U.S. patent application number 11/724313 was filed with the patent office on 2007-10-25 for securities settlement system.
This patent application is currently assigned to OMX Technology AB. Invention is credited to Bengt Lejdstrom, Oskar Sander, Johan Soderqvist.
Application Number | 20070250437 11/724313 |
Document ID | / |
Family ID | 38564012 |
Filed Date | 2007-10-25 |
United States Patent
Application |
20070250437 |
Kind Code |
A1 |
Lejdstrom; Bengt ; et
al. |
October 25, 2007 |
Securities settlement system
Abstract
A securities settlement system for clearing trades comprising an
input for receiving trade information, a selector for selecting a
group of trades to be cleared, an aggregation unit for determining
an aggregated obligation to be cleared by each user associated with
the group of trades and a settlement unit for executing the
aggregated obligations for each user to clear the trades in the
group of trades.
Inventors: |
Lejdstrom; Bengt;
(Sollentuna, SE) ; Sander; Oskar; (Stockholm,
SE) ; Soderqvist; Johan; (Nacka, SE) |
Correspondence
Address: |
NIXON & VANDERHYE, PC
901 NORTH GLEBE ROAD, 11TH FLOOR
ARLINGTON
VA
22203
US
|
Assignee: |
OMX Technology AB
Stockholm
SE
|
Family ID: |
38564012 |
Appl. No.: |
11/724313 |
Filed: |
March 15, 2007 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
60789574 |
Apr 6, 2006 |
|
|
|
Current U.S.
Class: |
705/37 |
Current CPC
Class: |
G06Q 40/04 20130101 |
Class at
Publication: |
705/037 |
International
Class: |
G06Q 40/00 20060101
G06Q040/00 |
Claims
1. A securities settlement system for clearing trades comprising an
input for receiving trade information, a selector for selecting a
group of trades to be cleared, an aggregation unit for determining
an aggregated obligation to be cleared by each user associated with
the group of trades and a settlement unit for executing the
aggregated obligations for each user to clear the trades in the
group of trades.
2. Securities settlement system according to claim 1, further
comprising a register indicating each user's obligation limit and a
comparator for comparing each user's aggregated obligation with the
obligation limit and if an obligation fails for a user, the
selector modifies the selected group of trades to arrive at a new
selected group where all obligations can be met by all users.
3. Securities settlement system according to claim 2, wherein the
selector modifies the selected group of trades by removing at least
one of the trades from the selected group of trades, the at least
one removed trade being associated with the user failing to meet
the obligation.
4. Securities settlement system according to claim 2, wherein the
selector modifies the selected group of trades by iterating through
a set of algorithms to determine a minimum number of trades
required to be removed from the selected group of trades in order
to meet the aggregated obligation for all users.
5. Method of clearing trades in a securities settlement system,
comprising the steps of: receiving information relating to a number
of trades; selecting a group of trades from said number of trades;
determining an aggregated obligation to be cleared for each user
associated with the group of trades; and executing the aggregated
obligation for each user, thereby clearing the group of trades.
6. Method of clearing trades according to claim 5 comprising the
further steps of: comparing the aggregated obligation for each user
with an obligation limit associated with each user; and modifying
the selected group of trades if at least one aggregated obligation
cannot be met.
7. Method of clearing trades according to claim 6, wherein the step
of modifying the selected group of trades includes removing at
least one trade associated with the aggregated obligation that
cannot be met.
8. Method of clearing trades according to claim 7, wherein the step
of modifying the selected group of trades includes iteratively
making random assignments of a new group of trades and saving an
optimal selection.
9. Method of clearing trades according to claim 8, wherein the step
of iteratively making random assignments of a new group of trades
includes successively increasing the new selected group of
trades.
10. A securities settlement system for clearing trades comprising
an input for receiving trade information, a sorting unit for
sorting out trades associated with a specific sorting criteria and
an aggregation unit for determining an aggregated obligation to be
cleared for each user.
11. Securities settlement system according to claim 10, wherein the
specific sorting criteria is one or more criterias selected from
the group of: user, instrument type and market.
12. Securities settlement system according to claim 10, further
comprising a register indicating a user's obligation limit and a
comparator for comparing the user's aggregated obligation with the
obligation limit and if an obligation fails for the user, at least
one of the sorted out trades is removed from the trades associated
with the user failing to meet the aggregated obligation.
