Notice2024-30916
Self-Regulatory Organizations; NYSE Arca, Inc.; Notice of Filing of Proposed Rule Change To Amend Rule 7.31-E To Adopt the Selective Midpoint Order
Primary source
Metadata and text below are from the Federal Register, a public-domain U.S. government work. Always verify the official published version before relying on it for any legal matter.
Published
December 30, 2024
Issuing agencies
Securities and Exchange Commission
Full Text
<html>
<head>
<title>Federal Register, Volume 89 Issue 249 (Monday, December 30, 2024)</title>
</head>
<body><pre>
[Federal Register Volume 89, Number 249 (Monday, December 30, 2024)]
[Notices]
[Pages 106630-106635]
From the Federal Register Online via the Government Publishing Office [<a href="http://www.gpo.gov">www.gpo.gov</a>]
[FR Doc No: 2024-30916]
=======================================================================
-----------------------------------------------------------------------
SECURITIES AND EXCHANGE COMMISSION
[Release No. 34-102005; File No. SR-NYSEARCA-2024-112]
Self-Regulatory Organizations; NYSE Arca, Inc.; Notice of Filing
of Proposed Rule Change To Amend Rule 7.31-E To Adopt the Selective
Midpoint Order
December 19, 2024.
Pursuant to Section 19(b)(1) of the Securities Exchange Act of 1934
(the ``Act''),\1\ and Rule 19b-4 thereunder,\2\ notice is hereby given
that on December 18, 2024, NYSE Arca, Inc. (``NYSE Arca'' or the
``Exchange'') filed with the Securities and Exchange Commission (the
``Commission'') the proposed rule change as described in Items I, II,
and III below, which Items have been prepared by the Exchange. The
Commission is publishing this notice to solicit comments on the
proposed rule change from interested persons.
---------------------------------------------------------------------------
\1\ 15 U.S.C. 78s(b)(1).
\2\ 17 CFR 240.19b-4.
---------------------------------------------------------------------------
I. Self-Regulatory Organization's Statement of the Terms of Substance
of the Proposed Rule Change
The Exchange proposes to amend Rule 7.31-E to adopt the Selective
Midpoint Order. The proposed rule change is available on the Exchange's
website at <a href="http://www.nyse.com">www.nyse.com</a>, at the principal office of the Exchange, and
at the Commission's Public Reference Room.
II. Self-Regulatory Organization's Statement of the Purpose of, and
Statutory Basis for, the Proposed Rule Change
In its filing with the Commission, the self-regulatory organization
included statements concerning the purpose of, and basis for, the
proposed rule change and discussed any comments it received on the
proposed rule change. The text of those statements may be examined at
the places specified in Item IV below. The Exchange has prepared
summaries, set forth in sections A, B, and C below, of the most
significant parts of such statements.
A. Self-Regulatory Organization's Statement of the Purpose of, and the
Statutory Basis for, the Proposed Rule Change
1. Purpose
The Exchange proposes to amend Rule 7.31-E(h)(3) to decommission
the Discretionary Pegged Order (``DPO'') and introduce the Selective
Midpoint Order (``SeMi Order''). The SeMi Order will be similar to the
DPO in that it is a discretionary order type, but will, unlike the
current DPO, provide price protection during periods of market
instability based on input from a gradient-boosting machine learning
model.
Background
The Exchange currently offers the DPO, which is a non-displayed
order to buy (sell) that is pegged to the same side of the PBBO and
assigned a working price equal to the lower (higher) of the midpoint of
the PBBO (the ``Midpoint Price'') or the limit price of the order.\3\
Any untraded shares of such order are assigned a working price equal to
the lower (higher) of PBB (PBO) or the order's limit price, which is
automatically adjusted in response to changes to the PBB (PBO) for buy
(sell) orders up (down) to the order's limit price. A DPO will exercise
the least amount of discretion necessary from its working price to its
discretionary price (defined as the lower (higher) of the Midpoint
Price or the limit price of the order) to trade with contra-side
interest.
---------------------------------------------------------------------------
\3\ See Rule 7.31-E(h)(3). As defined in NYSE Arca Rule 1.1,
``PBBO'' means the Best Protected Bid and the Best Protected Offer.
