Systems and methods for optimizing trade execution

Abstract

To optimize trade execution.SOLUTION: There are provided systems and methods for optimizing trade execution by performing the steps of: computing market reaction to recent trades of securities; calculating matching parameters for the securities in response to the computed market reaction and at least one of historical market data and real-time market data; calculating a trade window for a next match; and executing a trade during the window.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application is a continuation of U.S. Provisional Patent Application No. 62 / 566,789, filed October 2, 2017. No. 60 / 019,599, filed on May 1, 2006, the contents of which are incorporated herein by reference. [Background technology]

[0002] The efficiency of public markets can have a significant impact on investors in securities. Every penny of additional trading expense can make a difference for large traders such as mutual funds, pension funds and hedge funds. These transaction costs can cost brokers hundreds of millions of dollars per year. This is inevitably passed on to customers and business partners, directly reducing investor returns. Compounded over decades, even such relatively small transaction inefficiencies can , the total will become so large that everyone who owns stocks in the market, from individual retirees to the entire economy, Affects people.

[0003] Stock exchanges play a vital role in facilitating transactions between market participants. Mechanically, an exchange matches each investor's buy and sell orders and completes the transaction. The matching process reports on which orders are matched. Eligible for a particular security and subject to a set of official rules governing when it is available. Efficiency depends on how well such matching rules are designed and implemented. do.

[0004] In exchanges using the Limit Order Book (LOB) method, market participants Places an order to buy or sell a security at a specified price in a specified quantity. An order to buy a security at a specified price is called a "bid." An order to sell a security at a specified price or above for a specified quantity An order is called an "offer." The bid with the highest price is called the "highest bid," and the lowest An offer with a price is called a "low offer." During a booming market, there are many different prices. At any particular time, all bids and offers may be The set of export prices and their total quantities at those prices is the state of the LOB.

[0005] During an active trading day, market participants bid and offer for each security at various times. By issuing securities, the market generally provides an opportunity for immediate access to securities. This is known as liquidity. Market participants wishing to trade by placing an order can place an order immediately at the best available price. You can also place orders. These are known as market orders.

[0006] In LOB-based systems, the highest bid price is typically higher than the lowest offer price. Otherwise, the orders may be matched and traded. The size of the trade is the maximum quota available for matching, i.e. the maximum bid and the minimum bid. After the transaction is completed, the highest bid and The quantity of the low offer is effectively reduced by the size of the trade. This continues until the amount of matching available high bids or low offers is exhausted. After all available bids / offers are exhausted in the LOB-based system, the gap is It lies between the bid price and the lowest offer price.

[0007] Allows trades to occur as soon as bids and offers can be matched during the trading day The LOB method or system for achieving this is Continuous Limit Order B The order book (CLOB) is a continuous limit order book. Matching occurs at specific times during the trading day. The LOB method or system for limiting the occurrence of Discontinuous Limit It is called the Order Book (DLOB, discrete limit order book).

[0008] The most common rule set established in the US securities market is the CLOB implementation. and is generally optimized for directness of matching and speed of execution. One advantage is that it may allow market participants to "quickly price" in new information. Examples of such information include corporate earnings updates, government economic releases, and Recent financial market activity, news specials, and other events that have a specific impact on security prices. CLOBs also typically work well for smaller (retail-sized) orders. , allowing relatively smaller sized bids and offers to be matched within microseconds. To do so. [Prior art documents] [Patent documents]

[0009] [Patent Document 1] U.S. Patent Application Publication No. 2005 / 0015323 Summary of the Invention [Problem to be solved by the invention]

[0010] However, CLOB-based systems are used by 401(k) plan managers and mutual funds. Institutional investors and other investors seeking to trade relatively large amounts of securities often CLOB-based systems do not match orders as quickly as Certain features of CLOB-based systems that make this possible also have adverse side effects. That is, certain market participants develop an information advantage over other market participants. allowing them to trade based on that information to the detriment of another market participant. The existence of additional bids and offers based on that information advantage may be an additional While this type of trading creates transactions that provide liquidity to specific securities, it also creates They may impose significant costs on institutional investors seeking to trade large volumes of securities.

