DK201300346A1 - Method for intelligent strategy development and testing - Google Patents

Abstract

A computer-implemented method, to automate the process of developing and testing trading strategies by restructuring the steps of strategy development, where the user can control an automated strategy development system using a remote computer connected to a centralized computer hosting the system for live automated strategy development. The method including the user can select a range of parameters, which is send to the centralized server, which perform a live backtest of a set of strategies which meet the parameters received, and returns a set of strategies to the user. The user can select a count of strategies and create a portfolio of strategies. The user sends a definition of the selected strategies to the centralized server, which are then traded automatically on the centralized computer system. One or more Templates may be used as the base of the automated strategy testing system for back testing strategies.

Description

Method for intelligent strategy development and testing The scope of the invention:

The scope of this invention is for the development and testing of strategies for auto trading tradable products, such as but limited to; stocks, Forex, CFD, options, ETF, bonds, betting products - traded on but not limited to; exchanges, banks, betting providers. Automatically trading strategies sending orders to a broker, where a broker can be any type of order receiver, such as but not limited to; brokers, exchanges of financial instruments, social networks, order receiving servers.

Traditionally, strategy development and testing requires programming. Lately visual methods have been made available to be used to construct a trading strategy. The present methods require that the user is actively involved in the development process, by either programming, or routing building blocks, or a combination. The user is required to decide which instrument (s) to use in the strategy, which timeframe (s) to use, the technical indicators and their parameters to use, what type of strategy to develop; such as trend following strategy, break-out strategy, scalping strategy - and how the strategy should be tested and later perhaps optimized.

This process is very time consuming, and requires that the trader has an in-depth understanding of the development platform he is using, including the learning curve and time used to get the in-depth knowledge of the development platform. The strategy developer needs to have in-depth knowledge of how to perform technical analysis, perhaps mathematical skills, how technical indicators work, how to read trading charts, the type of strategy types and their use.

If a neural network is used, the user needs to know about neural networks, he needs to define what parameters and other settings to use in the neural network, or the user even needs to develop a neural network himself. If data mining is used, the user needs to maintain the data mining, the user needs to have full understanding of data mining and how to use data mining.

Further, developing and testing trading strategies is a learning process, where the developer in the process gains experience of how to select the possibilities, what kind of criteria, parameters, timeframes and strategy types works, which does not work, and why. If using programming in development, the developer needs to have programming capability. If the developer uses visual development, the developer needs the persistence and time in the process of visually constructing a trading strategy by combining building blocks; US patent application 20100057600. The typical development time for one trading strategy is measured in months.

Common to the mentioned development methods is that a fairly large screen is needed, or several crrppn «; in nrrlpr tn in an nnprstinnal mannpr Hpvplnn thp Qtratpøip «; Thp rptail i i <; pr <; mn \ / p p \ / prv day more towards using mobile units with relatively small screens for their daily computing, which makes the mentioned methods of development not operational.

This invention:

The present invention relates to a computer-implemented method, to automate the process of developing and testing trading strategies, where the user can control an automated development system using a remote computer connected to a centralized computer hosting an automated strategy development system. Where the user can select a range of parameters, which is sent to the centralized server, which test or select a set of strategies that meet the parameters received. The centralized server returns a set of strategies to the user. The user may select a count of strategies and create a portfolio of strategies. The user may use or develop a technical indicator to filter periods of trading. The user sends a definition of the selected strategies to the centralized server, which trades the strategies automatically. A template may be used as the base of the automated strategy testing system for back testing strategies.

Description:

This invention typically consists of, but not limited to; a mobile computerized system connected to at least one centralized computerized system. The mobile computerized system may be, but not limited to; mobile phones, tablets, gaming devices, notebooks, tower PCs, a computerized television - hereinafter referred to as a "mobile device". The mobile device consists of, but is not limited to, at least one microprocessor, at least one physical storage device / media, a UI / GUI, a communication line, such as but not limited to; phone-net connection, Wi-Fi connection, Bluetooth connection, internet connection, cable connection The centralized computer system consists of but not limited to; minimum one microprocessor, minimum one physical storage device / media - and have minimum one connection to other computerized systems. This connection may be, but not limited to; connecting over an internet connection, using a Wi-Fi connection, Bluetooth connection, a cable, a fiber, a phone-net. Drawing 1.