Rule 1.1 also defines ``PBB'' as the highest Protected Bid and
``PBO'' as the lowest Protected Offer.
---------------------------------------------------------------------------
Prior to November 2022, the DPO would not exercise discretion if
the PBBO was determined to be unstable via a ``quote instability
calculation'' that assessed the probability of a change to the PBB or
PBO.\4\ The Exchange used the quote instability calculation along with
real-time relative quoting activity of protected quotations to assess
the probability of an imminent change to the PBBO (the ``quote
instability factor''). When the quoting activity met certain predefined
criteria and the quote instability factor calculated was greater than
the Exchange's predefined quote instability threshold, the Exchange
treated the quote as unstable and restricted DPOs from exercising
discretion. In November 2022, the Exchange amended Rule 7.31-E(h)(3) to
eliminate the quote stability calculation, allowing DPOs to exercise
discretion even during potential periods of quote instability.\5\
---------------------------------------------------------------------------
\4\ See Securities Exchange Act Release No. 95154 (June 24,
2022), 87 FR 39134 (June 30, 2022) (SR-NYSEArca-2022-13) (Notice of
Filing of Amendment No. 2 and Order Granting Accelerated Approval of
a Proposed Rule Change, as Modified by Amendment No. 2, To Amend
Rule 7.31-E(h)(3) Relating to Discretionary Pegged Orders).
\5\ See Securities Exchange Act Release No. 96322 (November 15,
2022), 87 FR 69376 (November 18, 2022) (SR-NYSEARCA-2022-76) (Notice
of Filing and Immediate Effectiveness of Proposed Rule Change to
Amend Rule 7.31-E). The Exchange resumed offering the DPO in
November 2022, after previously filing to temporarily suspend its
use in August 2022. See Securities Exchange Act Release No. 95584
(August 23, 2022), 87 FR 52826 (August 29, 2022) (SR-NYSEARCA-2022-
54) (Notice of Filing and Immediate Effectiveness of Proposed Rule
Change To Amend Rule 7.31-E).
---------------------------------------------------------------------------
Proposed Rule Change
The Exchange proposes to modify Rule 7.31-E(h)(3) to replace the
DPO with the SeMi Order. The SeMi Order will share the same basic
attributes as the DPO. Like the DPO, the SeMi Order would be a non-
displayed order to buy (sell) that is pegged to the same side of the
PBBO and assigned a working price equal to the lower (higher) of the
Midpoint Price or the limit price of the order. Any untraded shares of
a SeMi Order would be assigned a working price equal to lower (higher)
of the PBB (PBO) or the order's limit price and automatically adjusted
in response to changes to the PBB (PBO) for buy (sell) orders up (down)
to the order's limit price. In order to trade with contra-side orders
on the NYSE Arca Book, a SeMi Order to buy (sell) would exercise the
least amount of price discretion necessary from its working price to
its discretionary price, which is defined as the lower (higher) of the
Midpoint Price or the SeMi Order's limit price.
SeMi Orders would not be displayed, must be designated Day, and
would be eligible to be designated for the Core Trading Session only.
SeMi Orders designated for the Early Trading Session or Late Trading
Session would be rejected.
When exercising discretion, SeMi Orders (like DPOs today) would
maintain their time priority at their working price as Priority 3--Non-
Display Orders and are prioritized behind Priority 3--Non-Display
Orders with a working price equal to the discretionary price of a SeMi
Order at the time of execution. If multiple SeMi
[[Page 106631]]
Orders are exercising price discretion during the same book processing
action, they would maintain their relative time priority at the
discretionary price.
Accordingly, the Exchange proposes to replace existing references
to the ``Discretionary Pegged Order'' in Rule 7.31-E(h)(3) and
subparagraphs (A) and (B) thereunder with references to the ``Selective
Midpoint Order.'' \6\
---------------------------------------------------------------------------
\6\ The Exchange also proposes conforming changes in Rules 7.18-
E(b)(1), 7.18-E(c)(1) and (5), 7.31-E(i)(3), and 7.34-E(c)(1)(A) to
replace references to ``Discretionary Pegged Orders'' with
references to ``Selective Midpoint Orders.''