[0011] Examples of additional costs that are particularly harmful to institutional investors involved in CLOB-based systems include: Examples of this include "adverse selection," colloquially known as "being singled out." Adverse selection occurs when another party (an "asymmetric counterparty") chooses to counter an investor's limit order for a security. and then, just before the price of the security is about to move, i.e., the market is This occurs when a transaction is made immediately after the release of information that would move the market. Typically, short-term traders who use accurate short-term statistical price forecasts, e.g., High-frequency traders use price forecast models to predict changes in market prices. For example, by placing an order, the investor may be able to modify or delete their order faster than the investor can. It matches relatively large orders from investors and then predicts price changes as they occur. When placing a trade, they will replicate the original order of the investor to secure profits. , the asymmetric counterparty profits at the investor's expense.

[0012] Adverse selection may be measured, for example, by the average change in price after a trade occurs. Radar buys stock at $100 / share and immediately after the purchase, the price of the stock drops to $95. If the price falls by $0.50, the trader may conclude that the $5 price difference represents some other superseding market factor. This may be considered adverse selection, which hinders the factors.

[0013] An alternative market design would implement a DLOB-based system, where order matching is , occur at scheduled times, rather than continuously, during the trading day. This has been attempted since the 1980s with considerable success. For example, having matching rounds occur at specific times reduces short-term adverse selection. However, this often creates liquidity problems, i.e., waiting until the next matching round. Orders that have been postponed will fail to match with orders that expire during the postponement.

[0014] Traditional DLOBs are typically updated every 100 milliseconds, every 5 seconds, or another time interval, for example. Some existing DLOBs match every 250-500 milliseconds. However, the DL for matching is slightly randomized by the Since the OB time interval does not depend on trading dynamics, it is usually For some securities, it is too rare, and for other securities, it is too frequent. The more frequent the market, the more liquid it becomes for that security, but the greater the adverse selection. Any calibration, especially matching frequency per security, volatility regime, price difference, time, Lacking dynamic calibration for other reasons, existing DLOBs are commercially unsuccessful. It was a success.

[0015] Another example of market inefficiency for institutional investors is the pricing of securities in response to securities orders and trades. This is known as "market influence." Institutional investors place orders and When involved in trading in exchanges and alternative trading systems (ATS), some Market Participants may submit orders (whether as bids or offers or a combination of the two) by detecting patterns in stock offerings and changes in security prices over time. Such participants are often then able to predict the direction of orders. cancel or adjust its own orders or anticipated agencies acting on the basis of The result is that institutional investors can trade their own positions in securities. Those market participants will receive poorer matches for statements. Effectively profit from investor expenditure. [Means for solving the problem]

[0016] One embodiment of the present invention creates a more efficient securities market, reducing adverse selection and investor bias. Reduce both market influence and simultaneously calibrate and control the matching engine rule set This allows for maximum liquidity through the use of novel machine learning to determine the According to an embodiment of the present invention, machine learning is used in the control loop to continuously monitor market and match Adjust the matching time window by incorporating new data from the matching engine. This results in better matching. Machine Learning Engine MLE (Machine Learning Engine) uses several parameters based on the operator's priorities. The data can be optimized.

[0017] For example, according to one embodiment of the present invention, machine learning is used to Calculate the optimal scheduled matching time by combining the advantages of OB, That is, the matching time should be long enough for each security to offer maximum liquidity. and also to create adverse selection for investors with relatively large orders. The transaction was completed in time to make it unprofitable for another market participant to systematically execute it. The net effect is experienced by investors with relatively large orders. Alternatively, MLE may be able to partially fill orders. to the matching engine to reduce market influence and minimize adverse selection. You may give instructions.

[0018] Another embodiment of the present invention may improve order book execution by certain means, The stage is the number of relatively larger orders of the investor for each match. By choosing sizes and prices that provide less information to the market, if possible , making it more difficult to predict the actual size and price of the order, The aim is to reduce the market impact when the

[0019] According to a further embodiment of the present invention, there is provided a method for optimizing trade execution, the method comprising: , calculating a market reaction to a recent transaction of the security; and At least one of the historical market data for all securities and real-time market data for all securities. calculating matching parameters for the securities in response to one of the and executing a transaction in the security in accordance with the parameters.