The centralized computer system has the ability to automatically test sets of criteria and parameters, and perform a backtest on these, using one or more sets of historical data from one or more instruments or / and timeframes and their derived technical indicators. This backtest may be done "bar" by "bar", over a defined time period of historical data, by reading the data from a physical storage device / media and testing against the selected or / and received definition of criteria and parameters.

The method may use one computer for hosting both the functionality of the mobile computer and the centralized computer (s).

The centralized computerized system may perform the test for strategies using all selected criteria or using a subset of the available criteria, may randomly select among the available criteria. Such combinations may consist of any type of data and may be derived from any source of information, such as but not limited to; derived from data mining, from pattern recognition, from a neural network, from market news, from social media, from technical indicators, from charts of a financial instrument, from sections of programming lines or code or script, from existing trading strategies. The possible data sources, parameters, methods, indicators, charts, neural networks, other strategies, data mining, and criteria, which can be used by the automated development method is called a "Template". By using such development method, it is possible to conduct strategy development on a small-screen device, and makes it possible for users with no knowledge of programming, technical analysis or strategy development, to "on-the-run", quickly "develop "and use signal or automated trading trading strategies.

Both the centralized system and the mobile system may receive sets of data for financial instruments, and may format the received data sets into timeframes and technical indicators.

The work flow in this invention is typically, but not limited to; the user of the mobile device uses a computer-implemented method, connects to the centralized system and receives a definition of data, such as timeframes, instruments, criteria, indicators, where the user can select among a set of filtering parameters, such as but not limited to; the instrument to test, the period to test, minimum trades per. day, maximum drawdown, minimum total gain, minimum gain per day. day, minimum gain per day. Trade, parameters indicating bullish or bearish types of strategies the Template to use.

The mentioned filtering parameters may be selected separately from the actual strategy and thereby the user does not necessarily need to have any knowledge of technical analysis, of trading strategies, of technical indicators, of mathematical formulas, of trading charts, or of strategy development. Drawing 2.

The mobile device can send this set of filter parameters to the centralized computer, which performs a search or / and backtest using a Template. This backtest is done by reading / using a set of historical information, such as but not limited to tick data of a financial instrument, formatted tick data, technical indicators, timeframes, timeframe charts, neural network, data mining, red from a physical storage media, and backtesting using a set of criteria and parameters defined either by the system or by the user or a combination. During the backtest, all strategies measuring the filtering parameters sent by the user are stored on a physical storage device. When the backtest has ended, the centralized computer sends the strategies and related statistical information to the mobile device, by reading the stored data from the physical storage media. The mobile device receives the data, and stores this data on a physical storage device / media on the mobile device.

The centralized computer system may save the backtest result together with the combination of parameters on a physical storage device / media for later use. The computerized system may have pre-stored historical data, and the settings, criteria and parameters to use in the backtesting, which is later saved from a physical storage device / media. Each strategy may have a unique identification for later testing the exact strategy, and in order to later trade this specific strategy by having an identification.

The result received by the mobile device can be a partial result or the full result, for example sending the top 10 best strategies / results. The result can be statistics or / and a full equity curve. The user can by stepping through a list of results, request the equity curve from the centralized computer or calculate the equity curve locally, and present the equity curve on a display. The user can step through the list of results, and save the most promising results on a physical storage media for later use. The user can select a result / equity curve and send a request to a service provider or to the centralized computer to send an SMS or an Email when the selected strategy triggers a signal. Such service provider may be, but not limited to; a stock broker, a social network, a bank, a trading platform, a betting provider. The user can send the selected strategies to the centralized server, which will automatically trade the strategies. The user can select and change strategy settings, such as but not limited to; how long to run the strategy, to group several strategies in a portfolio, to exit a strategy or portfolio when a gain is above a limit, or when loss is below a limit.