---------------------------------------------------------------------------
The Exchange proposes to add new Rule 7.31-E(h)(3)(C) to provide
that the SeMi Order, as proposed, would exercise price discretion to
its discretionary price, except during periods of quote instability as
identified by the Selective Midpoint Indicator (``SMI'') (as discussed
in further detail below). If the SMI determines the PBBO for a
particular security to be an unstable quote, both an arriving and
resting SeMi Order would wait for a PBBO that is stable before the
order's working price is adjusted and the order becomes eligible to
trade. In other words, a SeMi Order would be ineligible to trade when
the SMI identifies unstable market conditions and would remain in that
state until the SMI determines that market conditions have stabilized,
thereby preventing potentially undesirable executions during volatile
or unstable market conditions. Whereas the DPO previously relied on one
static logistical regression model to forecast market instability and
prevented DPOs in any symbol from exercising discretion to trade when
the model anticipated an unstable market, the SeMi Order, as proposed,
would rely on the SMI to predict market instability using a symbol-
specific gradient-boosting machine learning model and would protect
SeMi Orders from trading when the SMI predicts quote instability for a
given symbol.
The Exchange also proposes new Rule 7.31-E(h)(3)(D) to provide for
SeMi Orders to be optionally designated as Liquidity Providing.
Proposed Rule 7.31-E(h)(3)(D)(i) would provide that a SeMi Order
designated as Liquidity Providing will only execute on arrival against
resting orders that include a Non-Display Remove Modifier and are
priced within the Liquidity Providing SeMi Order's discretionary
range.\7\
---------------------------------------------------------------------------
\7\ The Exchange also proposes related conforming changes to
Rules 7.31-E(d)(2)(B), 7.31-E(d)(3)(F), and 7.31-E(e)(1)(C) to
provide that Non-Displayed Limit Orders, MPL Orders, and Non-
Routable Limit Orders designated with a Non-Display Remove Modifier
will trade as takers against Liquidity Providing SeMi Orders.
---------------------------------------------------------------------------
Proposed Rule 7.31-E(h)(3)(D)(ii) would provide that if a resting
contra-side order that does not include a Non-Display Remove Modifier
is priced within an arriving Liquidity Providing SeMi Order's
discretionary range, the Liquidity Providing SeMi Order will be placed
on the NYSE Arca Book, and its discretionary range will be adjusted to
equal the resting price of a non-displayed contra-side order or to one
MPV less aggressive than the resting price of a displayed contra-side
order.\8\
---------------------------------------------------------------------------
\8\ The Exchange notes that allowing Liquidity Providing SeMi
Orders to trade with resting orders with a Non-Display Remove
Modifier, as well as adjusting the discretionary range of such
orders, would be consistent with how similar discretionary order
types function on other equities exchanges. See, e.g., Cboe EDGX
Exchange, Rule 11.8(g)(5) (``[Midpoint Discretionary Orders
(``MDOs'')] that are not entered with a [Quote Depletion Protection
(``QDP'')] instruction, as defined in Rule 11.8(g)(10), will only
act as the liquidity provider. MDOs entered with a QDP instruction
will instead be allowed to remove liquidity, by default, unless the
User chooses to require that the MDO only act as a liquidity
provider. If the instructions included on an MDO do not permit the
MDO to remove liquidity, it will only execute on entry against
resting orders that include a Super Aggressive instruction priced at
the MDO's pegged price if the MDO also contains a Displayed
instruction, and against resting orders that include an NDS
instruction priced either at the MDO's pegged price or within its
discretionary range. If a resting contra-side order that does not
include [a Non-Displayed Swap] instruction is priced within the
discretionary range of an incoming MDO that is not permitted to
remove liquidity, the incoming MDO will be placed on the EDGX Book
and its discretionary range will be shortened to equal the limit
price of the resting contra-side order. Likewise, where an incoming
order with a Post Only instruction does not remove liquidity on
entry pursuant to Rule 11.6(n)(4) against a resting MDO, the
discretionary range of the resting MDO will be shortened to equal
the limit price of the incoming contra-side order with a Post Only
instruction.'').