[0020] In accordance with yet another embodiment of the present invention, a system for optimizing securities trade execution comprises: a processor for calculating market reactions to recent transactions of the securities, and and real-time market data. A machine learning engine for calculating all matching parameters and a matching parameter and a matching engine for executing trades in securities in accordance with the [Brief explanation of the drawings]

[0021] [Figure 1] 1 is a system architecture according to an embodiment of the present invention. [Figure 2] FIG. 2 is a block diagram of a system and method according to another embodiment of the present invention. [Figure 3] 10 is a flowchart illustrating a method for calculating an optimal matching time according to a further embodiment of the present invention. [Figure 4] 10 is a flowchart illustrating a method for executing a trade in accordance with another embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0022] An improved order matching system and method is disclosed. In particular, but not by way of limitation, embodiments of the present invention may be used to match securities such as company stocks. Embodiments of the present invention are described with respect to systems and methods for trading such securities. Including, but not limited to, fixed income (e.g., corporate, government, special purpose), currency, options, Liberals and other financial instruments (e.g., loans, leases, mortgages, notes, commercial paper, etc.) useful for trading commodities, real estate, other real assets, digital assets, cryptocurrencies, and more. It may be implemented conveniently.

[0023] FIG. 1 illustrates a system architecture 100 according to an embodiment of the present invention. In this example, a computerized exchange 101 and a client device 105 are connected to a network. The computerized exchange 101 is also connected to the network 102. 02 or via a different network (not shown) to one or more data sources 1 03, each of the data sources being optionally connected to application programming The network 102 preferably also includes an API 104. By connecting the data source 103 with the client device 105, such device The data can be accessed from a data source 103 to a computer. The same as, similar to, or a subset of the data received by Exchange 101 It is.

[0024] The computerized exchange 101 preferably comprises one or more processes or applications. The server 108 may be configured to process transactions. and may be configured to interoperate via an internal network structure; Or, for example, a presentation server, a database server, an application server, and other associated servers configured together to implement aspects of embodiments of the present invention. The server 108 is preferably a general-purpose computer, such as a A cloud server, or distributed computing network.

[0025] The interface 107 is preferably a local API, a web API, or a program. Application Program Interfaces (APIs) such as the Instead, it is a network interface that connects a computer to a computer network. Virtual Private Network (VPN) is a virtual private network that connects a computer to a virtual controller or Alternatively, the interface 10 may be a network interface. 7 may be an interface application, The application provides a user interface for users to interact with the computerized exchange 101. For example, by providing the Services to third parties to place orders, monitor transactions, review market data, etc. .

[0026] The network 102 is preferably TCP / IP, FTP, UPnP, NFS or C A communications network that uses one or more commercial communications protocols, such as IFS. The network 102 may be wireless or wired, and the network may be a local area network. Network (LAN), Wide Area Network (WAN), Virtual Private Network ( VPN), Internet, Intranet, Extranet, Public Switched Telephone Network ( PSTN), mobile phone networks, satellite networks, infrared networks, and other This includes any type of wireless network, other, or a combination of the foregoing.

[0027] The data source 103 is preferably a market, exchange, and / or market reporting service, The market reporting service provides historical and current market information for, for example, securities, bonds, currencies, derivatives, and the like. The API 104 preferably provides real-time price and trading data. APIs, web APIs, or programmatic APIs, and alternatively, a network interface controller for connecting the is a virtual network interface that connects your computer to a virtual private network. Alternatively, the API 104 may be an interface application. The interface application may be a user interface. provides a user with a data source 103 to interact with one or more data sources 103, for example, to monitor transactions. Monitor market conditions and scrutinize market data.

[0028] The client device 105 is preferably a conventional order management system (OMS) or A personal computer running or part of a traditional executive management system (EMS); A conventional computing device such as a tablet computer or a smartphone. Client devices 105 preferably allow individual users to place orders and monitor transactions. Provide a user interface for reviewing market data, reviewing account status, etc. The client device 105 optionally places orders and receives information about order status. Alternatively, this may include an internet browser or mobile application for , the client device 105 may, for example, purchase securities, bonds, currencies, derivatives, and / or other and one or more computers that execute trading algorithms, etc. for trading other .