13. A securities settlement system for clearing trades comprising
an means for receiving trade information, means for selecting a
group of trades to be cleared, means for determining an aggregated
obligation to be cleared by each user associated with the group of
trades and means for executing the aggregated obligations for each
user to clear the trades in the group of trades.
14. Securities settlement system according to claim 13, further
comprising means for determining each user's obligation limit and
means for comparing each user's aggregated obligation with the
obligation limit and if an obligation fails for a user, the means
for selecting a group of trades modifies the selected group of
trades to arrive at a new selected group where all obligations can
be met by all users.
15. Securities settlement system according to claim 14, wherein the
means for selecting a group of trades modifies the selected group
of trades by removing at least one of the trades from the selected
group of trades, the at least one removed trade being associated
with the user failing to meet the obligation.
16. Securities settlement system according to claim 14, wherein the
means for selecting a group of trades modifies the selected group
of trades by iterating through a set of algorithms to determine a
minimum number of trades required to be removed from the selected
group of trades in order to meet the aggregated obligation for all
users.
Description
CROSS-REFERENCE
[0001] This application is a new U.S. utility application claiming
priority to U.S. Provisional Application No. 60/789,574 filed Apr.
6, 2006, the entire content of which is hereby incorporated by
reference in this application.
BACKGROUND ART
[0002] Globalization has directly impacted various aspects of the
Central Securities Depository (CSD) market sector. A drive for
consistency and a more standardized approach in operations and
systems has resulted. In addition, there is continuous pressure on
participants and operators to improve efficiency in all elements of
the transaction value chain. CSDs are exposed to more complicated
securities and resultant activities, e.g. Corporate Actions,
because of the global perspective and market refinement. Processes
and systems must now be much more adaptable. In such a dynamic
environment "speed to market" of new products and services is
critical. The CSD systems must facilitate this "speed", which can
be achieved by flexible, generic functionality for the different
steps in the settlement or other processes.
[0003] Transactions between market participants require matching
and settlement. Trades can be captured directly from a trading
(exchange) system or other external system, such as a Central
Counterparty (CCP), matching service or back office system or
manually entered by participants. Trade matching and confirmation
prepare transactions for settlement according to applicable
settlement rules. Assets and financing are verified, and securities
may be locked-in in the investor CSD prior to settlement.
DESCRIPTION OF THE INVENTION
[0004] An overall object and purpose of the securities settlement
system (SSS) is to increase the flow of successful trades, by
minimizing the demands for liquidity of cash especially.
[0005] A securities settlement system that achieves the object and
purpose is obtained through a securities settlement system
comprising an input for receiving trade information, a selector for
selecting a group of trades to be cleared, an aggregation unit for
determining an aggregated obligation to be cleared by each user
associated with the group of trades and a settlement unit for
executing the aggregated obligations for each user to clear the
trades in the group of trades.
[0006] Grouping trades and aggregating the obligation to be cleared
for each user increases usability and speed of the clearing. Each
user's account is accessed a minimum of times (essentially once)
and a minimum of transfers of securities is also established.
Aggregating also provides the benefit of enabling clearing of a
series of trades where a traditional one by one clearing would have
resulted in trade removals or freezing the clearing. This because
in a group of trades a single user may have several trades that
together is clearable as they net out.
[0007] An advantageous improvement is achieved in that the system
further comprises a register indicating each user's obligation
limit and a comparator for comparing each user's aggregated
obligation with the obligation limit and if an obligation fails for
a user, the selector modifies the selected group of trades to
arrive at a new selected group of trades where all obligations can
be met by all users.
[0008] Hereby it is guaranteed that all trades in the selected
group of trades can be cleared. The re-selection process can be
performed in many ways, for instance by iterating through a set of
algorithms in order to determine a minimum of trades to be removed
from the group in order to obtain a group in which all trades are
clearable.
[0009] An alternative solution is obtained through the invention by
a securities settlement system comprising an input for receiving
trade information, a sorting unit for sorting out trades associated
with a specific sorting criteria and an aggregation unit for
determining an aggregated obligation to be cleared by each
user.
[0010] The sorting criteria may be one or more specific users
(prioritized users due to number of trades or amount of trades),
one or more specific instrument types and one or more markets. A
combination of these criteria or other trade specific critera are
also feasible.
[0011] A method that achieves an effective and improved settlement
clearing is obtained through the invention by the method steps of
receiving information relating to a number of trades; selecting a
group of trades from said number of trades; determining an
aggregated obligation to be cleared for each user associated with
the group of trades; and executing the aggregated obligation for
each user, thereby clearing the group of trades.