---------------------------------------------------------------------------
Proposed Rule 7.31-E(h)(3)(D)(iii) provides that a Liquidity
Providing SeMi Order resting on the NYSE Arca Book will not trade with
an arriving contra-side order that cannot remove liquidity.\9\ Once
such arriving contra-side order is placed on the NYSE Arca Book, the
discretionary range of the Liquidity Providing SeMi Order will be
adjusted to equal the resting price of a non-displayed contra-side
order or to one MPV less aggressive than the resting price of a
displayed contra-side order.
---------------------------------------------------------------------------
\9\ The Exchange notes that this proposed handling is also
consistent with the handling of similar discretionary order types by
other equities exchanges. See id.
---------------------------------------------------------------------------
Proposed Rule 7.31-E(h)(3)(D)(iv) would provide that, once resting
on the NYSE Arca Book, the discretionary range of a Liquidity Providing
SeMi Order will be adjusted based on resting contra-side interest as
described in proposed subparagraphs (ii) and (iii) of this Rule when
its working price Changes. In addition, proposed Rule 7.31-
E(h)(3)(D)(iv)(a) and (b) provide that a Liquidity Providing SeMi Order
to buy (sell) will not be eligible to trade at a price equal to or
above (below) any sell (buy) orders that are displayed and that have a
working price equal to or below (above) the working price of such
Liquidity Providing SeMi Order, or at a price above (below) any sell
(buy) orders that are not displayed and that have a working price below
(above) the working price of such Liquidity Providing SeMi Order.
Finally, the Exchange proposes to renumber current Rule 7.31-
E(h)(3)(C) as 7.31-E(h)(3)(E). The Exchange proposes that, as with
DPOs, if the PBBO is locked or crossed, SeMi Orders would wait for a
PBBO that is not locked or crossed before the working price is adjusted
and the order becomes eligible to trade. Accordingly, the Exchange
proposes to substitute ``Selective Midpoint Order'' for ``Discretionary
Pegged Order'' in the text of proposed Rule 7.31-E(h)(3)(E).
Selective Midpoint Indicator
As described in further detail in the white paper attached as
Exhibit 3 to this proposed rule change,\10\ the SMI was developed using
a decision tree model, which is a supervised learning algorithm used
for classification and regression tasks. A decision tree model is
trained sequentially, and each successive model seeks to improve on
errors in the previous model by focusing on accurately predicting where
the previous model performs poorly.
---------------------------------------------------------------------------
\10\ See Exhibit 3, Selective Midpoint Indicator (SMI): A
Gradient Boosted Signal Enabling Stable Order Executions (the
``White Paper'').
---------------------------------------------------------------------------
The Exchange determined to use a decision tree model in developing
the SMI in large part because of the transparency it offers to its
designers. The output of a decision tree is a hierarchy of questions
that allows a user to follow the model's decision-making process,
assess the importance of a given feature to the model's output, and
examine the reasons underlying a specific output from the model. The
Exchange believed it was important to build a model that could be
interpreted and understood in this way to allow for the evaluation of,
among other things, the relationships between market shifts, feature
selection, and feature weightings, as well as to be able to assess
overall model performance.
Market Instability Assessment
The Exchange set out to develop a model that would allow the SMI to
predict market instability. For purposes of the model, the Exchange
defined
[[Page 106632]]
``instability'' at a high level as relatively large price moves during
a relatively short time frame. The Exchange chose PBBO updates as the
fundamental data points for the model based on the evenly distributed
nature of PBBO updates throughout the trading day and the granular
level of information such updates offer. The model's objective is to
identify windows where changes to the PBBO are unstable, in order to
predict unstable markets in real time and prevent SeMi Order trading
accordingly.
The Exchange established the concept of a ``price jump'' to further
understand and categorize periods of instability. A price jump is
defined as a PBBO mid-price change of a pre-defined percentage of a
symbol's spread (referred to as the spread threshold X), in either
direction, within a configurable time interval (referred to as the time
horizon G). The Exchange sought to identify price jumps following a
PBBO update, whether positive or negative. A price jump is marked at a
given point in time if, looking back to the start of the time horizon,
the mid-price was at least the spread threshold in difference from the
current mid-price. This approach identifies discrete price jumps over
configurable time intervals, thereby offering the Exchange flexibility
to adjust the spread threshold and/or time horizon parameters according
to symbol-specific dynamics.\11\
---------------------------------------------------------------------------
\11\ Figure 1 in Section 4.1 of the White Paper provides a
detailed example of how a price jump is calculated.