[0029] In preferred operation, a user accesses one or more client devices 105. The client device 105 processes the order via the network 102. Optionally, the user may send the data to a computerized exchange 101 via a network 102. Access market data from 103. Computerized Exchange 101 is a network 1 02 receives historical and current market data from data source 103. Orders received by the exchange 101 are subject to a matching process by the server 108. After the match is performed and the order is filled, information about the filled order is via network 102 to a data source 103 and / or to another market participant, etc. (not shown) ) will be sent.

[0030] FIG. 2 is a functional block diagram of a system or method according to an embodiment of the present invention. A preferred embodiment of the digitalized exchange 101 is shown as trading system 205. The system 205 includes a machine learning engine 206, a market reaction module 207, and a matching and a machine learning engine 208. Preferably, the machine learning engine 206 is a real-time Market data 201, past order data 202, past market data 203, and market reaction data and the response data 207, and then use each to execute the matching engine 20 8. The matching engine 208 calculates the order matching parameters provided in , by matching real-time orders 204 using matching parameters. , minimizing adverse selection.

[0031] The real-time market data source 201 may include, for example, securities, bonds, currencies, derivatives, and other Providing real-time market data on price, size, timing, and more. Such data may be provided by a stock exchange, its alternative trading platform, or another trusted market. The historical order data source 202 may be obtained from any of the following data sources: securities, bonds, , currency, derivatives, and other orders, including price, size, timing, and other historical data. Historical market data sources 203 provide data on, for example, securities, bonds, currencies, derivatives, Other historical market data on price, size, timing, etc. The market data provided to the machine learning engine 206 is preferably time-varying (historical) ) and the published prices of all orders and transactions that are flowing during trading (real-time) and the market price of such orders and transactions. and summary statistics calculated from the market data. Such statistics may include, for example, price volatility. ,changes in price gaps, buy / sell order imbalances in active markets,size changes between Timing, last trade size, ratio of trade size to order book size, and the book size at the time of the trade The ratio of the transaction price to the price of the stock may also be included.

[0032] Data sources 201, 202 and 203 are preferably data sources 103. The time order 204 may preferably be provided by a client device 105 .

[0033] The market reaction module 207 may also be used to determine if the most recently filled order is a traded item, e.g. How did the transaction affect the market price of the security traded at a specific time after the transaction was completed? As a simple example, the amount of a traded item is quantified in terms of another superseding market price. When there is no reason, the transaction increases or decreases after it is considered to be the market impact of the transaction. In a market reaction implementation, multiple price movements in response to multiple trades are used to calculate the performance of the market reaction. Optionally, a market reaction module may be included to identify and quantify market impact. Rule 207 quantifies the market impact of past orders on the market value of the traded item. In this case, past order data 202 and / or past market data 203 are used.

[0034] Preferably, the market reaction is determined, for example, by the order being submitted to the matching engine 208. or by the change in the ledger after a specific event, such as the execution of a specific trade. The difference between the ledger before and after these events may be measured in a number of ways. For example: For example, the difference in the ledger is the difference between the respective prices / levels / sizes in the bid and the corresponding amounts in the offer. Another example measure is the asking price / level / size ratio. This is a comparison of the weighted total number of shares offered for purchase against the weighted total number of shares distributed.

[0035] To track post-trade market reactions, the market reaction module 207 uses available on-site The site monitors changes in order price and size, and the site monitors the contents and related information of those order books. The market reaction module 207 preferably Examples of data that may be monitored include the timing, size and / or price of subsequent trades. In the market, for example, the market reaction module 207 preferably reacts to new high bids and new Tracks the best available bid and offers and displays the security's best available price for an instant transaction. Reflect.

