 |
Description  |
|
|
FIELD OF INVENTION
The present invention is directed to an automated crossing network (also
known as a matching system) for trading instruments, and in particular, a
continuous crossing network that matches buy and sell orders based upon a
satisfaction and size profile.
COPYRIGHT NOTICE
A portion of the disclosure of this patent document contains material which
is subject to copyright protection. The copyright owner has no objection
to the facsimile reproduction by anyone of the patent document or patent
disclosure as it appears in the Patent and Trademark Office, patent file
or records, but otherwise reserves all copyright rights whatsoever.
BACKGROUND OF THE INVENTION
Traditionally, traders and investors who desired to buy or sell securities
placed orders with brokers who traded on the floor of organized stock
exchanges, such as the New York Stock Exchange or the NASDAQ market.
Traders and investors, particularly institutional investors, are
increasingly balking at the high cost of trading on organized exchanges
and in the OTC (Over-The-Counter) market. Discontent with the expense of
using intermediaries and the cost of market impact has contributed to the
development of the electronic fourth market for crossing trades. See
"Reshaping the Equity Markets, A Guide for the 1990s" by Robert A.
Schwartz, Harper Business, 1991, especially at pp. 93-95.
Various companies and exchanges operate computerized crossing networks,
also called anonymous matching systems. By way of example, crossing
networks used in connection with the trading of trading instruments are
disclosed in U.S. Pat. No. 4,412,287, which discloses an automated stock
exchange in which a computer matches buy and sell orders for a variety of
stocks; U.S. Pat. No. 3,573,747, which discloses an anonymous trading
system for selling fungible properties between subscribers to the system;
U.S. Pat. No. 3,581,072, which discloses the use of a special purpose
digital computer for matching orders and establishing market prices in an
auction market for fungible goods; U.S. Pat. No. 4,674,044, which
discloses an automated securities trading system; U.S. Pat. No. 5,136,501,
which discloses an anonymous matching system for effectuating trades
through automatic matching in which buyers and sellers who are willing to
trade with one another based on specified criteria, such as price,
quantity and credit, may automatically trade when matching events occur
satisfying these criteria; and U.S. Pat. No. 5,101,353, which discloses an
automated system for providing liquidity to securities markets in which
orders are entered by the system and executed in real time either
internally between system users or externally with stock exchanges and
markets.
Crossing networks have a number of advantages, including: (a) traders need
not search for a contraparty; and (b) anonymity is preserved.
Existing facilities for crossing trades include Instinet's Crossing Network
and POSIT (Portfolio System for Institutional Trading) which is jointly
owned by Jefferies and BARRA. The Instinet Crossing Network has an
equities trading service to match buyers and sellers anonymously at set
times. Computers pair buyers with sellers on a time priority basis. Trades
are executed at the closing price for exchange-listed issues, and at the
midpoint of the inside market (best bid and ask) for OTC issues.
POSIT, for example, enables large investors to trade baskets of stocks
among themselves. The orders are sent to a central computer where they are
electronically matched with other orders. Unlike Instinet's Crossing
Network, POSIT crosses are done during the trading day. The prices are
obtained from those quoted on the exchanges, a practice known as
"parasitic pricing." See "Reshaping the Equity Markets, A Guide for the
1990s" cited above.
Instinet, owned by Reuters, also operates an electronic block-trading
system that facilitates the negotiation of block trades between
institutional investors and brokers. Instinet allows parties to trade
anonymously, entering bids electronically. Instinet subscribers can
respond to an "order" entered into the system either by matching a
displayed price or by making a counter bid or offer that is transmitted
instantaneously to the contraparty's terminal. The trades that result from
these negotiations become public information only when they are executed.
This procedure provides an alternative to the direct human-to-human
negotiation of orders in the upstairs market or on the trading floors.
Instinet provides a limit order book for over-the-counter (OTC) securities
and listed securities and also provides inside quotes for exchange listed
securities for the seven U.S. exchanges on which stocks can be traded and
for NASDAQ listed securities.
Many crossing networks function independently of existing stock exchanges.