---------------------------------------------------------------------------
To build the SMI as a continuous signal of market instability, the
Exchange next applied the price jump definition to delineate continuous
periods of market instability. At each PBBO update (at index time i),
the Exchange determined whether there had been a price jump. If there
was no price jump, the market would be considered stable. If there was
a price jump, the Exchange continued to look for price jumps within the
time horizon G, until no price jump occurred. At that point, the
Exchange would mark an unstable time window starting at the PBBO update
that occurred at i-1 or 50 microseconds prior to the PBBO update at
time i, whichever is closest to i, and ending at the last price jump
identified.\12\
---------------------------------------------------------------------------
\12\ Figure 2 in Section 4.2 of the White Paper provides an
example of the identification of unstable windows.
---------------------------------------------------------------------------
Finally, the Exchange employed an additional parameter for a
minimum time increment g to establish that unstable windows identified
using the spread threshold and time horizon parameters are indicative
of persistent price instability. If multiple price jumps occur within a
small timeframe, such that the distance between the first and last
price jumps is less than g, the Exchange would not mark the window as
unstable. The application of the minimum time increment parameter
allows the Exchange to more accurately identify true periods of
instability by filtering out temporary price jumps that are quickly
followed by a reversion to the price prior to the observed price
jump.\13\
---------------------------------------------------------------------------
\13\ Figure 3 in Section 4.2 of the White Paper demonstrates the
operation of the minimum time increment parameter.
---------------------------------------------------------------------------
In developing the SMI to facilitate the SeMi Order's ability to
provide protection against potentially unfavorable executions, the
Exchange also wanted to be able to differentiate between quote
instability on the bid side and ask side. For example, the Exchange
would not want to prevent executions of buy orders based on upwards
instability of the mid-price of the PBBO because executions under those
conditions would likely be favorable. Accordingly, if the mid-price of
the PBBO is higher at the end of an unstable window, the Exchange would
only mark PBBO updates during such window as unstable for the ask side
(and vice versa for the bid side).
Model Development
To develop and train the models underlying the SMI, the Exchange
used data from the NYSE Arca Book from August 29, 2024 through October
22, 2024 for a set of 500 symbols selected to reflect a representative
sample of the U.S. equity markets.\14\ The Exchange created evaluation
data points for each PBBO update that include the closest book depth
state at the time of that update. This merged data set of PBBO updates
and book depth data has the benefit of the information from both types
of data without overinflating the size of the data set. All features
and parameters used by the models are calculated based on this merged
data set.
---------------------------------------------------------------------------
\14\ The full symbol list is included in Appendix B of the White
Paper. Symbols were chosen based on criteria including absolute
price level, spread in dollars, spread in basis points, and
liquidity (daily ADV). Because stable periods generally far
outnumber unstable periods in the U.S. equity markets, the Exchange
used under-sampling methods where appropriate to reduce the number
of stable data points in the data set and randomly shuffled data
before training.
---------------------------------------------------------------------------
The Exchange identified a set of features based on NYSE Arca Book
data--e.g., book depth information, PBBO updates, number of IOC
orders--that would contribute to the models. The Exchange selected 83
unique features that could be considered for incorporation into the
model.\15\ These features were identified based on an iterative
process, and features were selected based on their ability to explain
unstable periods identified by the application of the parameters
discussed above and, going forward, to predict market instability on a
real-time basis so that the SMI can effectively protect SeMi Orders
from potentially unfavorable executions. In production, the Exchange
proposes to select a subset of these 83 features to be weighted in the
model's assessment of market instability (rather than having the model
utilize all available features), because focusing on the features that
have been identified as most important in predicting market instability
for a given symbol would both optimize prediction accuracy and
processing speed.\16\
---------------------------------------------------------------------------
\15\ The full list of features is included in Appendix A to the
White Paper. Appendix A also identifies the subset of features that
were selected for the model training described in the White Paper.
