will multiply them by the hidden layer weights and produce a predicted output. See, Neural Networks in Finance: Gaining Predictive Edge in the Market, Paul D. McNelis, Academic Press Advanced Finance, 2005; Neural Networks: A Systematic Introduction, Raul Rojas. Springer, 1996; and Neural Networks and the Financial Markets: Predicting, Combining and Portfolio Optimization (Perspectives in Neural Computing), Jimmy Shadbolt and John G. Taylor, Springer, 2002. The use of neural networks described in U.S. Pat. Nos. 5,303,328; 5,444,819; 6,247,001; 6,735,580; and 7,082,420 might be adapted for use in the current invention.

In one embodiment of the invention, the predicted output can be a prediction of a future value in a time series. As long as the data being run through the neural network is similar to the training data, the prediction will be approximately as accurate as the results obtained during training The accuracy of the prediction depends on many factors, the most important of which can be the presence of patterns in the data and relationship between the inputs and the value being predicted. Neural networks can be trained on electronic exchange data as well as technical indicators to predict the future price of a market.

Referring now to FIG. 3, the trading application 14 can include a watch lists component 50, a charts component 52, a grids component 54, a reports component 56, an intermarket analysis component 58, a technical analysis component 60, a predictive technical analysis component 62, and a predictive trader database 64. The watch lists component 50 can present predictive technical indicators to a trader in a grid with one market on each row and one indicator in each column. The charts component 52 can present the predictive technical indicators to a trader in graphical chart format. The grids component 54 can present the predictive technical indicators to a trader in a grid with one-time period on each row and one indicator in each column. The reports component 56 can present the predictive technical indicators to a trader in a variety of report formats. The intermarket analysis component 58 can use the intermarket relationships stored in the predictive trader database 64 to generate inputs to the neural networks. The technical analysis component 60 can use financial time series data to calculate technical indicators for use as inputs to the neural networks. The predictive technical analysis component 62 can calculate predictions of future values of the time series data. The predictive trader database 64 can store data used by the trading application 14.

A method for producing predictive technical indicators is illustrated in FIG. 4. In a first step 70, financial time series data is analyzed. In a second step 72, a series of neural network predictions is designed which represents the financial time series data one or more periods in the future. This step 72 can include one or more (e.g., 1, 2, 3, 4, 5, or more) neural network training sessions which can involve using data from different time periods (e.g., all historical, previous five years, previous 1 year, previous 30 days, previous 10 days, previous 5 days, etc.) as inputs to train and retrain a neural network. Intermarket analysis is then performed using the trained neural network in a third step 74 to enhance the accuracy of the predictions. The technical indicator calculations are analyzed in a fourth step 76, and the neural network predictions are then integrated with the technical indicator calculations in a fifth step 78. Finally, in a sixth step 80, predictive technical indicator algorithms are used to produce predictive technical indicators.

As one example, in the first step Analyze Time Series Data 70, the financial time series data can come in two forms, real-time quotes and historical bars. A quote can be for a bid to buy, an ask to sell, or a trade which represents a buy and sell transaction. Each quote has a price and a quantity. An example of a quote would be a bid to buy 500 shares of XYZ stock at $30 per share. Historical bars summarize quote activity over a period of time such as a minute, hour, or day. Each bar contains four prices and one or two quantities. The four prices are open, high, low, and close. The open price represents the price of the first trade of the instrument during the time period covered by the bar. The high price represents the highest price the instrument traded at during the time period covered by the bar. The low price represents the lowest price the instrument traded at during the time period covered by the bar. The close price represents the price of the last trade of the instrument during the time period covered by the bar. The volume represents the quantity of instruments traded during the time period covered by the bar.

In the second step, Design Predictions 72, as exemplified in Table 1 below, predictions can be defined as daily bars, hourly bars, and minute bars; however, any duration from real time quotes, such as up to monthly bars, could be used. Within each duration, the four components of a bar are predicted-open price, high price, low price, and close price. In addition, in the case of commodities, the next bar's volume and open interest could be predicted as well. Finally, the short, medium, and long term price trends are predicted. These could be expressed as moving averages of the price or any technical indicator. The trends could be calculated using any of the price components or a combination of price components, such as typical price which can be the average of high, low, and close.