[0036] The machine learning engine 206 and the matching engine 208 preferably operate in a single logical system. The machine learning engine 206 preferably operates as a system for optimizing the Matching parameters, such as when to enter a trade or series of trades The sentence should be matched, how many orders should be matched, and at what price. Optimize matching parameters for how matching should occur in the matching engine 2 Inform 08.

[0037] The term "machine learning" refers to the process of "training" a computer to learn a set of predefined functions. Training refers to the process of generating a desired output for a set of inputs. "Learning" involves systematically presenting examples of the All of this is integrated into a large-scale model of the relationships between inputs and outputs, and a set of new occurs when a person acquires the ability to predict the correct output with increasing accuracy depending on the input The accuracy of the output of the machine learning engine 206 can be calculated by, for example, by testing the output and measuring the "error" between the predicted output value and the "correct" output value. It may be quantified by

[0038] Many machine learning methods are known in the art. by the manner in which the data is presented to the computer and by the system This may vary depending on the type of training process the system uses to "learn." Examples of machine learning systems and methods include neural networks, regression, Bayesian These include the neural network method, the deep learning method, and the neural network algorithm.

[0039] In one embodiment of the present invention, the machine learning engine 206 processes the real-time orders 204 as data from orders submitted by the market, as well as other data contained in real-time market data 201 Preferably, the machine is trained on market data from exchanges and trading venues. The learning engine 206 uses real-time market data 201, historical order data 202, A combination of past market data 203 and real-time orders 204 is used to determine the order placement for a particular transaction. Create predictive models for items, e.g., securities, that are traded on historical and It may include the raw orders as well as adverse selection and market reactions observed after the trades are executed. The goal of the machine learning engine 206 is to create a predictive model, The model determines which matching times, order prices, and order volumes minimize adverse selection and market reaction. This will predict how the situation will develop.

[0040] In this context, adverse selection is preferably a series of events occurring after a transaction has been executed. It is measured by the price difference between the new market price of the security and the actual market price of the security. For example, ,Machine learning engine 206, AdvSel_at_time_0 = price_at_time_0 - trade_price AdvSel_at_time_1 = price_at_time_1 - trade_price … AdvSel_at_time_n = price_at_time_n - trade_price where n is the time step (e.g., microseconds) after the trade. ) and trade_price is the price at which the particular trade was executed. Alternatively, a top may count significant events such as a price change or another transaction. stomach.

[0041] "price_at_time_n" is the National Best Bid and d Offer (NBBO, National Best Offer), or calculated from the current and most recent listed prices. Some alternative statistical methods include Recent Volume Weighted Average Price (VWAP, the most recent volume-weighted average price) ) and Weighted-Mid Price (weighted sum price). According to the WAP method, "recent" is over the previous N transactions or over K periods. , or V is calculated over the nearest volume. Weighted-mid price method Therefore, prices are calculated using all available quote prices on exchanges and their order books. Alternative Trading Systems (ATS) that enable In this case, the contribution of each price to the result is calculated from the displayed Weighted by sentence size.

[0042] As the machine learning engine 206 processes more data over the trading day, matches The accuracy of the learning parameters should improve. Machine learning methods do not rely on any static rules. is superior to a set because it automatically adapts to changing market conditions while At the same time, they attempt to minimize adverse selection and / or market reaction. 206 matches orders more quickly on days of high volatility than on days of low volatility. A static rule set may be trained to adapt to the amount of volatility on a given day. Regardless of the order, orders will be matched at the same speed.

[0043] The machine learning engine 206 may, in accordance with an alternative embodiment of the present invention, implement a conventional machine learning algorithm. The preferred embodiment utilizes reinforcement learning and supervised learning methods. The method is used to create an optimal matching model for each security.

[0044] Using reinforcement learning methods, the machine learning engine 206 continually learns by itself using trial and error. The body is trained to make certain decisions by being exposed to an environment that trains it. The machine learning engine 206 learns from past experiences and attempts to make decisions. It learns which decisions produce better results. Examples of reinforcement learning methods include Markov decision There is a way to follow the process.