\16\ Once the SMI is implemented in production, the full list of
features currently calculated in real-time and available for
evaluation for inclusion in the SMI models will be published daily
on the Exchange's website.
---------------------------------------------------------------------------
Symbol-Specific Models
The Exchange proposes that the SMI will rely on symbol-specific
models to leverage the ability of the models to incorporate different
features and weightings to respond to individual symbols' unique
profiles (e.g., the features most likely to accurately predict
instability for a given symbol). To explore how symbol-specific models
should be distributed, the Exchange ranked each of the 500 symbols in
the representative data set according to the total number of unstable
data points. The Exchange trained a model for each of the 500 symbols
and used the model trained for the symbol SPY as a default ``market
model.''
The Exchange found that the SMI performed better with a symbol-
specific model for more active symbols (i.e., those with more unstable
data points) than on the market model (i.e., testing showed sharp
declines in precision and increases in overlocking behavior using the
market model), whereas the SMI with a market model performed well for
less active symbols (i.e., high recall with relatively small loss of
precision and minimally more overlocking). The Exchange concluded that,
for the SMI to provide optimal information and protection to SeMi
Orders, more active symbols would benefit significantly from symbol-
specific models, while less active symbols (which have fewer unstable
data points to inform a symbol-
[[Page 106633]]
specific model) could successfully default to a market model.
The Exchange believes that models tailored to individual symbols'
specific characteristics would provide for better performance by the
SMI and thus enhanced price protection by the SeMi Order. Accordingly,
in production, the Exchange anticipates that at least 200 (and up to
approximately 1,000) symbols that trade on the Exchange will have an
individualized model that incorporates features that have been
specifically identified for predicting market stability for that
symbol. The remaining symbols that trade on the Exchange would use the
market model, which would apply the same features and weightings for
all symbols.\17\ Each day, the Exchange will identify the 1,000 symbols
with most unstable data points and evaluate those symbols to determine
whether a symbol-specific model or market model would yield better
performance. The Exchange will publish on its website a list of the
symbols that have an individual model to provide transparency to market
participants regarding the operation of SeMi Orders.
---------------------------------------------------------------------------
\17\ The Exchange proposes that a new symbol will operate on the
market model until the Exchange has gathered at least three days'
worth of data to be able to train a symbol-specific model and
determine whether it outperforms the market model for that symbol.
---------------------------------------------------------------------------
Performance Metrics
The Exchange strategically prioritized developing the models to
have a high rate of recall, which was intended to maximize the models'
ability to accurately capture unstable PBBO updates while accepting
that the models might identify more periods of instability than would
exist in realistic market conditions. The Exchange focused on three
metrics: (1) recall, or the model's ability to accurately identify true
unstable data points; (2) precision, or the model's ability to identify
only true unstable data points (i.e., to not misidentify stable data
points as unstable); and (3) overlocking, or the model's ability to
minimize the amount of time (measured in seconds) that the model
incorrectly predicts unstable market conditions.\18\ The Exchange
intends for the models to maximize recall and precision, while
minimizing overlocking.
---------------------------------------------------------------------------
\18\ Recall, precision, and overlocking are discussed in more
detail in Section 6.1 of the White Paper.
---------------------------------------------------------------------------
To evaluate the performance of the models, the Exchange selected
initial baseline values for each the three parameters X, G, and g to
maximize these performance metrics. To define an unstable PBBO period,
the Exchange selected a minimum time increment g of 100 microseconds;
spread threshold X of 25%; and time horizon G of one millisecond.\19\
---------------------------------------------------------------------------
\19\ The analysis the Exchange performed to arrive at these
parameter values is discussed in more detail in Section 6.2 of the
White Paper. The Exchange expects that the parameter values may
change over time to ensure proper calibration. The Exchange
anticipates implementing the SMI in production with these parameter
values, but will continue to analyze data and train the models until
the date of implementation and may update these values to the extent
that its analysis suggests that different values would improve
performance.
