US 20070192230 A1
Match engine systems and methods that allow for rapid processing of orders are provided. Pending orders are arranged as a non-indexed collection of orders and may be stored in the cache memory of a processor. The physical locations of pending orders stored in the cache memory may correspond to the order in which they were received at a match engine. New orders are compared to orders in the non-indexed collection of orders in an attempt to match orders.
1. A computer-implemented method of matching orders for financial instruments, the method comprising:
(a) receiving at a match engine an order for a financial instrument; and
(b) comparing the order in (a) to a plurality of pending orders arranged in a non-indexed collection of substantially sequential orders located on a computer-readable medium.
2. The method of
(i) encoding the order before (b).
3. The method of
4. The method of
(ii) storing the encoded order in a memory module; and
(iii) at the second processor periodically checking for new orders stored in the memory module.
5. The method of
6. The method of
7. The method of
8. The method of
(i) formatting and distributing market data.
9. The method of
10. The method of
11. A match engine configured to match orders for financial instruments, the match engine comprising:
a computer-readable medium containing a non-indexed collection of substantially sequential pending orders; and
a first processor configured to sequentially compare a new order to at least a portion of the orders included in the non-indexed collection of substantially sequential pending orders.
12. The match engine of
a second processor configured to receive orders and encode the received orders.
13. The match engine of
a memory module coupled to the first processor and the second processor and that stores encoded orders.
14. The match engine of
15. The match engine of
16. The match engine of
17. The match engine of
18. The match engine of
a backup computer-readable medium containing a backup copy of the non-indexed collection of substantially sequential pending orders; and
a backup processor configured to sequentially compare a new order to at least a portion of the orders included in the backup copy of the non-indexed collection of substantially sequential pending orders.
19. The match engine of
20. The match engine of
21. The match engine of
22. The match engine of
23. A system for matching orders for financial instruments, the system comprising:
a first match engine comprising a first computer-readable medium containing a first non-indexed collection of substantially sequential pending orders and a first processor configured to sequentially compare a new order to at least a portion of the orders included in the non-indexed collection of substantially sequential pending orders;
a second match engine comprising a second computer-readable medium containing a second non-indexed collection of substantially sequential pending orders and a second processor configured to sequentially compare a new order to at least a portion of the orders included in the second non-indexed collection of substantially sequential pending orders; and
an output gateway coupled to the first match engine and the second match engine and configured to filter redundant messages.
24. The system of
25. The system of
26. The system of
27. The system of
The present application is a continuation-in-part application of U.S. Ser. No. 11/234,697, filed Sep. 23, 2005 and entitled “Non-Indexed In-Memory Data Storage and Retrieval,” the entire disclosure of which is hereby incorporated by reference.
The present invention relates to systems, methods and user interfaces that are utilized in connection with the trading of financial instruments. More particularly, the invention provides mechanisms for matching orders against a non-indexed collection of pending orders.
Modern exchanges must process and monitor a large volume of trading data, such as orders for financial instruments. Financial instruments include options contracts, futures contracts, options on futures contracts, interest rate products, securities, equity contracts, etc. Large exchanges are required to process and store large amounts of trading data every second of the trading day. Moreover, upon executing trades, exchange computers must continually access and distribute market data, which is another form of trading data. The distribution of market data facilitates necessary market-driven decisions. Indeed, risk analysis of current and future exposure is vital to ensuring a stable financial exchange. Often large databases are utilized to store and retrieve this trading data.
To select and aggregate trading data, conventional databases often use sorts, searches, indexes, and/or disc lookups. These requirements result in substantial chip clock cycles and lead to delayed query results. Current analysis systems utilized to aggregate large quantities of trading data are often executed in batch mode overnight because of the computing resources that are consumed by these activities. Often the aggregation and retrieval of trading data is not efficient enough to allow adequate information to be retrieved within the desired timeframe. Indeed, under traditional approaches, large amounts of trading data cannot be adequately analyzed in real-time, thereby preventing many uses of the data.