The mobile device may have a more advanced section, where the user can select and send more specific parameters for the computerized system to use and backtest, such as but not limited to; use trailing stops, use partial profit taking, use money management, the level of risk for the strategy, which days to trade, which hours to trade, the quantity to trade.

The mobile device may have a section for signing up for an account with a provider, or for the user may start and stop trading strategies, or may send a request to the provider or the centralized system to enter a trade or to exit a trade, or set a price / time where an open position will be closed.

The centralized computer system may receive a live data feed, over an internet connection or other type of connection from one or more stock exchanges or from data providers of data feeds and news information. This data is stored on a physical storage device / media as they arrive. The stored data is later used for creating criteria and parameters, which the system uses to combine and test. Each combination tested can be considered a trading strategy. The received data feed may be formatted as timeframes, derived from technical indicators may be calculated, and saved on a physical storage device / media for later use by the backtesting.

The method of testing can include, but is not limited to; neural network, data mining, simple criteria and parameters, groups of criteria and parameters, using existing data mining data done in the past, using data from a neural network. The centralized computer system may be adapted to randomly select a count of criteria and parameters, which the system will use to search and test. By having the centralized computer randomly or in an ordered manner to select a set of criteria and parameters to combine, the user is freed from the actual development process.

The data feed stored may also be used when the centralized computer checks for sending messages or trade action when combinations are met. When the user requests to receive a message when a specific combination triggers a signal, the centralized system will send a message to the user. If the user has requested a service from the centralized computer or a service connected to the centralized computer, the signal can be traded without interference from the user. The message and trade information is stored on a physical storage media at all parties in order to keep a copy.

The mobile device system may have several levels of skills, for the user to select from simple, quick, skilled or advanced levels of settings, each level providing different filters to send the centralized server. The user may select one strategy or further optimize a set of strategies for the central computer. The user may select several instruments that are traded as one, hedging these strategies by sending this information to the centralized computer, which combines this set of strategies into one hedging strategy. The user may select strategies from different instruments, and define a portfolio

The mobile device may be adapted to have a touch screen, where the user can drag his fingers on the equity curve and thereby go through the equity curve / result received from the centralized server.

The mobile device may provide a feature where the user can send one or more backtesting results to, but not limited to; a social network, an email address, as an SMS destination, to a strategy store, where this strategy is stored on a physical storage device / media and can be sold or shared with others. The mobile device may be adapted to take a screen shot of the equity curve with is sent along with the strategy identification and result and stored on a physical storage device / media at the destination. The mobile device may be adapted to connect to a social network, to a strategy store, where strategy results can be retrieved from a physical storage media at the "source", download and save this information to a physical storage device on the mobile device, or on a physical storage device on a centralized computer where the mobile device has an account. The mobile device may be adapted to receive strategy results and / or their identification by SMS or email, and save this information on a physical storage media.

The mobile device may be adapted to draw an amount from the users account at any time a request for backtest is sent, or when the user requests the signal service via SMS or Email messages, saving this information on a physical storage media for later status .

The centralized computer system may serve a count of users, storing information from requests and backtest results to service on a physical storage device / media, in sections for each user.

The method may use different Templates, the user may select a template to use from a list of templates. The method may further include a development environment for defining and developing templates. A template can include any type of data and method. A user may use this development environment to define / develop his own personal template, which he can save on a physical storage device, or / and transfer the template to a server which stores the template on a physical storage device. The user can later select this template and use this template for automated strategy search and testing. The user may add his template to an app store or any place where other users can select and use this template. Template providers may develop templates for various uses; such as but not limited to, for his own use, for a group of users, or available to all users.

The user may select a technical indicator or define a technical indicator which is calculated on the basis of the equity curve of the portfolio of strategies, attempting to find a technical indicator which can indicate when the portfolio will be traded, and when the portfolio will be paused . An example of such indicator is show in drawing. 2 where an SMA is calculated using the equity curve of the portfolio, to be used by the centralized system to know when the portfolio should not trade, which could be for example when the SMA indicator has a negative slope. A count of technical indicators can be used as an indication of when to trade the portfolio. A definition can be sent to the centralized computer and stored on the centralized computer, which the centralized computer uses to calculate values at different time values based on the equity curve of the strategy or the full portfolio. Several standard definitions of such indicators can be avaliabe on the centralized computer, which the user can select, and perhaps even modify.