However, some crossing networks are operated by stock exchanges.
For example, the Match Market Exchange ("MMX") is operated by the Chicago
Stock Exchange. All matched orders are executed at a random time within a
predetermined ten minute window at the market price at such time. The
market price is calculated based upon the spread of a particular issue.
Rather than matching orders on the basis of time priority, thee system
uses liquidity fees and liquidity credits to determine the level of
priority for order matching. Those users willing to pay the highest
liquidity fee have the highest execution priority. See 59 F.R. 5451 (Feb.
4, 1994).
Crossing networks that automatically match buy and sell orders often
concentrate trading at a single point of time, and can be called a batch
process matching system. There is a need, however, for an anonymous
crossing network that continuously, and in real-time, satisfies the buying
and selling desires of an arbitrary number of market participants.
A major problem encountered in the design of crossing networks is that of
determining how to match buyers and sellers. Existing approaches to this
problem include:
Take-out strategies, where overlapping bids and offers are matched at the
midpoint of the overlapped bid and ask prices, with priority given to
buyers and sellers in order of price. This assumes a significant quantity
of non-disclosed orders in the system; otherwise, there would be no
incentive for overlap, and take-out would start at the disclosed best
bid/offer prices, just like the Instinet book.
Single price auction strategies, where a single, size-weighted average
price is computed from overlapping bid and offer prices, and everyone is
filled at that price. Again, traders would have to be confident of a
significant number of non-disclosed orders in the system to have the
incentive to enter orders at a better price than the best disclosed price.
Premium strategies (as in the Chicago MMX system), where bids and offers
have an associated positive or negative premium, and crossing takes place
at the midpoint of market spread or at the minimum necessary premium
differential from the midpoint, with priority given in order of premium.
Here, the premium-based priority in matching provides the incentive for
offering higher premiums.
Each of the above approaches is a batch process that relies upon ad hoc
rules of competition among a relatively small set of discrete orders as
being the means of arbitrating the crossing network participants' buy/sell
entries. In the real world of trading, orders to buy or sell can enter the
market at any time, and discrete orders in a crossing network often
represent only an approximate and partial expression of the order fill
that would satisfy the trader. For institutional traders in particular, an
individual order seldom represents the full desired fill size, and the
trader must then employ multiple orders at different prices (and generally
in different markets) to achieve his ultimate fill.
Typically, existing crossing networks allow discrete buy or sell orders to
be entered, e.g., "sell 10,000 IBM at 64." However, as stated above many
traders, particularly institutional traders, wish to deal in baskets of
securities, so that, for example, a portfolio is as far as possible,
"balanced." Existing crossing networks do not easily allow traders to
enter combinations of orders, such as "sell 10,000 IBM at 64 only if I can
buy 20,000 DEC at 32". Furthermore, existing crossing networks do not
allow traders to enter combinations of orders, such as "sell 10,000 IBM at
64 or sell 100,000 IBM at 63." Traders often have trading strategies such
as, for example, "buy 3,000 IBM at 33, but if I can buy 5,000, I would be
prepared to pay 33 and 1/2", that cannot be handled by existing crossing
networks.
Given the above limitations of conventional crossing networks, a more
satisfactory approach to the overall problem of continuously satisfying
the buying and selling desires of an arbitrary number of market
participants is needed. Furthermore, a crossing network is needed that
will guarantee mathematical optimality of the matching process, so that
each participant is assured that the overall outcome of the process (in
terms of the price and size of all fills) has maximized the joint
satisfaction of all participants.
SUMMARY OF THE INVENTION
The present invention is directed to a computerized crossing network that
allows traders to input as orders a satisfaction density profile and
maximum size limit which at once characterizes the trader's degree of
satisfaction to trade at any and all prices and sizes, up to the aggregate
(or size) limit, and that matches orders (as represented by each trader's
satisfaction density profile) so that each trader is assured that the
overall outcome of the process (in terms of average price and size of
fill) has maximized the mutual satisfaction of all traders.