[0012] The essence of the new type of clearing is to combine
multiple trades into one group and to net the effects of multiple
debits and credits against the same account or cash record.
[0013] For example, consider two trades; one which debits the cash
record A with $10 and the other that credits A with $10. By
combining the two, the net effect is to do nothing against A.
Therefore, even if A has no balance, the debiting trade can be
completed.
[0014] Therefore, this system affects the areas around clearing and
settlement. In summary, the following sections of the SSS are
affected: [0015] Clearing of obligations. [0016] Handling of queue.
[0017] Settlement of instructions.
BRIEF DESCRIPTION OF THE DRAWINGS
[0018] FIG. 1 illustrates how aggregation of trades according to
the invention reduces the number of necessary transactions;
[0019] FIG. 2 illustrates an example of an overall clearing and
settlement process according to the invention;
[0020] FIG. 3 shows a flow chart exemplifying one possible
algorithm for optimizing a group of trades to be cleared;
[0021] FIG. 4 shows a graph exemplifying one possible selection for
optimizing a group of trades to be cleared;
[0022] FIG. 5 illustrates the generic clearing and settlement
process;
[0023] FIG. 6 illustrates the process at waypoints in the generic
clearing and settlement process in FIG. 5;
[0024] FIG. 7 further exemplifies the process at different
waypoints;
[0025] FIG. 8 examplifies an optimised process at waypoints;
[0026] FIG. 9 examplifies an alternative optimized process at
waypoints;
[0027] FIG. 10 shows an embodiment of a securities settlement
system in accordance with the invention; and
[0028] FIG. 11 shows an alternative embodiment of a securities
settlement system in accordance with the invention.
DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION
[0029] The basis for all clearing in the settlement system is the
settlement obligation, which contains one or several debit or
credit instructions. In the following, we assume each settlement
obligation contains exactly one such instruction.
[0030] A trade is converted into a set of obligations, called a
settlement obligation group. Hence, a group is a list of debit and
credit obligations that must be settled automatically and
simultaneously. Thereby, a list of entered trades give rise to a
list of settlement obligation groups.
[0031] Clearing a group thus means to reserve the necessary amounts
induced by the contained debit obligations, before these can be
settled together with the corresponding credit obligations.
[0032] For each such list of groups, the settlement system is
required to clear and settle as many as possible whenever it is
invoked. The normal mode is to invoke the system repeatedly and
settle the groups reactively in real-time. However, it may also be
that the system is invoked periodically, thus confronted with large
sets of groups that, of course, could be settled one by one, but
where there are several opportunities to do "better" than that.
[0033] For example, consider the scenario of FIG. 1. Assume the
participant T is a great trader. He buys a security S for $X from a
market-maker M, and sells it for $X+10 to investor I. But assumes
he has no money. Hence, he cannot buy S, nor complete the sell.
But, if the system aggregated the two trades, it could deliver S to
I from M, and $10 to T from I, and $X to M from T.
[0034] Such deadlock situations occur frequently in a market, and
require special attention from the settlement system to improve the
completion rate of trades (as this is the true purpose of any
market). In the general case, the deadlock may occur via a chain of
trades between multiple participants.
[0035] The requirement then is to make the settlement system
aggregate trades to minimize the demand for liquidity by resolving
deadlocks.
[0036] In the following we look into the details of the clearing
and settlement process, to better illustrate the changes made in
the settlement system and process.
Settlement Definitions
[0037] We use the following definitions from here on. Rather than
referring to netted we prefer the term aggregated, since netting is
a term somewhat misused. [0038] (Settlement) obligation, i.e. a
tuple of four values: (A, I, V, P), where A refers to an account or
cash record, I to a security or currency, V to a volume or amount
(negative or positive), and P to a priority (Active, Deferred).
[0039] (Settlement obligation) group, i.e. a list of obligations,
where it is assumed that for any two obligations (A, I, V, P) and
(B, J, W, Q) in a group, either A<>B or I<>J
(otherwise, they are aggregated into one). A two-party trade is
typically represented by a group of four obligations (one
debit/credit per participant and counterpart). [0040] Obligation
aggregation, i.e. two obligations (A, I, V, P) and (A, I, W, Q) are
aggregated by replacing them with (A, I, V+W, Pm), where Pm is the
minimum priority of P and Q. [0041] Group graph, i.e. an undirected
graph of groups constructed as follows. Two groups G1 and G2 are
connected if at least one obligation in G1 can be aggregated with
one obligation in G2 (i.e. they refer to the same
holding/position). [0042] Group component, i.e. given a graph of
groups, the component is simply a connected component in the graph.