Prior art attempts to solve these problems have focused on building more intelligent indexes to speed up selection and analysis of the data stored within a database. Yet other systems have attempted to reduce response time to users through the use of precomputed summary data. These and other attempts to more efficiently store and retrieve trading data do not adequately address the problem. For example, precomputed indexes cannot be rapidly adapted for changing user needs or changing data. Additionally, precomputed data requires the user to specify the data that needs to be precomputed. When there is a need to analyze data from different angles or perspectives, these conventional systems fail to deliver results in a rapid fashion.
Existing order matching systems have similar drawbacks. As the volume of orders increases, existing order matching systems struggle to compare and match orders in an efficient manner.
Therefore, there exists a need in the art for systems and methods that allow for the efficient processing, storage and searching of large amounts of data in a time sensitive manner.
Aspects of the invention overcome at least some of the problems and limitations of the prior art by providing systems and methods that allow for the efficient storage and searching of large amounts of data. Pending orders may be arranged as a non-indexed collection of data records within one or more computer-readable media. Exemplary computer-readable media include processor cache memories, magnetic memories, hard disk drives, and optical disk drives. Solid-state memory modules allow for rapid queries due to the lack of moving parts, such as those associated with hard disk drives. The physical locations of pending orders stored in a computer-readable medium may correspond to the order in which the pending orders were received at a match engine and the order in which queries are performed. New orders are compared to orders in the non-indexed collection of orders in an attempt to match orders. Order matching may be performed without the speed limitations and overhead associated with indexed order collections.
Of course, the methods and systems disclosed herein may also include other additional elements, steps, computer-executable instructions, or computer-readable data structures. The details of these and other embodiments of the present invention are set forth in the accompanying drawings and the description below. Other features and advantages of the invention will be apparent from the description and drawings, and from the claims.
The present invention may take physical form in certain parts and steps, embodiments of which will be described in detail in the following description and illustrated in the accompanying drawings that form a part hereof, wherein:
Aspects of the present invention are preferably implemented with computer devices and computer networks that allow users to exchange trading information. An exemplary trading network environment for implementing trading systems and methods is shown in
An exchange computer system 100 receives orders and transmits market data related to orders and trades to users. Exchange computer system 100 may be implemented with one or more mainframe, desktop or other computers. In one embodiment, a computer device uses a 64-bit processor. A user database 102 includes information identifying traders and other users of exchange computer system 100. Data may include user names and passwords. An account data module 104 may process account information that may be used during trades. A match engine module 106 is included to match bid and offer prices. Match engine module 106 may be implemented with software that executes one or more algorithms for matching bids and offers. A trade database 108 may be included to store information identifying trades and descriptions of trades. In particular, a trade database may store information identifying the time that a trade took place and the contract price. An order book module 110 may be included to compute or otherwise determine current bid and offer prices. A market data module 112 may be included to collect market data and prepare the data for transmission to users. A risk management module 134 may be included to compute and determine a user's risk utilization in relation to the user's defined risk thresholds. An order processing module 136 may be included to decompose delta based and bulk order types for processing by order book module 110 and match engine module 106.
The trading network environment shown in
Computer device 114 is shown directly connected to exchange computer system 100. Exchange computer system 100 and computer device 114 may be connected via a T1 line, a common local area network (LAN) or other mechanism for connecting computer devices. Computer device 114 is shown connected to a radio 132. The user of radio 132 may be a trader or exchange employee. The radio user may transmit orders or other information to a user of computer device 114. The user of computer device 114 may then transmit the trade or other information to exchange computer system 100.
Computer devices 116 and 118 are coupled to a LAN 124. LAN 124 may have one or more of the well-known LAN topologies and may use a variety of different protocols, such as Ethernet. Computers 116 and 118 may communicate with each other and other computers and devices connected to LAN 124. Computers and other devices may be connected to LAN 124 via twisted pair wires, coaxial cable, fiber optics or other media. Alternatively, a wireless personal digital assistant device (PDA) 122 may communicate with LAN 124 or the Internet 126 via radio waves. PDA 122 may also communicate with exchange computer system 100 via a conventional wireless hub 128. As used herein, a PDA includes mobile telephones and other wireless devices that communicate with a network via radio waves.