The satisfaction density profile is a two-dimensional grid or matrix (which
could also be represented as a two-dimensional graph or in another
two-dimensional format), one dimension being price and the second
dimension being size of transaction, that as a whole characterizes the
trader's degree of satisfaction for a transaction at each (price, size)
coordinate. Each element of the satisfaction density profile, called a
satisfaction density value, indicates the trader's degree of satisfaction
to trade that size order at that price. In the representative embodiment,
each satisfaction density value is a number between zero and one, with
zero representing no satisfaction (i.e., will under no circumstances trade
that quantity at that price) and one representing total satisfaction.
Each trader can input one or more satisfaction density profiles. In the
representative embodiment, each satisfaction density profile can be
represented and input as a graph of at least two dimensions, such as, for
example, as a temperature grid or two dimensional matrix.
Once the satisfaction density profile is complete, the trader causes the
satisfaction density profile to be transmitted to a central matching
controller ("CMC"), which anonymously matches buy and sell orders as
discussed below.
For the purpose of explanation, assume a batch process in which multiple
traders enter satisfaction density profiles that represent either buy or
sell orders for a particular stock. Upon transmission of the satisfaction
density profiles to the CMC, the CMC will cause buy profiles to be stored
in a buy profile database and sell profiles to be stored in a sell profile
database. The CMC will then calculate, for every buy/sell profile pair, a
mutual satisfaction cross product. The mutual satisfaction cross product
represents the degree to which that buy/sell pair can satisfy each other.
Next, the individual grid values of the mutual satisfaction cross products
for all buy/sell combinations are ranked in order, starting with the
highest value of mutual satisfaction. The buy/sell orders represented by
the ranked grid values of the mutual satisfaction cross products are then
matched in order, and matching trades are aggregated by the CMC system.
The matching process then continues down the ranked list.
The present invention can easily handle trades of baskets of securities.
For example, in the representative embodiment, if a trader wishes to
simultaneously both sell IBM and buy DEC, the trader would create a sell
satisfaction density profile representing the IBM part of the transaction
and a buy satisfaction density profile representing the DEC part of the
transaction, and indicate that these profiles are to be linked together.
The CMC will then combine these two satisfaction density profiles into
one. This combining process can be accomplished, for example, by utilizing
a connection matrix and performing matrix manipulation on the two
profiles. It will be apparent that other, more complex, combined trades
can be accomplished in the same manner.
The present invention can be operated as a batch crossing network, where
orders are matched at set times. Alternatively, the present invention can
be operated as a continuous crossing network by treating each new
satisfaction density profile that is entered as a triggering event that
causes a new round of computation as described above.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram of a representative system operating according to
the present invention.
FIG. 2 is an example satisfaction density profile for a sell order.
FIGS. 3A to 3D are example satisfaction density profiles in contour plot
format.
FIG. 4 is an example interface window, used to set parameters for and
enable input of satisfaction density profiles.
FIG. 5 is a load pattern interface window, used to load parameterized
satisfaction density profile templates.
FIGS. 6A and 6B show an example screen display showing an interface window,
a load pattern interface window and a satisfaction density profile.
FIGS. 7A and 7B show an example buy satisfaction density profile displayed
as a temperature graph.
FIG. 8 is a description of the matching process in flow chart format.
FIGS. 9A to 9C are example mutual satisfaction cross products, represented
as contour graphs.
DETAILED DESCRIPTION
The present invention is described below in the context of trading equity
securities. However, the invention is not so limited and can be easily
adapted to allow the trading of other liquid assets such as futures,
derivatives, options, bonds, currencies and the like. Accordingly, where
the context permits, the terms "securities", "stock", and "shares" when
used herein includes other instruments that can be traded, such as, for
example, futures, derivatives, options, bonds and currencies. The terms
"buy" and "sell" include, where appropriate, put and call, bid and offer,
etc.
Intended users of the representative embodiment system of this invention
are typically investors, such as institutional investors (e.g., a pension
fund) but may also be individual investors, brokers or others who deal in
or trade securities. As used herein, the term "user", "trader" or
"investor" means that person or entity who wishes to make a trade.