---------------------------------------------------------------------------
In final performance testing of the model, the aggregate results
demonstrate that the model achieves an average recall rate or 90% and
average precision of 30%, with overlocking occurring for an additional
3.8 seconds on average.\20\
---------------------------------------------------------------------------
\20\ Additional discussion of the Exchange's performance testing
of the models appears in Section 6.4 of the White Paper.
---------------------------------------------------------------------------
Production Integration
The models' compact size and average prediction speed of
approximately two microseconds allows for seamless integration of the
model's prediction process into the NYSE Pillar trading platform
(``Pillar'') on which the Exchange operates.\21\ The Exchange believes
that it has designed the SMI to produce an output rapidly enough to
keep up with real-time trading, without overburdening Exchange systems
or otherwise impacting current system performance.
---------------------------------------------------------------------------
\21\ Section 7 of the White Paper discusses the Exchange's
analysis of the model's prediction time in more detail, as well as
the integration of the model into the Pillar platform (see Figure
9).
---------------------------------------------------------------------------
Pillar will have access to full real-time trading data and will
continuously maintain required features for the model, including PBBO
updates, order entries, order cancellations, and book depth
information. The models are invoked as soon as an evaluation trigger
(e.g., a PBBO update) is received.\22\ The evaluation process concludes
by sending a message the Pillar matching engine to indicate the
beginning and end of an unstable period, which would inform whether
SeMi Orders are eligible to trade. For example, when Pillar receives a
SeMi Order, the SMI will indicate whether the market is stable or
unstable. If the market is stable, the SeMi Order will be allowed to
post to the NYSE Arca Book and trade. If the market is unstable, the
order will be prevented from trading until the SMI next predicts that
the market is stable. For as long as the SMI predicts that the market
will be unstable, a SeMi Order will remain ineligible to trade and will
not execute until the SMI evaluates the market as stable.
---------------------------------------------------------------------------
\22\ In addition to PBBO updates, Pillar will respond to a
timer-based evaluation trigger. The timer-based trigger is intended
to ensure that the model remains updated when the NYSE Arca Book
changes in the absence of a PBBO update (such as when non-displayed
liquidity is added to the NYSE Arca Book).
---------------------------------------------------------------------------
On a given trading day, the SMI models will use the feature weights
determined from the previous night's training, and the features will be
calculated using real-time intraday data. The list of symbols with
individualized models will be dynamically constructed daily and
published before the start of trading each day.
The model will not change intra-day. The Exchange proposes to
continue to retrain the model within the parameters described in this
filing and the White Paper daily, outside of the Core Trading Session
(on days when trading takes place on the Exchange). Retraining will be
based on the last three trading days' worth of historical data.
Retraining may result in changes to the features used by the model and/
or the weighting of such features. The values assigned to the three
parameters X, G, and g will not be adjusted as a result of regular
model training but may be updated periodically based on the Exchange's
analysis of overall model performance. Retraining is a standard and
accepted process in the use of machine learning models like the ones
underlying the SMI. The retraining process is not intended to result in
significant or unexpected changes to the performance of the SMI or the
behavior of the SeMi Order. Rather, retraining would help ensure that
the SMI continues to perform well in dynamic circumstances, by allowing
the models to learn from and incorporate more recent data points and
would facilitate improved model performance over time. The Exchange
also notes that retraining would build on the models' existing state
(i.e., existing data inputs and knowledge base) and would not alter the
model's objectives; retraining would result in new behaviors only to
the extent that the model had not previously encountered a given
scenario, and even then, any new behavior would be consistent with the
model's objectives. If the Exchange determines that a retrained model
would not be as successful as an existing model in achieving its
objectives based on the metrics defined above, the Exchange will not
implement the retrained model in production.
The Exchange will file a subsequent proposed rule change if it
seeks to modify the underlying structure of the models underlying the
SMI, such as the parameters X, G, and g used to label unstable windows
or new features that
[[Page 106634]]
could be incorporated into the models, but will not seek Commission
approval prior to retraining the models to adjust the weighting of
features that have been disclosed as potential inputs for the models or
modifications to the value of any of the three identified parameters.
The Exchange will also retain copies of each historical iteration of
the models as part of its books and records and will make such records
available to the Commission upon request. The Exchange will also
publish a Trader Update in advance of implementing a retrained version
of the SMI models when the Exchange has a reasonable belief that the
retrained version(s) would yield results that differ materially from
the prior version(s).