One or more market makers 130 may maintain a market by providing constant bid and offer prices for a derivative or security to exchange computer system 100. Exchange computer system 100 may also exchange information with other trade engines, such as trade engine 138. One skilled in the art will appreciate that numerous additional computers and systems may be coupled to exchange computer system 100. Such computers and systems may include clearing, regulatory and fee systems.
The operations of computer devices and systems shown in
Of course, numerous additional servers, computers, handheld devices, personal digital assistants, telephones and other devices may also be connected to exchange computer system 100. Moreover, one skilled in the art will appreciate that the topology shown in
In step 210, at least a portion of the data from the trading record received in step 205 is stored on a computer readable medium. As used herein, a computer-readable medium may include, for example, RAM, flash memory, a hard disk drive, and/or an optical disk (CD ROM, DVD or other optical media). Solid-state memory modules allow for rapid queries due to the lack of moving parts, such as those associated with hard disk drives. In one embodiment, the medium is within the Exchange Computer System 100 and may be associated with one or more modules, such as the match engine module 136.
In step 220, the second trading record (or a portion thereof) is sequentially stored in relation to the first trading record 305, wherein the data from the second trading record is stored substantially physically adjacent to the data received from the first trading record 305. One of the advantages of locating records close to one another is faster reading times. Existing database systems may have data scattered throughout a memory device. Reading data arranged in this manner is time consuming because the reading process has to skip from one physical location to another physical location. For example, a hard disk drive must physically move a reading head from location to location.
While the exemplary embodiment does not have an index as used in conventional databases, the collection of data as represented by first trading records 305, 310 and 315 can perform row selection at the same speed as an indexed database. Indeed, in some embodiments, eliminating conventional databases indexes results in queries that have the same speed for all columns in the table. Conventional database typically have an index on every column. This results in deleterious effects on the insertion speed. Conversely, a collection of data organized according to the various embodiments of the present invention allows for rapid insertion speeds and is particularly useful and advantageous in real-time insertion situations, such as those routinely encountered in the trading industry. Moreover, by providing a collection of data without an associated database-type index, more space is available on the computer readable medium to store data, such as that present in the data of the first trading record and second trading record (305, 310). An increases in data storage may be achieved by eliminating the use of a conventional database-type index. In at least one implementation, the elimination of an index may double the amount of data that may be stored on the computer readable memory.
As one skilled in the art will appreciate, a computer-readable medium may comprise a plurality of linked drives and/or modules, wherein the data is stored in a sequential fashion as described above. In at least one embodiment, the plurality of drives and/or modules may be at different physical or remote locations. For example,
As briefly mentioned above, one or more trading records may comprise a plurality of fields. In one such embodiment, optional step 225 may be implemented to determine if one or more of the trading record comprise more than one field. In one embodiment, upon determining that more than one field exists, step 230 may optionally be initiated to parse a first field from one of the trading records. As shown in
In one embodiment, data associated with a first field of a trading record, such as financial instrument data 405 a may be stored in a first location on the computer readable medium (step 245). Data associated with other fields of the same record is not stored substantially sequential to financial instrument data 405 a, but may be placed on the computer readable memory at a different location (see exemplary embodiment of step 250). Upon receiving another trading record, such as trading record 410, it too may be parsed into a plurality of data associated with different fields. For simplicity,
As seen in
Embodiments of the invention also relate to methods of performing a query on a computer readable medium, such as computer readable mediums having data stored in accordance with several or all of the steps and embodiments discussed in regards to
For example, looking back to
Yet in other embodiments, only distinct portions of trading records may be queried. For example, in one embodiment the pending orders may be organized as set forth in
This can be more readily seen when reviewing
The speed at which queries may be performed when trading records are arranged as described above may be taken advantage of for other exchange and trading related activities. For example, a match engine may match trades using aspects of the inventions. In another example traders, trading firms and exchange regulatory or enforcement divisions may wish to recreate the state of a market, such as pending bids and offers, at a given time. One conventional approach includes recording a snapshot of the state of the market for every change in the market. These snapshots require large amounts of storage space, even for data parameters that may not have changed since the last snapshot.