TABLE 1 Daily Predictions 1. Predicted Open Price Next Day 2. Predicted High Price Next Day 3. Predicted Low Price Next Day 4. Predicted Close Price Next Day5. Predicted Volume Next Day 6. Predicted Short Term Daily Trend 7. Predicted Medium Term Daily Trend 8. Predicted Long Term Daily Trend Hourly Predictions 9. Predicted Open Price Next Hour 10. Predicted High Price Next Hour 11. Predicted Low Price Next Hour 12. Predicted Close Price Next Hour 13. Predicted Volume Next Hour 14. Predicted Short Term Hourly Trend 15. Predicted Medium Term Hourly Trend 16. Predicted Long Term Hourly Trend Minute Predictions 17. Predicted Open Price Next Minute 18. Predicted High Price Next Minute 19. Predicted Low Price Next Minute 20. Predicted Close Price Next Minute 21. Predicted Volume Next Minute 22. Predicted Short Term Minute Trend 23. Predicted Medium Term Minute Trend 24. Predicted Long Term Minute Trend

The neural networks do not have to predict the desired future price directly. Instead, they can predict any information which can be used to calculate the desired future price. For example, they could predict the change in price from a current or past period to the future period. As another example, they could predict volatility or probability of the price going up or down and that information could be used in a calculation which produces the predicted future price.

How far in the future useful predictions can be made will vary from market to market? It will also vary depending on the current state of changing market conditions. In a preferred embodiment, a goal is to maximize the amount of future information captured by the predictions while minimizing the amount of error introduced. The error of the predictions can be measured across a variety of markets and conditions to determine which ones should be integrated with the technical indicator calculations and how heavily they should be weighted. In addition, the amount of lag inherent in the technical indicator calculation should be factored in as the purpose of the predictions is to overcome the lag.

The third step, Perform Intermarket Analysis 74, processes financial market data to identify, classify and grade relationships between financial markets and/or financial market segments. Neural networks use these relationships and historical data to generate predictions of future market prices. The fourth step 76 of analyzing technical indicator calculations can be applied to a wide variety of technical indicators. An illustrative example is the Simple Moving Average (SMA) technical indicator. A simple moving average is the average of a price series over a selected time period and gives an equal weight to each price during the period. To calculate the SMA, the sum of prices in the selected time period is divided by the number of prices in the selected time period. As a new price becomes available, it is added to the price series and the oldest price is dropped from the calculation, which allows the average to move over time.

In a hypothetical example, Table 2 below displays the daily close price with predictions of 100 troy ounces of gold traded on the CME Group exchange under the symbol GC during a selected month.

Column A indicates the trading day for the price data. Trading days are determined by the exchange and exclude weekends and some holidays. Column B indicates the close price, which is the final price the gold contract was traded on the day referenced. Column C is the 3-Day SMA Close, which is the 3-day simple moving average of the close price that is calculated each day by averaging the close price for that day and the two prior days. For example, on January 2nd, the close price was $879.50 and the close prices on the prior two trading days, December 31st and December 30th, were $884.30 and $870.00. The total price is $879.50+$884.30+$870.00=$2633.80. The average is calculated by dividing the total by three. $2633.80/3=$877.93. When the SMA moves forward to the next day, it is calculated by averaging the close prices for January 5th, January 2nd, and December 31st. This process is repeated until the moving average is calculated for all of the days in January. Column D is the 3-Day SMA Close Tomorrow. This column stores a 3-day simple moving average of tomorrow's close price.

This represents the perfect predictive version of this technical indicator. It is used as the basis for measuring the error of the technical indicators in the next table. Column E is the Close Price Tomorrow, which is simply the close price for the next day. It is used as the basis to calculate a sample prediction of the close price for the next day. A perfect prediction would equal this value. Unfortunately, it is not possible to predict the future with perfect accuracy. Column F, the Daily Price Change, is the amount the price went up or down from the previous day. This is used as the basis for the error in the prediction. The error will be some percentage of the daily price change, not a percentage of the price. Column G is the Random Error +/-25% of Change. This is a random value from -25% to +25%. This simulates the error between the perfect prediction and the real prediction. The error percent is multiplied by the daily price change to calculate the error amount. The error amount is added to the close price tomorrow to calculate the sample prediction for the day.