Hence, if two groups G1 and G2 are in different components, they
have no obligations that can be aggregated. [0043] Holding, i.e. a
holding AI is defined by the volume of the instrument/currency I
held in account/cash record A. Clearing And Settlement Process
[0044] The overall clearing and settlement process is shown in FIG.
2 and can be described as follows. [0045] 1. First two lists of
settlement rules of purpose LockinIncremental and LockinAggregated
are retrieved.
[0046] This determines the selection criteria of which groups
should be cleared in the current subsession. The criteria consists
of a list of sources, instrument classes, currencies and types
(message/operation). [0047] 2. From the selection criteria, choose
a subset of all unsettled groups (or all, depending on the number
of inserted trades). [0048] 3. For those groups to be cleared
incrementally, the clearing proceeds by clearing the groups, one by
one, and within a group, clearing the obligations one by one.
Whenever an obligation is Deferred, or overdraws an account/cash
record, the clearing of the group is frozen (to be reconsidered in
later invokations of the system). When all debit obligations of a
group are cleared, the group is marked as fully cleared. [0049] 4.
For those groups to be cleared in aggregation, the clearing
proceeds by aggregating the groups (see below) into net
obligations. For each net obligation, neither Deferred nor
overdrawing, each group aggregated into it is marked as fully
cleared. [0050] 5. For each fully cleared group, the system
proceeds by adjusting each account and cash record against the
obligations in the group. Such groups are then Settled.
[0051] Incremental clearing is performed as follows. [0052] A. Let
(G1, . . . , Gn) be a list of groups. [0053] B. For each group Gi
in the list, let (o1, . . . , ok) be its list of obligations.
[0054] 1. For each obligation oi=(A, I, V, P) in the list, where
the available volume of AI is Av: [0055] 2. if P=Deferred and
V<0, return. [0056] 3. if P<>Deferred and V<0, check if
Av+V>=0. If not, return. Otherwise, make the reservation of V in
AI, and continue in 1. [0057] 4. Continue in 1.
[0058] Hence, incremental clearing considers one group at a time,
obligation by obligation, and reserves each debit in order. If, not
all debits are successfully locked in, because of a Deferred
obligation, or an overdraft, the group is left partially cleared.
It remains this way, until either the deferral is removed or the
affected account/cash record has more available volume.
[0059] Aggregated clearing is performed as follows. [0060] 1. Let
G=(G1, . . . , Gn) be a list of groups, where no group contains a
Deferred obligation (these are filtered out). [0061] 2. For each
group Gi in the list, let Oi=(o1, . . . , ok) be its list of
obligations. [0062] 3. Let O be the aggregated union of all Oi.
[0063] 4. For each obligation oi=(A, I, V, _) in O, where the
available volume of AI is Av: [0064] 5. if V<0, check if
Av+V>=0. If so, continue in 4. Otherwise, remove m groups H1, .
. . , Hm (let m be the smallest such number) from G, where Hi
contains an obligation (A, I, Wi, _), Wi<0, such that Av+V-(W1+.
. . +Wm)>=0, and continue in 3. [0065] 6. Continue at 4,
selecting the next obligation.
[0066] Hence, aggregated clearing considers all groups
simultaneously, and aggregates all the contained obligations.
[0067] The result is a list of netted obligations that all must be
cleared together. If this fails, because of one debit making an
overdraft against an account (cash record), a number of groups that
debit the account are removed to make the clearing successful.
[0068] Once some groups have been removed, the aggregation is
re-computed, and the clearing repeated. Eventually, all netted
obligations can be cleared, and the groups that still remain in the
aggregation are cleared and settled together.
[0069] For example, consider the following scenario, consisting of
4 groups, and 9 obligations. Assume A2 has no holding, and that A3
has a holding (of the selected instrument) of 40. TABLE-US-00001
Group A1 obligations A2 obligations A3 obligations G1 +30 -30 G2
-10 +10 G3 +20 -20 G4 +20 -20 Total +50 -10 -40
[0070] First, note that using incremental clearing, either G1 or G4
can be cleared, but not both. Also, neither G2 nor G3 can be
cleared incrementally (since A2 is empty).