In accordance with one embodiment of the invention, a trading firm, exchange or other entity may record trading records in a non-indexed collection of data, as described above. The speed at which such a collection may be queried and processed allows such entities to quickly recreate the state of the market for any time period. For example, an initial state of the market may first be determined and then all of the orders placed at an exchange may be processed in the same manner that they would be processed by an exchange until the desired point in time. For example, graphical user interface 500 (shown in
Graphical user interface 520 represents the state of a market at some time after the state represented with graphical user interface 500. If we assume that graphical user interface 500 represents the state that existed at 9:00 on Monday morning and graphical user interface 520 represents the state that existed at 2:00 on the following Wednesday, the state represented with graphical user interface 520 may be recreated by starting with the state representing with graphical user interface 500 and processing orders in the sequential non-indexed collection of orders received until 2:00 on Wednesday.
Unlike conventional indexed databases storing and retrieving trading data according to one or more methods of the present invention does not require large quantities of trading data to be executed in batch mode overnight. Indeed, under traditional approaches, large amounts of data could not be adequately analyzed in real-time, thereby preventing many uses of the data. Under select embodiments of the invention, the analysis of the data sequentially stored on the computer readable memory can be continually processed in real-time to monitor activity while new data is being written to the computer readable medium, all without having to create, update, and maintain a space-consuming database index and constant interruption to jump physical locations within the computer readable medium to locate a certain data piece.
Recreating market conditions may be readily accessible by querying methods, for example, as described above. Indeed, by following one or more embodiments of the invention, the analysis of the data sequentially stored on the computer readable memory can be continually processed in real-time to monitor activity while new data is being written to the computer-readable medium, all without having to create, update, and maintain a space-consuming database index and constant interruption to jump physical locations within the computer readable medium to locate a certain data piece.
A second processor 716 may be configured to periodically check shared memory 714 for new orders. Processor 716 may include a cache memory 718 that includes one or more order books, such as order book 720. The physical locations of pending orders stored in cache memory 718 may correspond to the order in which they were received at the match engine. When a new order is received, the order may be compared to orders that are included in order book 720. The use of two processors 712 and 716 allows for fast operation when matching of trades. One skilled in the art will appreciate that in an alternative embodiment the functions performed by processors 712 and 716 may be performed by a single processor. In other embodiments, the functions performed by processors 712 and 716 may be performed by three or more processors. An exemplary processor that may be used with aspects of the invention is an Intel Itanium II, which contains a 9 MB byte cache memory. Another exemplary processor is the AMD Opteron processor, which utilizes Hypertransport™ technology.
After trades are matched, market data may be generated by processor 716 and then transmitted to shared memory 714. Processor 712 may periodically check shared memory 714 for market data messages. When market data messages are received, processor 712 may format or expand the market data message and then distribute the market data message to trading entities, such as those represented by computer devices 702, 704 and 706. Processor 712 may also distribute other information to trading entities, such as acknowledgement messages.
The speed at which processor 716 may process orders and the low-cost of such processors facilitates the use of redundant components and backup mechanisms. For example a backup processor 722 may include a cache memory 724 that includes an order book 726. Order book 726 may be synchronized with order book 720, such that in the event that processor 716 fails, backup processor 722 may resume matching of trades. A synchronization module 728 may be used to synchronize order books 720 and 726. In one embodiment, processor 716 and backup processor 722 transmit information identifying the states of order books 720 and 726 to synchronization module 728. Synchronization module 728 may then compare the states of the order books and make any adjustments that are necessary.
In one alternative embodiment, match engine 708 may include several different processors that are configured to match a variety of different trades. Shared memory 714 may group new orders such that each of the processors knows which order to process. For example, a first processor may maintain a first order book and match trades against the first order book and a second processor may maintain a second order book for a different financial instrument and match trades for those financial instruments.