The chart in FIG. 5 displays the price of gold over three months using daily bars, a 5-day simple moving average, and a 10-day simple moving average. The thinner line represents the 5-day simple moving average while the thicker line represents the 10-day simple moving average. The moving averages have a negative effect called lag. Lag is a delay in periods between when a price change occurs and when the change is reflected in the moving average. The vertical cursor on the chart is positioned just after the bar for December 8th. The price of gold was going down prior to that and started going up after that. The bars, which have no smoothing effect from a moving average, reflected that change in direction on December 8th. The 5-day SMA reflected that change two trading days later on December 10th. The 10-day SMA reflected the upturn in the market even later on December 11th. The larger the number of periods in the moving average, the greater the lag. All moving average systems use historic data only, therefore will always suffer from lag and introduce uncertainty to the trader.

Maximum profit is achieved by buying at the bottom of a cycle and selling at the top of a cycle or vice versa. The delays caused by the lag of moving averages may cause a trader to react late to changes in the market, reducing profit or increasing losses. Techniques such as giving the recent periods more weight than the older periods can also be used to reduce the lag in moving averages. There is, however, a tradeoff between the beneficial smoothing effect of moving averages and the harmful lag effect. In general, anything which helps one will harm the other.

In the fifth step 78, predictions are integrated with technical indicator calculations. Every technical indicator is based on data for a range of periods. A predictive technical indicator can remove the oldest period from the beginning of the range and add the next period's prediction to the end of the range. The example shown in Table 3 below compares a technical indicator calculated using just the predictions with a predictive technical indicator that combines actual prices with predictions. The error is measured as the difference from a technical indicator calculated with the perfect predictions displayed in column D of the previous table.

Column H, the Predicted Close Price Tomorrow, is simulated by taking the close price tomorrow and adjusting it by random plus or minus twenty-five percent of the price change from the previous day. Column I is the 3-Day SMA Predicted Close, which is calculated by averaging the current predicted close with the prior two predicted closes each day. Column J, the 3-Day SMA Predicted Close Error, is calculated by subtracting the 3-day SMA of predicted close (column I in Table 3) from the 3-day SMA Close Tomorrow (column D in Table 2). Column K is the predictive 3-day simple moving average of close, which is calculated by averaging the predicted close price for tomorrow with the actual close prices for today and yesterday. Column L, the predictive 3-day simple moving average of close error, is calculated by subtracting the 3-day SMA of close tomorrow (column D) from the predictive 3-day SMA close (column K).

In the sixth step 80 of producing predictive technical indicators, predictions of the future price are integrated into the technical indicator calculation to reduce the lag without sacrificing smoothing. This can be illustrated by the following example. Assume the price is going up one dollar per day: Day 1-$10, Day 2-$11, Day 3-$12. A 3-day simple moving average calculated on day 3 would be (10+11+12)/3=$11. This represents one day of lag because the price was $11 on Day 2. If one assumes the price on Day 4 is correctly predicted to be $13, then the 3-day simple moving average on Day 3 using Day 2, Day 3, and Day 4 can be calculated as follows: (11+12+13)/3=$12. The moving average price on Day 3 then matches the actual price on Day 3, without sacrificing smoothing. In this example, integrating predictions into technical indicators reduces their negative side effects while retaining their benefits.

The predictive technical indicator method can, in some cases, also reduce the error in the result dramatically. Each prediction contains error. The technical indicator calculated in column J includes the error from three predictions. The predictive technical indicator in column L only uses one prediction. As a result, two-thirds of the error is eliminated. This is reflected in the error of column L, being one-third the error in column J. Most technical indicators use more than three periods worth of data, which can make the reduction in error even more significant.

FIG. 6: illustrates the primary components of the system in summary form. At 100, 102, 104, 106, and 108, the system's 5 advisors are shown, which are comprised of both machine learning components, common trading models in the field that are enhanced with embedded machine learning components, and non-leaming proprietary (to the applicant) and common trading models. At 110, a group of proprietary and common indicators and compound indicators are shown. All of these components, described in more detail later, produce outputs which are then combined by a neural net combiner as shown at 112, producing a final prediction as shown at 114.