[0071] Now, using aggregated clearing, three obligations result:
+50 against A1, -10 against A2, and 31 40 against A3.
[0072] Since A2 is empty, these three obligations cannot clear
simultaneously.
[0073] Using the principle in step 5 above, either G2 or G3 is
removed, since this suffices to make the clearing of A2
succeed.
[0074] Let us assume G2 is removed. The aggregation then results
in: TABLE-US-00002 Group A1 obligations A2 obligations A3
obligations G1 +30 -30 G3 +20 -20 G4 +20 -20 Total +50 .+-.0
-50
[0075] That is, two obligations result: +50 against A1, and -50
against A3.
[0076] Since A3 holds an available volume of 40, these two
obligations cannot clear simultaneously.
[0077] Using the principle in step 5 above, either G1 or G4 is
removed, since this suffices to make the clearing of A3
succeed.
[0078] Let us assume G1 is removed. The aggregation then results
in: TABLE-US-00003 Group A1 obligations A2 obligations A3
obligations G3 +20 -20 G4 +20 -20 Total +20 .+-.0 -20
[0079] These net obligations (+20, -20) can be cleared, and the
aggregated clearing thus completes.
[0080] However, note that if we in the original list of groups were
to remove G3 instead of G2, the result is: TABLE-US-00004 Group A1
obligations A2 obligations A3 obligations G1 +30 -30 G2 -10 +10 G4
+20 -20 Total +30 +10 -40
[0081] This time, the net obligations (+30, +10, -40) can be
cleared without further removal of groups. That is, the choice of
which groups to remove affects the convergence of the aggregated
clearing.
[0082] Thus, aggregated clearing does not remove the minumum number
of groups, but rather tries to remove the minimal number in each
step.
[0083] As the simplest possible heuristic of what to remove in the
clearing, consider the following version of the removal, which has
the advantage of being easy to explain. if V<0, check if
Av+V>=0. If so, continue in 4. Otherwise, remove all groups H1,
. . . , Hm from G, where Hi contains an obligation (A, I, _, _),
and continue at 3.
[0084] That is, all groups referencing the overdrawn account are
removed. This leads to quick convergence, at the expense of
removing a larger number of groups.
[0085] Consider the example again: TABLE-US-00005 Group A1
obligations A2 obligations A3 obligations G1 +30 -30 G2 -10 +10 G3
+20 -20 G4 +20 -20 Total +50 -10 -40
[0086] The new heuristic would thus remove G2, G3, and G4, giving:
TABLE-US-00006 Group A1 obligations A2 obligations A3 obligations
G1 +30 -30 Total +30 .+-.0 -30
[0087] The settlement system could also provide another clearing
mechanism that can be invoked pre-business to complete as many
pending trades as possible ahead of the market opening (this to
cater for previously received trades to be settled during the
current business date).
[0088] It is important to include as many groups as possible (to
complete as many trades as possible) in the clearing, which is why
we introduce the notion of batched clearing. This to allow us to
use a more advanced clearing algorithm than can normally be invoked
during business hours (for performance reasons).
[0089] However, we do not intend to produce the optimal result in
the batched run, i.e. clearing as many groups as possible, since
the clearing problem is an integer programming problem. Such
problems lack efficient optimization algorithms for large
instances, and instead we seek an approximation algorithm.
[0090] First, consider an integer programming problem. The
aggregated clearing problem is formalized as follows. [0091] 1. Let
G=(G1, . . . , Gn) be a list of groups. [0092] 2. For each group Gi
in the list, let Oi=(o1, . . . , ok) be its list of obligations.
[0093] 3. Let xi in (0, 1), such that xi=1 if Gi is cleared by the
aggregation (netting). [0094] 4. Maximize x1+. . . +xn (i.e.
maximize the number of cleared groups), given the following
constraints. [0095] 5. For a given account A and instrument I,
where the available volume of AI is Av, the clearing constraint is
Av+v1.times.1+. . . +vnxn>=0, where vi=V if (A, I, V) in Gi, and
0 otherwise. That is, this constraint states that the aggregation
of each debit and credit operation of any group in the solution is
less or equal to the available volume of the affected account/cash
record. There will be one such constraint per account/cash
record.
[0096] A solution to the above is such that each volume constraint
is satisfied, and the number of groups included in the clearing is
the maximum.