[0045] Supervised learning is a machine learning method that infers functions from training data. The training data consists of a set of training examples. Machine Learning Engine 2 Preferably, the method performs supervised learning using each training example to A training example is a set of input objects (typically vectors) and desired output values ("trainers"). The training process is performed by the machine learning engine 206 until it has tuned its model sufficiently to achieve the desired level of predictive accuracy. ,Continue. Examples of supervised learning include regression, decision trees, random forests, KNN, Other common methods of machine learning include, but are not limited to, logistic regression and logistic regression. The law is https: / / en.wikipedia.org / wiki / Machine _learning (last accessed October 2, 2017), It is incorporated herein by reference.

[0046] Another feature that may be advantageously implemented within the machine learning engine 206 in embodiments of the present invention is Machine learning methods include variations of deep learning methods. Learning (or hierarchical learning) is based on training data representations, as opposed to algorithms that perform specific tasks. Deep learning methods can be supervised, partially supervised, or unsupervised. It is also possible.

[0047] In another embodiment of the present invention, the machine learning engine 206 uses historical market data as input. real-time market data, statistics calculated from market data (current volatility, current Recent returns, recent trades, book pressure, trader pressure), and market reaction (each trade (and subsequent transactions) It receives, and as output, a match for when to perform the next match. The matching time range (lowest price, highest price), how many orders should be matched Matching size range (minimum, maximum), each note to participate in the matching Matching of future orders regarding how long sentences should be kept on the books Grediency targets (lowest price, highest price) and next matching orders In a further embodiment of the present invention, the machine learning engine 206 determines the random matching time, size, and and prices, reducing the ability of other market participants to predict its behavior. .

[0048] In preferred operation, the machine learning engine 206 is manually configured by an expert. The machine learning engine 206 starts in either an initial state or a random state. , then historical order data 202 and historical market data 203, as well as real-time Connected to market data 201 and used by matching engine 208 Each transaction begins generating matching parameters. After the transaction is completed by the market reaction module 208, the market reaction module 207 and the market impact data is provided to the machine learning engine 206. ,updates its internal model with new information, e.g., input-output pairs.

[0049] The machine learning engine 206 may be configured to minimize adverse selection and / or market influence and to optimize subsequent training. The learning process is repeated until a local optimum is found where no further iterations significantly improve the results. Optionally, the machine learning engine 206 then starts from a new state and continues learning. By continuing, new optima may be found. The algorithm also has a different representation of its state, its threshold for updating its learning. For example, in a neural network, the state of the learning process may be represented by a neuron. the weights in the links between them, the firing thresholds for each neuron, and and thresholding functions.

[0050] In a further alternative embodiment, the machine learning engine 206 may be configured to match the matching engine 208. By providing a switching parameter, you can specify partial fills for one or more orders. Diagrammatically involved.

[0051] The machine learning engine 206 updates its internal model continuously or at different times to The matching engine 208 may be provided with the best matching parameters. The matching engine 206 not only adapts quickly to new market conditions, but also It operates frequently enough to react to activity. Traditional matching logic cannot handle such changes. It does not fit the situation.

[0052] The matching engine 208 receives buy and sell orders and generates trades through a traditional ATS or For example, a "buy" order is placed at the immediately available price. It may be a "market" (to buy at a price below a certain limit) or a "limit" (to buy at a price below a certain limit). A "sell" order can be placed either in the "market" (to sell at a readily available price) or at a pre-determined limit. There may be a "limit" for selling at a price above which the match is made. Typically, the match is made by There is at least one buy order and one sell order with a price, and The matching engine 208 generates the matching result when the local adjustment for the matching result is satisfied. In the United States, Regulation NMS requires that a site be designated as a "protected" site with a custom price. NBBO, for example, if it is outside the exchange at the time of matching, it cannot be matched. Prescribe.

[0053] Orders that qualify for matching by price are paired in a designated order. Typically, matching is based on size priority, time priority, or proportional allocation. For example, an order must be in the "Resident" position in the order book to be eligible for matching. It may be required that the "cache" has a certain duration (e.g., a minimum number of microseconds). Thus, if an order is "too new" it may not qualify for matching. Instead, orders must meet certain criteria to be eligible for matching. A minimum size may be required. The minimum size may be determined by the rules of the exchange. It may be specified by the entity submitting the statement or by a matching algorithm.