In a preferred embodiment as illustrated by FIG. 6, raw time-series stock data is entered into the process at 2, where all raw data is stored in a database as shown at 4. At 6, the first process step uses mathematical indicators to pre-process the raw time-series data. Each of the stocks for which the system is producing a prediction has a minimum indicator value which is equal to the change over the prior closing price for each respective stock. Additionally, each stock has its own value for each indicator it is pre processed with.

At 8 and 10, the raw time-series data values and the indicator output values are shown as being entered into the Data Base 1, at 12. Data Base 1 stores all raw time-series data and indicator output histories for further use in subsequent processes by more complex components called Advisors, as described in more detail later.

Advisors comprise static or non-static mathematically based routines with embedded logic, which are generally more complex than the mathematical indicators used in the pre-processing of the raw time series database. In the context of embodiments of the present invention, static advisors do not have any learning function that causes changes in how the outputs are derived (i.e., they have fixed parameters), where non-static advisors have a degree of freedom generally governed by a learning mechanism and parameter ranges (e.g., as in a neural network). Different Advisors and combinations of advisors can have profound impact on the accuracy of predictions. Embodiments of the present invention employs specific implementations of machine learning components with unique proprietary enhancements described in more detail later.

As shown at 14, the Nearest Neighbor Advisor comprises a component that creates a vector of the input values, and using table lookup finds the vector of values in previous periods of time that is most similar (based on a selected distance metric) and "assumes" what happened then will happen again; thus, its prediction can be said to be reasoned by analogy with past data or "case based" reasoning. Usually the more periods the nearest neighbor has to consider the more reliable it will be. Embodiments of the present invention uses normalized indicator values (e.g., using percentage moves rather than raw values, and standard deviations to normalize the size of moves) to allow case data on different stocks to be relevant candidates for the current query. For example, what IBM did on May 22, 1998 may be viewed as a relevant case for predicting the MEX (Mexican Stock Index) on Jun. 11, 2004, if their normalized indicator value vectors are similar

As shown at 16, a Decision Tree Advisor, is informally a conditional series of "tests" that are applied to the input, where depending on the outcome of the tests (a path through the tree), a prediction is made. Given n samples of prior instances of the classification path of the data as seen in the input history, the system uses a traditional "minimum entropy" heuristic that attempts to approximate the smallest "explanation" of the data over that period. For example, a small decision tree may, by way of example only, look like the following: If 13mvag is > close if 23ema is < high then expect 2.2% gain next period (5 samples) else expect 0.1% loss next period (2 samples)else if up 3 days in a row expect 4.5% drop next period (1 sample) else expect 0.5% gain (7 samples).

Embodiments of the present invention comprises the use of decision trees in a manner to identify and then possibly "mimic" or "fade" what it expects other trading systems may have discovered about the current period. To mimic means, take prediction as is explained by decision tree as a final output, and to fade means to multiply its prediction by negative 1.

Additionally, which tests are asked of the data depends on the outcome of their parent tests, thus producing a tree structure. Unlike conventional use of decision trees used in the art, which utilize just one back-tested static tree forward in time, embodiments of the present invention continually create new decision trees for each new period (e.g., each day). Further, the decision trees operate on normalized data, such as like the data produced from the implementation of nearest neighbor, in order to allow rules to be learned across differing types of data, e.g., individual stocks and stock indices.

As shown at 18, the Bob Advisor combines all of the indicator outputs to create an intermediate prediction for each respective stock. The Bob Advisor is an example of a static advisor because is not adaptive, treating every stock the same given a historical record of indicator values. Note it takes each stock's data and computes a "score" which is based on the applicant's personal heuristics. The score is initialized to 0 and then adjusted for each rule. For example, if the 5 Day Average Facilitation is greater than the 34 Day Average Facilitation the score is increased by 1. If the 18 Day Average is greater than the Day Average the score is increased by 5.0 else decreased by 5.0 etc. Finally, the total score is normalized into a range of -3.0% and 3.0% which represents expected change in each respective stock's price as of the next day's Close.