[0097] Consider the sample used above. TABLE-US-00007 Group A1
obligations A2 obligations A3 obligations G1 +30 -30 G2 -10 +10 G3
+20 -20 G4 +20 -20
[0098] The corresponding equations are (assuming A1 holds 100, A2
holds 0, and A3 holds 40), using the formalization: Maximize
x1+x2+x3+x4, where 30.times.1+20.times.1<=100 (A1)
-10.times.2-20.times.3+20.times.4<=0 (A2)
-30.times.1+10.times.2-20.times.4<=40 (A3)
[0099] The optimal solution assigns x1=1, x2=1, x3=0, and x4=1,
i.e. G3 is excluded.
[0100] In general, there may be more than one answer to the
optimization problem.
[0101] Given that we may encounter more than 100 000 trades, the
number of variables could exceed 100 000, which is very difficult
to solve. Therefore, we need to look at large-scale algorithms that
can handle the worst-case. [0102] 1. Let (AI, . . . ) be a list of
all holdings/positions referenced by any group to be settled,
sorted in decreasing order by the number of pending debit
obligations against the holding. [0103] 2. Enumerate over the list,
choosing AI each time. [0104] 3. From AI, retrieve the list of all
groups (G1, . . . , Gn) that contains an obligation (A, I, _, _).
[0105] 4. For each obligation (B, J. _, _) and retrieved group,
retrieve the list of all groups that contain an obligation (B, J,
_, _). Eliminate BJ from the holding list accordingly. [0106] 5.
Repeat 4 until no more groups are retrieved or until a threshold
has been reached (configurable). Hence, the strongly connected
component that originates with the list in 3 has been computed.
[0107] 6. From the list of retrieved groups, run aggregated
clearing, using the aggregated clearing algorithm above, and settle
the cleared groups. [0108] 7. Go back to 2, enumerating over the
next holding AI. [0109] 8. When the enumeration is finished, all
groups remaining as unsettled make up the trades that failed to
settle.
[0110] As an alternative in step 6, use the following algorithm.
Let there be k constraints in the following, and p a pre-chosen
probability (set to around 0.80). [0111] 1. Initially, let all x1,
. . . , xn be set to 0. [0112] 2. Let the current assignment be the
optimal solution. [0113] 3. Flip 50% of the variables, chosen
randomly among those set to 0, hence, setting them to 1. [0114] 4.
If all k constraints are true, and fewer than n variables are set
to 1, continue in 2. [0115] 5. If some constraints are false, 1
say, then with probability p, flip the value of xi, where the new
value of xi reduces 1 more than flipping the value of any other xj
does, and with probability 1-p, flip the value of some randomly
chosen xi. [0116] 6. Continue at 4, unless n*log(n) flips have been
made. [0117] 7. If time allows (configurable), repeat at 1, keeping
the current optimum. Otherwise, return the best of the current
optimum and the previous optimum.
[0118] That is, starting from a random assignment of which groups
to include in the aggregation, given a connected component, we
search locally for improvements to the current assignment. See the
flow outline shown in FIG. 3.
[0119] The algorithm iterates making partly random choices to
include as many groups as possible in the final solution. It starts
from a initial solution, where no groups are included, and randomly
includes as many groups as possible until at least one constraint
is violated (using a Zeno-like inclusion).
[0120] Once a constraint has been violated, the algorithm tries to
improve the current assignment by flipping the variable that
improves the assignment the most, but occasionally flipping some
variables chosen randomly. Hence, this strives to come closer to a
solution.
[0121] This repeats until too many iterations have been made, or a
solution is found. At this point, the solution is saved, and the
process is repeated by greedily trying to add yet other groups.
[0122] Eventually, the limit of iterations is reached, and if time
allows, the whole process starts over, in hope of finding an even
better solution.
[0123] Consider the graph sketch in FIG. 4 for an example.
[0124] FIG. 4 is meant to illustrate a graph of groups, where G11
is particularly connected (to G1, G3, G5, G7), and hence possibly a
candidate to eliminate if it contains debit operations that collide
with those of its neighbors. ##STR1##
[0125] Hence, two steps in the iterative algorithm may thus be:
[0126] Where at some point, G11 is first included, i.e. x11 is set
to 1, and then later its value is flipped, i.e. G11 is excluded by
setting x11 to 0.
[0127] This algorithm is known to be very effective in quickly
finding good solutions.