[0054] FIG. 3 illustrates the optimal matching time for the machine learning engine 206 according to an embodiment of the present invention. 1 is a flowchart illustrating a method for calculating the optimal matching time. Calculations are based on publicly available data and real-time data submitted to the matching engine 208. This is done for individual securities based on the customer's order.

[0055] The published data and the order data submitted to the matching engine 208 are processed in step 30 This data is used to calculate the stock price at various times during the trading day. Calculate historical volatility (step 302). Volatility is a statistically significant change in the security's price. A measure of fluctuation, typically calculated as of a specific time. For each point in time, the instability may be calculated over various time periods. Used by the matching engine 208 to identify periods when instability is below a threshold. The optimal matching time for the security is calculated by:

[0056] FIG. 4 is a flowchart according to an embodiment of the present invention. The steps shown in FIG. Alternatively, the machine learning engine 206 and / or the matching engine 208 may First, the market reaction to the last trade of a particular security is calculated, and then, preferably, The range of any of the reverse selections is determined (step 401). Matching parameters include market reaction, historical market data, historical order data, real-time Calculated using order data and / or real-time market data.

[0057] One example of a matching parameter is the The optimal matching time is the time when the outstanding offers and bids are matched by the machine learning engine 206. Based on the matching parameters sent to the matching engine 208, Matched and filled, or partially matched and partially filled, or later In step 404, each match is postponed until the The matched order (or a portion of the order) is executed by the matching engine 208. The executed orders are then forwarded to the Trade Reporting Facility (TRF) by the matching engine 208. ), and sent to an exchange and / or another trading system (step 405).

[0058] The various implementations described above may be used in many different operating environments and for collection purposes for processing and storage. The present invention is applicable to one or more electronic devices incorporating the integrated circuit, chip. Appropriate configurations of hardware, software and / or firmware are available for trading in the marketplace. To improve the ability of a computer to interface with data, a method is disclosed herein. The systems or methods shown may also be used in conjunction with one another to perform the same functions disclosed herein. Some of the above exemplary systems operate in the same manner.

[0059] Most of the above exemplary embodiments use TCP / IP, FTP, UPnP, NFS and At least one communication network using one or more commercial communication protocols such as CIFS and The network 102 may be wireless or wired. The network is a local area network (LAN), a wide area network (WAN), a virtual private network (VPN), Private network, Internet, Intranet, Extranet, Public access Telephone networks, infrared networks, wireless networks, and combinations of the above networks It includes one or more of the following:

[0060] Examples of the present invention include data formed from various data stores and other memory or storage media. These components may be located on one of the servers, as described above. may be located within multiple or even within a network of servers. In certain embodiments, the information may be in a storage area network (SAN). Similarly, any content belonging to such a computer, server or other network device may be used for the purpose of Files to perform the functions described are stored locally and / or remotely as needed. Each of the above computing systems, including client devices, may The system may incorporate hardware elements electrically coupled via a control and / or power bus. For example, one or more processors in such a computing system may It may also be a central processing unit (CPU) for one or more of the ant devices. A client device is a device that contains at least one user device (e.g., a mouse, keyboard, a touch-sensitive display), and at least one It may further include an output device (e.g., a display, a printer, or a speaker). A client device may also include one or more storage devices, The device may include disk drives, optical storage devices, and random access memory (RAM) or This includes solid-state storage devices such as read-only memory (ROM), as well as removable media. This includes media devices, memory cards, flash cards, and more.

[0061] The computer system also includes, as described above, a computer-readable recording medium reader; Communication devices (e.g., modems, network cards (wireless or wired), or infrared communication The computer-readable recording medium reader may include a remote, Local, fixed and / or removable storage devices and computer-readable information temporarily stored and / or more permanently representing recording media for containing, storing, transmitting and retrieving The computer-readable storage medium may be connected to or configured to receive the computer-readable storage medium. Systems and various devices also typically include several software applications. application, module, service, or at least one operating memory device The other elements include the operating system and the client. This includes application programs such as mobile applications or web browsers. It should be understood that alternative embodiments may have numerous variations from those described above. For example, customized hardware may also be used and / or specific The elements of hardware, software (including portable software such as applets) It may also be implemented in another computer, such as a network I / O device. A connection to a routing device may also be used.