Referring now to the drawings, and initially FIG. 1, there is illustrated a
block diagram of the overall architecture of the crossing network
according to the present invention. A central matching controller ("CMC")
2 matches buy and sell orders transmitted to the CMC from various trader
terminals, e.g., 10, 12, 14. The CMC 2 is preferably a supercomputer that
can perform matrix calculations at rates of multiple gigaflops, such as,
for example with present technology, an IBM SP2 or an Intel PARAGON
supercomputer. A storage device 4 is coupled to the CMC 2. The storage
device 4 comprises a database for storing buy and sell satisfaction
density profiles. In the representative embodiment, buy satisfaction
density profiles can be stored in a buy profile database 6 and sell
satisfaction density profiles can be stored in a sell profile database 8.
Each trader who wished to trade will have access to a trader terminal, such
as trader CPU 10, 12, 14. Ideally, the trader terminals 10, 12, 14 are
high powered personal computers or workstations. The trader terminals 10,
12, 14 are coupled to the CMC 2. Trader terminals may be coupled to the
CMC 2 over a wide area network (WAN) and/or over the Internet. Each trader
terminal includes one or more input/output devices 16, 18, 20 that allow
for the entry of satisfaction density profiles and the display of output,
such as matching trades.
Turning now to FIG. 2, there is illustrated an example satisfaction density
profile for a sell order. In the representative embodiment, each security
that a trader wishes to trade requires the creation of a satisfaction
density profile. The satisfaction density profile as shown in FIG. 2 is a
two-dimensional grid. Quantity is represented on the vertical axis and
price is represented on the horizontal axis. (The ranges and scale of each
axis can be set by the trader or automatically set by the present
invention.) The satisfaction density levels are described by the numbers
shown.
The satisfaction density profile characterizes the trader's range of
interest in the transaction at each (price, size) combination. The trader
is required to enter in one or more places on the satisfaction density
profile a satisfaction density value, indicating the trader's willingness
to trade that size order at that price. In the representative embodiment
shown in FIG. 2, each satisfaction density value is a number between and
including zero and one, with zero representing no satisfaction (i.e., will
under no circumstances trade that quantity at that price) and one
representing total satisfaction.
Thus, for example, FIG. 2 shows that the trader is fully willing to sell
7,500 at $73, but under no circumstances is willing to sell any quantity
at $70.25. The satisfaction density value to sell 2,250 at $72.25 is 0.3,
indicating a limited degree of satisfaction. FIG. 2 shows that the trader
is more willing to sell 8,250 at $72.75 then 1,500 at $73.25.
In FIG. 2, a boundary is shown, marked B1-B2-B3-B4. Using known drawing
program techniques, the user is able to draw a boundary on the
satisfaction density profile. On one side of the boundary, the user can
enter satisfaction density values, On the other side of the boundary, the
satisfaction density values are automatically set to zero.
In FIG. 3A, there is illustrated an example satisfaction density profile
for a buy order. The satisfaction density profile as shown in FIG. 3A is a
contour plot on a two-dimensional grid. Price is represented on the
vertical axis and quantity is represented on the horizontal axis. The
satisfaction density levels are described by the contours shown. In the
representative embodiment of FIG. 3A, each satisfaction density value is a
number between and including zero and one, with zero representing no
satisfaction (i.e., will under no circumstances trade that quantity at
that price) and one representing total satisfaction.
Thus, for example, FIG. 3A shows that the trader would be highly satisfied
to buy 1,500 at $20.50, but under no circumstances is willing to buy any
quantity at $24. The satisfaction density value for a buy of 3,000 at $20
is 0.6, indicating a more limited degree of satisfaction. FIG. 3A shows
that the trader is more willing to buy 2,000 at $21 then 4,000 at $21.
FIG. 3B shows an example satisfaction density profile for a sell
transaction in the same format as the contour plot in FIG. 3A. The
satisfaction density profile of FIG. 3B could represent, for example, an
agency sell order.
As another example, FIG. 3C shows a satisfaction density profile for a sell
transaction. FIG. 3C illustrates a "soft" limit order.