[0128] Another challenge with this algorithm is to make its
flip-evaluation efficient, since it will be based on quickly
propagating the effects of including/excluding a group from the
aggregation. However, this is rather straightforward, since the
function is simply based on applying obligations to a set of
accounts/cash records (this is what happens when an xi is flipped,
i.e the group Gi is evaluated, either by removing its effect when
setting xi=0, or adding its effect when setting xi=1).
[0129] For example, the flip evaluation can be described as
follows. TABLE-US-00008 New value A1 A2 x1 = 0 +10 -10 x1 = 1 -10
+10 . . .
[0130] where G1 is described as follows: TABLE-US-00009 Group
Obligations G1 (A1, I, -10), (A2, I, +10)
[0131] That is, whenever x1 is set to 0, 10 is added of I to A1,
and 10 is subtracted from A2. On the other hand, when x1 is set to
1, 10 is subtracted from A1, and added to A2.
[0132] Thus, the representation of the group component is such that
a list of affected accounts and cash records is maintained, and
whenever a group is evaluated either for inclusion or exclusion,
the corresponding accounts, and any violated volume constraint are
quickly identified. This should make the performance of the
iterative algorithm sufficient.
The Settlement System
[0133] The settlement system can further be described with the
following features.
[0134] For any generic and flexible piece of functionality there is
a need to define the actual behaviour in a certain scenario. This
can be achieved in two ways: [0135] by implementing specific
adaptations of the generic functionality, or [0136] by having a
number of parameters defining the actual "flavour" of the generic
functionality
[0137] For a CSD system product to be efficient, both for the
system vendor and the operators of the system, the actual solution
could be one of the two ways or a mix of both. This part of the
description is aimed at the need for a generic functionality to
define, control and run the different settlement processes, which a
CSD operator may need to support.
[0138] In one known settlement system there is the concept of
settlement rules. A settlement rule is an object, which defines how
the matching, clearing and settlement process of the system will
behave for one specific settlement instruction (SI), settlement
obligation (SO) or settlement obligation group (SOG). The different
settlement rules can be administered to adjust and adhere to market
needs and behaviour.
[0139] In order for the system to choose the correct rule to apply
at a given step in a process, the system must filter out which rule
to use. For the applicable, valid, settlement rule of a specific
purpose (e.g. matching or lock-in of securities) the combination of
the following attributes (the "fingerprint") of the SI/SO/SOG must
be unique in order to perform the filtering: [0140] Source (e.g. an
external system or an internal system module) [0141] Payment
Currency [0142] Transaction type, i.e. the purpose of the
transaction based on: [0143] i. Message Type (e.g. DvP, FoP or PvN)
[0144] ii. Operation Type (e.g. CSD link, maturity payment or
normal trade settlement) [0145] Settlement Method (e.g. RTGS or BIS
Model 2) [0146] External Instrument Class (e.g. fixed income
securities or equities)
[0147] The actual rules (i.e. how the system should behave) are
given by the system implementation (as specified by the CSD).
[0148] To control what dates and times the different settlement
rules are valid, they are listed in the subsessions as valid
settlement rules for that sub session.
[0149] For the disclosed settlement system it is suggested that a
number of matching rules and optimization routines are added as
well as the concept of grouping settlement instructions that are
due to settle together (but not necessarily simultaneously).
[0150] The generic clearing and settlement process in the
settlement system, which is not explained in details here, could be
summarized as in FIG. 5.
[0151] At a number of steps in the process there is a need to make
a decision on how to proceed. These "waypoints" are marked in FIG.
5 as "Initiation based on rules". It should be possible to filter
out any settlement rule at the waypoints in the process, i.e. if a
SI had a fingerprint based on a certain source, transaction type
and instrument class it could e.g. filter out a rule for matching
based on the source only and then filter out a rule for
optimization based on transaction type and instrument class. The
process at the waypoints can be described in FIG. 6.
[0152] The number of possible combination of fingerprints times the
number of waypoints makes the potential mix of rules complex.
However the potential flexibility makes the solution very dynamic.
In order to reduce the complexity some or all of the steps can be
given in the system for certain types of transactions, e.g. for SIs
related to corporate actions. That is, on certain conditions the
system will select the appropriate rule to use in the different
steps without filtering out the rules based on the fingerprint.
Examples of Transaction Flows
[0153] In order to visualize the rather complex flow at the
different waypoints a number of examples are given in FIG. 7.