The method used on the users computer may receive price data, order data, information of the position of each strategy, from the centralized server, or other source, where the method is calculating the PLN of each strategy displaying this PLN and other information next to each strategy for the user to follow the automated trading of his strategies, drawing 3. The method may display equity curves on each strategy, or of full portfolios - the user may develop technical indicators on the basis of an equity curve, and send a definition of this indicator to the centralized server for filtering the periods where a strategy or / and portfolio is allowed. The user may select a strategy, and send a message to a centralized server to stop, pause, delete a strategy or a portfolio, drawing 5.

The user may choose to send information along with the selected strategies of how to trade the strategies, drawing 4, such as if to stop trading when a profit of 100, and top when a loss of 100.

Claims (18)

1. A computer-implemented method for automatically developing and automatically trading a plurality of trading strategies of financial instruments, wherein a user can develop trading strategies for automated trading said method including: receiving, by a centralized computer, a set of data for at least on financial instrument, wherein the set of data includes price data; connecting, by a user using a computer, to the centralized computer receiving a definition including at least one financial instrument and criterion; selecting, by a user using a computer, at least one financial instrument of a plurality of user selectable financial instruments; selecting, by a user using a computer, at least one criteria of a plurality of user selectable criteria; sending, by a user using a computer, a definition which is used to backtest or search for strategies to a centralized computer; back testing, but the centralized computer, a count of trading strategies using the received definition testing using a data set from at least one financial instrument; sending, by the centralized computer, at the definition of at least one trading strategy derived from the back testing; selecting, by the user using a computer, at least one of the received definitions; sending, by a user using the computer, a data set defining the selected definition to the centralized server; receiving and storing, by the centralized computer, the definition of the selected strategy; testing, by the centralized computer, each of the stored strategies at new time values using price data from at least one financial instrument; sending, by the centralized computer, trading orders to an order receiving server when a test indicates a trade;

2. The method of 1) wherein the centralized computer sends at least one Template to the user for selecting as Template to use during back testing.

3. The method of 2) wherein a Template consists of a definition of any of; timeframe to use, instruments which can be testing using the Template, technical indicators used in the back testing, chart to use, slippage to use, spread to use.

4. The method of 2) wherein a user can define a Template and register this Template at the centralized server as any of; for private use, for public use, for use of a group of users;

5. The method of 4) wherein the centralized computer stores the registered Templates, and manage the Templates, sending Templates to the users according to the defined user-priviledge of the Template.

6. The method of 1) wherein the back testing is performed on a data set testing a set of criteria for each time value of said data set, using the calculation of at least one technical indicator.

7. The method of 6) wherein the result of the backtest is send to the user as statistics of said backtest.

8. The method of 6) wherein a technical indicator is calculated for each time value using the data set from an equity curve using said technical indicator used to indicate when the strategy is allowed to trade.

9. The method of 6) wherein the equity curves of a set of strategies are added to one equity curve representing the joint equity curve, calculating the technical indicator for filtering trade indications using the joint equity curve of the portfolio of strategies to indicate when the strategies in the portfolio are allowed to trade.

10. The method of 2) where the centralized computer manage a plurality of portfolios for a plurality of users.

11. The method of 1) wherein the centralized computer test all stored strategies for new time values sending orders to an order receiving server, storing the order information.

12. The method of 11) where a user can connect to the centralized computer and receive status on all his strategies.

13. The method of 12) where the user can at least one of a) pause a strategy or a full portfolio b) delete a strategy or a full portfolio c) add new strategies to an existing portfolio.

14. The method of 12) where the user can receive stored a definition of at least one strategy and request a back test of at least one of the received strategies.

15. The method of 1) wherein the user can use a selector to enable and disable a parameter.

16. The method of 1) wherein the user can use at least one setting to select the value of a parameter.

17. The method of 1) wherein the computer of the user receives price data of at least one financial instrument used to update the status of the strategies and their PLN.

18. The method of 1, wherein the user can select which Template to use.