As another example, FIG. 3D shows a satisfaction density profile for a buy
transaction.
Turning now to FIG. 4, there is illustrated an example interface window 21
that can be used by a trader to set parameters when inputting a
satisfaction density profile. In the representative embodiment, the
interface window 21 is displayed on an Input/Output device (e.g., 16, 18,
20). The interface window 21 in the representative embodiment can be
manipulated according to standard graphical user interface (GUI) commands
by a user using, for example, a mouse and a keyboard.
A Side indicator 22 allows the trader to set the type of transaction, e.g.,
buy, sell, sell short etc. A Symbol indicator 24 allows the trader to set
the instrument being traded, e.g., for stocks, the ticker symbol; for
futures, the futures symbol, etc. A Maximum Quantity indicator 26 allows
the trader to set the maximum quantity to be traded. In FIG. 4, the trader
has entered a buy profile to buy a maximum quantity of 50,000 Motorola
shares.
The trader can also set a Time-in-Force indicator 28 that determines for
how long the order will be valid. Examples of valid Time-in-Force settings
include "good until canceled", "good until end of day" and for a set
period of time.
A Surface Description section 30 of the interface window 21 allows the
trader to define the dimensions of the satisfaction density profile. A
Center Price parameter 32 is used to set the price that is displayed in
the center of price axis of the satisfaction density profile. A Price
Range parameter 34 shows the range for the price on the price axis. A
Price Interval parameter 36 enables the trader to set the price interval
(i.e., scale) on the price axis of the satisfaction density profile. For
example, as shown in FIG. 4, the center price is set at 48 and the price
range at 6. Thus, the price axis will vary from a price of 45 to a price
of 51 (with 48 being in the center) in price increments of one quarter. A
Price parameter 37 can be used to set whether the price is absolute or
relative (e.g., pegged to the bid/ask midpoint or to the average price of
the last match of that security).
A Min Size parameter 38 and a Max Size parameter 40 allows the trader to
set the minimum and maximum limits for the quantity axis of the
satisfaction density profile.
Mismatches of grid resolution (e.g., due to different traders using
different settings) is handled by interpolation to a common grid
resolution. For example, the CMC 2 could determine the smallest grid
resolution used and interpolate all coarser grid resolutions to that
resolution. Alternatively, the CMC 2 could specify a minimum resolution
(e.g. 100 shares and 1/8 in price) and interpolate all coarser grid
resolutions to the specified minimum resolution.
A Cursor Position segment 42 will display the coordinates (price, size) of
the position of the cursor as the cursor is moved by the user over the
satisfaction density profile being created. A Cursor Type section 44 has a
number of buttons to control the function of the cursor, for example, a
set button, a more button, a less button, a boundary button, and an edit
button. The more and less buttons allow the user to increase or decrease
the satisfaction density level on the satisfaction density profile at the
location of the cursor. In the representative embodiment, the selected
satisfaction density value is increased or deceased each time the mouse
button is "clicked." The boundary button allows the user to create a
boundary, as discussed above at FIG. 2. The edit button allows the user to
edit the boundary. The set button allows the user to set a satisfaction
density value, so that each time the mouse is "clicked", the set value is
placed in the satisfaction density profile at the location of the cursor.
In the representative embodiment, there is a Display section 46 that has
two buttons, namely, an Update button and a Display Surface button. The
Update Button will redraw the satisfaction density profile. The Display
Surface button will allow the user to change the way the satisfaction
density profile is displayed, e.g., as a contour plot (as shown in FIG. 2)
or a temperature graph (as shown in FIGS. 7A and 7B).
The interface window 21 can also have a quit button 50, a clear button 52,
a reset button 54 and an apply button 56. The quit button 50 enables a
user to exit from the program that controls the interface window 21. The
clear button 52 clears the display of the graph (e.g., the display of the
satisfaction density profile) and all other parameter values, such as, for
example, the Min Size parameter 38 and the Max Size parameter 40. The
reset button 54 "undeletes" the most recent change to the graph, resetting
the state of the graph to the immediately preceding state. The apply
button 56, when selected, causes the satisfaction density profile to be
submitted for matching. A remove boundary button 59, when selected, will
remove all boundaries from the satisfaction density profile.