Realt Time Gross Settlement (RTGS)
[0154] An RTGS would consist of two settlement instructions,
created based on e.g. a stock exchange trade or an OTC trade. The
two SIs would be grouped in one SIG and when selected for SO/SOG
creation they would be mapped to one SO each without any kind of
intermediate calculation.
[0155] The lock-in would be on a gross volume/amount and when time
for settlement (typically as soon as possible on S), the accounts
and cash records would be debited/credited simultaneously.
Optimized RTGS
[0156] The settlement of a group of RTGS transactions is valid as
an example of an optimized settlement process: Out of four matched
pairs of SIs, three are selected to participate in one SIG aimed
for optimization. The selection criteria may be e.g. Source=Stock
Exchange. When creating the SOs to lock-in the system would try to
optimize to settle e.g. as many transactions as possible with as
little liquidity use as possible.
[0157] Let us assume that the optimization process resulted in four
obligations, two to lock-in securities and two to lock-in cash,
calculated as the netted obligations per instrument/currency and
account/cash record. These four SOs volumes/amounts would be enough
to cover all the six SIs included in the SIG.
[0158] When time for settlement the locked-in assets of the four
SOs would be transferred (as one SOG) and the six SIs would be
considered settled (gross) delivery versus payment(DvP) at the same
time, as shown in FIG. 8.
Netting According to BIS Model 2
[0159] In a BIS (Bank of International Settlement) Model 2 netting
scenario the securities would be settled gross and the cash would
be settled net, by means of novated netted cash obligations.
[0160] Out of four matched pairs of SIs, all four are selected to
participate in one SIG aimed for the netting. The selection
criteria may be e.g. External Instrument Class=FIS (fixed income
securities). When creating the SOs to lock-in the system would net
the cash obligations of each involved cash record.
[0161] Let us assume that the netting process resulted in two net
cash obligation and four to lock-in securities. These six SOs
volumes/amounts would cover the eight SIs included in the SIG.
[0162] When time for settlement the locked-in assets of the six SOs
would be transferred (as one SOG) and the eight SIs would be
considered settled (net) DvP at the same time, as shown in FIG.
9.
[0163] An embodiment of a securities settlement system 1001 capable
of performing the clearing and settlement methods and procedures
disclosed above is shown in FIG. 10.
[0164] The securities settlement system (SSS) 1001 can communicate
with a trading system 1002 in which matching of deals or trdes take
place. These trades origin from orders made by users 1003, using
trading stations or similar tools for sending in orders to buy/sell
to the trading system 1002. Within the SSS 1001 there is an input
1004 which, apart from receiving orders from the trading system
1002, may sort, modify and store the received information in an
appropriate way to facilitate further processing.
[0165] In a selector 1005 at least some of the received trades will
be selected and grouped together. Selection may be made on a number
of parameters, such as user (trade parties) and security type.
[0166] The selected group is forwarded to an aggregation unit 1006
for determination of an aggregated obligation required to be met in
order to clear (or settle) all the trades. The possible processes
for doing so have already been explained above so no further
description of this is necessary at this point.
[0167] The actual accounts held by the users of the system may be
positioned outside of the SSS 1001, as shown by data account memory
1007.
[0168] In FIG. 11 an alternative embodiment of a securities
settlement system 1101 according to the invention is shown.
[0169] The securities settlement system 1101 comprises an input
1102 for receiving trade information. Received trade information is
then sorted in a sorter 1103 in accordance with a set of sorting
criteria. It may thus select all trades involving a certain user or
group of users, all trades related to a specific instrument type or
instrument types, all trades related to a specific market or
markets or a combination of two or more of these. Other selectable
criteria can also be made.
[0170] Once a selection has been made, the selected group of trades
proceeds to an aggregation unit 1104 for aggregation of obligations
for each user (or account). Following that all aggregated
obligations are compared with each user's (account's) obligation
limit to find out if all trades can be cleared at
simultaneously.
[0171] Should any obligation not be met, one or more trades must be
removed, which is done in the selector unit 1106 inaccordance with
what has been discussed above. Removed trades are sent back to the
selector 1103 and the other trades are once again sent to the
aggregator unit 1104 for aggregation of obligations. Once a
complete group can be cleared, it is sent to a finalizing unit 1107
for settlement.
[0172] The above embodiments are only examples of how the invention
can be realized. A full rendering of the invention is embodied in
the accompanying claims.
* * * * *