Backup processor 722 may be included within match engine 708. Alternatively, backup processor 722 may be connected to match engine 708 via a local area network or wide-area network. Backup processor 722 may be in a different geographic location than processor 716. For example processor 716 may be located within a first premises and backup processor 722 may be located in a different premises to prevent all processors from failing because of a fire or other event at the first premises. Two or more processors may also be geographically distributed and may be configured to process orders originating from different geographic regions. For example, processor 716 may be located in Chicago and may process orders originating in the United States and a second processor may be located in London in may be configured to process orders originating in that region.
In step 812 the new order is compared to the plurality of pending orders arranged in the non-indexed collection of substantially sequential orders. It is then determined whether the new order matches one of the pending orders in step 814. When the new order does not match one of the pending orders, in step 816 the new order is added to the non-indexed collection of substantially sequential orders. Step 816 may include adding the new order to an existing order book. If the new order does match one of the pending orders, a trade for the matching orders may be executed in step 818. Finally, in step 820 market data may be formatted and distributed.
The match system shown in
In the embodiment shown, match engines 902 and 914 receive orders from input gateway 926 and process those orders in parallel. Match engines 902 and 914 include the same data in order books 912 and 924 and are configured to produce the same results. Output messages from processors 904 and 916 are transmitted to an output gateway 932. Output gateway 932 may be implemented with a computer device configured to route data. In one embodiment of the invention input gateway 926 and output gateway 932 are implemented with the same computer device. Output gateway 932 may include a message memory 934 that stores information identifying messages that have been transmitted by output gateway 932. At least part of the identifying information may be assigned by input gateway 926. The identifying information may be used to make sure that data for an event is only distributed once. For example, input gateway 926 may assign and identification number to a new order and then transmit the order to match engines 902 and 914. Match engines 902 and 914 may process the order in parallel and generate market data messages. The market data messages include the identification number assigned by input gateway 926. When market data messages are received at output gateway 932, output gateway may be configured to search identification numbers stored in message memory 934 to determine if a corresponding market data messages has been transmitted. When a message is received that includes an identification number not stored in message memory 934, the message is distributed to trading entities, such as to computer device 936 via network 938. When a message is received that includes an identification number that is stored in message memory 934, the message may be discarded.
One skilled in the art will appreciate that a variety of different schemes may be used to assign identifying information to data received at input gateway 926 and then filter data at output gateway 932. In one embodiment of the invention a new order may be assigned an identification number at input gateway 926 and derivative identification numbers may be assigned to messages associated with that order. For example, a new order may be assigned a 15 digit identification number and an acknowledgment message may be assigned an identification number that consists of the 15 digit identification number followed by a character or number that identifies the type of message. This allows multiple messages associated with a single order to all be assigned unique identification numbers that are related and ensures that match engines 902 and 914 assign the same identification numbers to created messages.
In the event of a failure of one of match engines 902 or 914, output gateway 932 would receive messages from only one match engine and continue to process messages without any failure being apparent to trading entities. In embodiments that include more than two match engines operating in parallel, the failure of one or two match engines would not be apparent to trading entities. Match engines 902 and 914, as well as any additional match engines, may be located in the same location or may be distributed to prevent a fire, network failure or other catastrophic event from halting the operation of all match engines.
The speed at which incoming orders may be processed when pending orders are arranged in a substantially sequential non-indexed collection in the cache memory of a processor allows for the processing of many types of orders. Existing match engine systems limit the type of orders that traders may make. For example, it not practical to process an order that does not have a standard format, such as an order to buy four particular contracts and sell six other contracts, because of the time required to match a single novel order having several legs to several different orders.
With the system described above the processing of orders can be quickly performed, which allows for many types of orders. For example, a trader may enter a nonstandard order that contains multiple legs in different markets. The systems shown in
The present invention has been described herein with reference to specific exemplary embodiments thereof. It will be apparent to those skilled in the art that a person understanding this invention may conceive of changes or other embodiments or variations, which utilize the principles of this invention without departing from the broader spirit and scope of the invention as set forth in the appended claims. For example, aspects of the invention may be applied to data collections that are not related to exchanges or trading. All are considered within the sphere, spirit, and scope of the invention.