A patterns button 58 allows a user to retrieve predefined or user-defined
parameterized templates for common types of satisfaction density profiles.
Upon selection of the patterns button 58, a load pattern window 60 (as
shown in FIG. 5) will be displayed, that allows the user to select saved
patterns. In the example of FIG. 5, there are two saved patterns, namely
ibm.sub.-- seller.sub.-- 1 and ibm.sub.-- buyer.sub.-- 1, that store
patterns that enable a user to quickly load and use satisfaction density
profiles to buy and sell IBM stock.
Certain common order types (e.g., market-maker quotes, limit orders, agency
orders) can be expressed as parameterized templates that could be selected
directly using the Patterns button 58 and the Load Pattern window 60. Once
the corresponding template pops up on the display, the user could morph
the display to suit the user's preferences as described below.
An even more elegant approach is to use a family of templates that
represent two-dimensional basis functions for the surface profile. These
functions can be selected from any of the standard two-dimensional
orthogonal basis sets (e.g., Fourier, Cosine, Walsh-Hadamard, Haar
transforms) or from more generalized and/or data specific basis functions
such as those associated with the Gabor and other wavelet transforms or
the Karhunen-Loeve transform. Analogous to their role in image data
compression, the use of such functions as templates in this application
would provide a substantially lower dimensional representation of the
buy/sell surface profile, which in turn will correspondingly reduce the
dimensionality of the coupling matrices needed to link together the
profiles of multiple securities.
FIGS. 6A and 6B show a full screen display that includes an interface
window 21 and the satisfaction density profile of FIG. 2.
As will be appreciated, the entry of buy/sell profiles can be facilitated
by a combination of a user-friendly graphical interface and
user-tailorable templates. Those illustrated above are merely examples of
the types of interfaces that could be used to enter satisfaction density
values. In an alternative embodiment, the GUI would provide a set of tools
for quickly entering buy/sell profile "peg point" values at selected
price/size grid coordinates, along with an arbitrarily drawn boundary
outside of which the profile values would be set to zero at all grid
points. The CMC 2 would then calculate a two-dimensional (or
multi-dimensional) profile that exactly matches the specified points and
interpolates between these values to compute all other grid values. This
interpolation can be accomplished by a number of mathematical algorithms,
including but not limited to triangular tessellations, spline functions,
and surface and/or contour plotting programs. The GUI would also include
the ability to "morph" a surface profile created in this manner, using
mouse hook-and-drag type operations or other similar methods, so that the
profile can be altered by the user as desired to achieve a final
configuration.
FIGS. 7A and 7B show a buy satisfaction density profile displayed as a
temperature graph. The user can cause any satisfaction density profile
that has been inputted as a contour plot (or other representation) to be
displayed in the form of a temperature graph by selecting the Display
Surface button. In the representative embodiment, each (price, quantity)
coordinate is represented by a colored square (e.g., 80), with the color
of the square signifying the satisfaction density value. For example, in
the representative embodiment, a satisfaction density value of 0 is
represented by a black square, a satisfaction density value of 1 is
represented by a white square, a satisfaction density value of 0.1 is
represented by a blue square, a satisfaction density value of 0.2 is
represented by a red square, a satisfaction density value of 0.9 is
represented by a yellow square, and so on. Using this display, the trader
can view the satisfaction density profile as a meaningful and colorful
graphical display.
As shown in FIGS. 2, 3, 6A, 6B, 7A and 7B, there are many ways and forms
for creating and representing a satisfaction density profile on a computer
interface and for inputting satisfaction density values.
FIG. 8 is a flow chart representing the overall steps performed in
accordance with the present invention. Traders enter satisfaction density
profiles, as discussed above, at steps 100 and 102. It will be appreciated
that many traders can each enter one or more satisfaction density
profiles. For convenience, FIG. 8 shows that the buy and sell profiles are
entered at different steps (i.e., steps 100 and 102); however, | | |