As shown at 20, an embodiment of the Retracement Advisor is based upon Joe DiNapoli's published trading methods which makes use of specific settings for MACD (moving average convergence divergence using several different moving averages each with different period parameters) stochastic indicators and delayed moving averages to generate a buy and sell signal of varying strength. The Retracement Advisor is designed to mimic the behavior of day traders who are following traditional stochastics and moving averages on their trading screen, thus exploiting any impact on stock price formation that results directly or indirectly from large populations of market participants using the same common trading signals. The unique implementation comprises the use of only a subset of the published method, excluding Fibonacci support and resistance levels. Rather, only the formulas (i.e., not chart patterns) are used, to produce a magnitude prediction rather than a trading signal, which is done by normalizing the trading signals the formulas produce and resealing the outputs to be within the range of typical market movements as measured by standard deviations.

As shown at 22, the Complex Retracement Advisor comprises a machine learning mechanism embedded into a traditional Fibonacci retracement analysis system. The

Complex Retracement Advisor foregoes the assumptions of stock price retracements (i.e., rebounds) based on Fibonacci ratios (e.g., 0.382 and 0.500 and 0.618 ratios) of the most recent trends, and learns non-linear effect of a stock's price reaching a support level (the price a stock trades at or near, but does not go lower than, over a certain period of time, e.g., the floor) and resistance level (the price a stock trades at or near, but does not go higher than, over a certain period of time, e.g., the ceiling). That is, rather than assuming the traditional ratios hold true, it learns what actually happens. Resistance and support price levels are defined as prices at which short-term trends changed. The Complex Retracement Advisor is a non-linear neural network (specifically, a multi-layer gradient descent with 100 non-linear interior nodes representing products of proprietary variables). It learns a non-linear combination of the 3 most recently identified support levels and the 3 most recently identified resistance levels and attempts to predict the next daily change in a stock's Closing price.

At 14, 16, 18, 20, and 22, the Advisors that are shown further process the indicator output data stored in Data Base 1, producing output values that are representative of each Advisor's respective prediction for the next day's closing price. At 24, the outputs of all Advisors are entered into the second database called UPD, shown at 26. UPD Neural Net Combiner, shown at 34, is responsible for the next step in the prediction process. This Combiner is a neural net which reviews all of the new Advisor predictions for each stock's closing price, and then compares them to the actual closing prices stored in Data Base 1, updating the weights for each Advisor (each stock has negative and positive weights for each advisor), which weights are stored in a table in UPD as shown at 28. The weights represent what the Combiner has learned (i.e., its memory) about the accuracy of the Advisor predictions, where the final prediction for each respective stock is a learned linear combination of all advisor outputs for that stock. The Combiner comprises a traditional gradient descent neural network that attempts to learn a linear combination of its input weights to produce predictions that minimize their error. In the context of embodiments of the present invention, the Combiner creates an output which is a linear combination of all Advisor predictions for each respective stock.

Some portions of the detailed description are presented in terms of algorithms and symbolic representations of operations on data bits within a computer memory. These algorithmic descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. An algorithm is here, and generally, conceived to be a self-consistent sequence of operations leading to a desired result. The operations are those requiring physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated. It has proven convenient at times, principally for reasons of common usage, to refer to these signals as bits, values, elements, symbols, characters, terms numbers, or the like.

It should be borne in mind, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise as apparent from the following discussion, it is appreciated that throughout the description, discussions utilizing terms such as "processing" or "computing" or "calculating" or "determining" or "displaying" or the like, refer to the action and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system's registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission or display devices.

The present invention, in some embodiments, also relates to apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, or it may comprise a general purpose computer selectively activated or reconfigured by a computer program stored in the computer. Such a computer program may be stored in a computer readable storage medium, such as, but is not limited to, any type of disk including floppy disks, optical disks, CD-ROM's, and magnetic-optical disks, read-only memories (ROM's), random access memories (RAMs), EPROMs, EEPROMs magnetic or optical cards, or any type of media suitable for storing electronic instructions, and each coupled to a computer system bus.

The algorithms and displays presented herein are not inherently related to any particular computer or other apparatus. Various general purpose systems may be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatus to perform the required method steps. The required structure for a variety of these systems will appear from the description below. In addition, the present invention is not described with reference to any particular programming language, and various embodiments may thus be implemented using a variety of programming languages.

From the foregoing, it will be appreciated that specific embodiments of the invention have been described herein for purposes of illustration, but that various modifications may be made without deviating from the spirit and scope of the invention. In some instances, reference has been made to characteristics likely to be present in various or some embodiments, but these characteristics are also not necessarily limiting on the spirit and scope of the invention. In the illustrations and description, structures have been provided which may be formed or assembled in other ways within the spirit and scope of the invention.