[0062] Recording media and other non-transitory computer readable media for carrying code or portions of code The retrievable medium may include any suitable medium known or used in the art. The media may include computer-readable instructions, data structures, programs, etc. Implemented in any method or technology for storage of information such as a system module or other data. This includes, but is not limited to, volatile and non-volatile, removable and non-removable media. Such media may include, but is not limited to, volatile and non-volatile, removable and non-removable media. RAM, ROM, EEPROM, flash memory or other storage technology, C D-ROM, Digital Versatile Disk (DVD) or other optical storage device, magnetic cassette , magnetic tape, magnetic disk storage or another magnetic storage device, or a storage medium storing the desired information another medium that may be used to Based on the disclosure and teachings provided herein, one of ordinary skill in the art would be able to It will be recognized that there are alternative aspects and / or ways to implement the embodiments.

[0063] The specification and drawings are to be regarded in an illustrative rather than a restrictive sense. However, it should be understood that any deviation from the broader spirit and scope of the present invention as set forth in the claims is expressly construed as a misrepresentation of the invention. Obviously, various modifications and variations can be made without departing from the scope of the invention.

Claims

1. 1. A method for optimizing trade execution in a trading system, comprising: The trading system receives real-time orders for securities from client devices. Tep and the trading system calculating market reactions to recent trades of the security; the trading system processes the market reaction and historical market data for the security and In response to at least one of real-time market data for the security, calculating a match time and a match size for the ticket; The trading system issues the certificate according to the matching time and the matching size. executing a transaction for the security; A method comprising:

2. The calculating step may include: , a machine learning step based on multiple calculated market reactions and multiple historical market data. Including 2. The method of claim 1 .

3. The calculating step may include: In addition, machine learning strategies are based on multiple calculated market reactions and multiple historical market data. Including top 2. The method of claim 1 .

4. The calculating step may include: , selecting the matching time and the matching size; 2. The method of claim 1 .

5. The calculating step may include: selecting the matching time and the matching size; 2. The method of claim 1 .

6. the matching time is a trading window; 2. The method of claim 1 .

7. The match size is a maximum threshold for a partial execution of an order.

2. The method of claim 1 .

8. The calculating step is performed by the trading system comparing the market reaction and the historical and real-time market data for the security. calculating a matching price for said securities; The executing step includes the trading system executing the securities in accordance with the matching price. including executing transactions on 2. The method of claim 1 .

9. the matching time is a matching residency time threshold; 2. The method of claim 1 .

10. 1. A trading system for optimizing securities trade execution, comprising: a processor for calculating market reactions to recent trades in the security; The market reaction and at least one of historical and real-time market data Calculate the matching time and matching size for the securities according to both. and a machine learning engine for receiving real-time orders for the securities from client devices; and a matching engine for executing a trade of said security in accordance with said match size. and, A system comprising:

11. The machine learning engine utilizes multiple calculated market reactions and multiple historical market data. The matching time and the matching size are calculated using the reducing adverse selection after the transaction; The system of claim 10.

12. The machine learning engine utilizes multiple calculated market reactions and multiple historical market data. The matching time and the matching size are calculated using the reducing the market impact after said transaction; The system of claim 10.

13. The matching time and the matching size are calculated, and the pre-matching time of the securities is calculated. Reduce adverse selection after the transaction The system of claim 10.

14. The matching time and the matching size are calculated, and the pre-matching time of the securities is calculated. Reduce the market impact after the transaction. The system of claim 10.

15. the matching time is a trading window; The system of claim 10.

16. The match size is a maximum threshold for a partial execution of an order. The system of claim 10.

17. The machine learning engine compares the market reaction with the historical market data for the security and the matching prices for the securities based on at least one of real-time market data for the securities; Calculate the rank, the matching engine executes the trade for the security in accordance with the matching price; The system of claim 10.

18. the matching time includes a matching residency time threshold; The system of claim 10.

Images