In particular, the separate modules of the various block diagrams represent functional modules of methods or apparatuses and are not necessarily indicative of physical or logical separations or of an order of operation inherent in the spirit and scope of the present invention. Similarly, method have been illustrated and described as linear processes, but such methods may have operations reordered or implemented in parallel within the spirit and scope of the invention.

The foregoing description of illustrated embodiments of the present invention, including what is described in the Abstract, is not intended to be exhaustive or to limit the invention to the precise forms disclosed herein. While specific embodiments of, and examples for, the invention are described herein for illustrative purposes only, various equivalent modifications are possible within the spirit and scope of the present invention, as those skilled in the relevant art will recognize and appreciate. As indicated, these modifications may be made to the present invention in light of the foregoing description of illustrated embodiments of the present invention and are to be included within the spirit and scope of the present invention.

Thus, while the invention has been described herein with reference to particular embodiments thereof, a latitude of modification, various changes and substitutions are intended in the foregoing disclosures, and it will be appreciated that in some instances some features of embodiments of the invention will be employed without a corresponding use of other features without departing from the scope and spirit of the invention as set forth. Therefore, many modifications may be made to adapt a particular situation or material to the essential scope and spirit of the present invention. It is intended that the invention not be limited to the particular terms used in following claims and/or to the particular embodiment disclosed as the best mode contemplated for carrying out this invention, but that the invention will include any and all embodiments and equivalents falling within the scope of the appended claims.

Claims (5)

WE CLAIM

1. Our Invention "IUML- Stock Prices Predictor" is a suite of predictions is defined to model the financial data commonly used to calculate technical indicators one or more periods in the future. The invented technology using machine learning and neural networks are trained to make these predictions and the predictions are then integrated with the standard technical indicator calculations to produce predictive technical indicators which are superior because they lead more and lag less. The invented technology is also apparatus and method for a stock investment method with intelligent agents is described and illustrated. The invented technology is a stock prediction system that through experience learns to make money based on short-term stock predictions and due to inherent flexibility continues to be profitable in virtually all market environments.

2. According to claims# the invention is to a suite of predictions is defined to model the financial data commonly used to calculate technical indicators one or more periods in the future.

3. According to claiml,2# the invention is to technology using machine learning and neural networks are trained to make these predictions and the predictions are then integrated with the standard technical indicator calculations to produce predictive technical indicators which are superior because they lead more and lag less.

4. According to claim,2,3# the invention is to apparatus and method for a stock investment method with intelligent agents is described and illustrated.

5. According to claim,2,4# the invention is to the invented technology is a stock prediction system that through experience learns to make money based on short term stock predictions and due to inherent flexibility continues to be profitable in virtually all market environments.

FIG. 1: IS BLOCK DIAGRAM ILLUSTRATING AN INFORMATION PROCESSING SYSTEM OF THE INVENTION.

FIG. 2: IS BLOCK DIAGRAM ILLUSTRATING THE ARCHITECTURE OF A PREDICTIVE SERVER.

FIG. 3: IS BLOCK DIAGRAM ILLUSTRATING THE COMPONENTS OF A TRADING APPLICATION FOR USE IN THE INVENTION.

FIG. 4: IS FLOW CHART SHOWING A METHOD OF GENERATING A PREDICTIVE TECHNICAL INDICATOR.

FIG. 5: IS A CHART DISPLAYING THE PRICE OF GOLD OVER THREE MONTHS USING DAILY BARS, A 5-DAY SIMPLE MOVING AVERAGE, AND A 10-DAY SIMPLE MOVING AVERAGE.

FIG. 6: ILLUSTRATES RELATIONSHIPS BETWEEN AN EMBODIMENT OF AN APPLICATION AND VARIOUS OTHER MODULES, DATA STORES, AND INTERFACES, SUCH AS MAY BE EMBODIED IN A MEDIUM OR IN MEDIA. (PRIOR ART)

FIG. 7: ILLUSTRATES AN EMBODIMENT OF AN APPLICATION UTILIZING INTELLIGENT AGENTS. (PRIOR ART)