* * * * *
Because of the technology changes associated with this proposed
rule change, the Exchange will announce the implementation of this
change by Trader Update. Subject to approval of this rule filing, the
Exchange is prepared to implement the proposed rule change in 2025.
2. Statutory Basis
The proposed rule change is consistent with Section 6(b) of the
Act,\23\ in general, and furthers the objectives of Section
6(b)(5),\24\ in particular, because it is designed to prevent
fraudulent and manipulative acts and practices, to promote just and
equitable principles of trade, to foster cooperation and coordination
with persons engaged in facilitating transactions in securities, to
remove impediments to, and perfect the mechanism of, a free and open
market and a national market system and, in general, to protect
investors and the public interest.
---------------------------------------------------------------------------
\23\ 15 U.S.C. 78f(b).
\24\ 15 U.S.C. 78f(b)(5).
---------------------------------------------------------------------------
The Exchange believes that the proposed change to eliminate the DPO
and introduce the SeMi Order would remove impediments to, and perfect
the mechanism of, a free and open market and a national market system,
as well as protect investors and the public interest, by continuing to
provide market participants with the benefits of an order type that can
exercise discretion to trade with contra-side interest. The SeMi Order
will operate in a substantially similar manner to the existing DPO,
with the benefit of the SMI to provide price protection to SeMi Orders
during periods of market instability. The Exchange also believes that
the proposed rule change would remove impediments to, and perfect the
mechanism of, a free and open market and a national market system and
protect investors and the public interest because the SMI would provide
improved functionality as compared to the regression model previously
used by the DPO to predict market instability, which used static
parameters for all symbols. Specifically, the Exchange believes that
the SMI would provide enhanced price protection for SeMi Orders because
its fast, lightweight, and transparent models can be seamlessly
integrated into Pillar to predict future microsecond-level market
stability on a symbol-specific basis. The Exchange believes that SeMi
Orders, as proposed, would perfect the mechanism of, a free and open
market and a national market system and protect investors and the
public interest by relying on the SMI to restrict SeMi Orders from
trading during times of predicted high market volatility, thereby
avoiding potentially undesirable executions and increasing the
potential for price improvement for such orders at the cost of slightly
reduced fill rates.
The Exchange further believes that the proposed change to allow
SeMi Orders to be designated as Liquidity Providing (an option that was
not previously available to DPOs) would remove impediments to, and
perfect the mechanism of, a free and open market and a national market
system and protect investors and the public interest because it would
afford increased flexibility to users of the order type.
The Exchange believes that the proposed change would remove
impediments to, and perfect the mechanism of, a free and open market
and a national market system and promote just and equitable principles
of trade because the SeMi Order and SMI will operate within strict,
well-defined, and transparent parameters. Although the SMI models will
undergo daily retraining (outside of market hours), such retraining
will aim to improve the performance of the SMI in achieving its stated
objectives; retraining is not intended to alter the basic design
parameters, features, or objectives of the models without prior
Commission approval.\25\ Moreover, the Exchange will not deploy a
retrained model if it fails to achieve performance improvements based
on the metrics described above. As noted above, a list of all features
that may be incorporated in the models will be publicly available, and
the Exchange will publish on its website daily the full list of
features used for real-time calculation and available for inclusion in
the SMI models. The Exchange will also retain each historical iteration
of models employed by the SMI as part of its books and records and make
such information available to the Commission upon request. The Exchange
will also publish a Trader Update in advance of implementing a
retrained version of an SMI model when the Exchange has a reasonable
belief that the retrained version(s) would yield results that differ
materially from the prior version(s).
---------------------------------------------------------------------------
\25\ As discussed above, the Exchange will not seek Commission
approval prior to allowing the models, as part of its retraining
process, to vary the weighting of the features it uses. The Exchange
believes this is appropriate because such variance will only occur
to the extent that it will improve a model's performance with
respect to pre-defined objectives.
---------------------------------------------------------------------------
The Exchange notes that neither the SMI nor the SeMi Order are
designed or intended to further the performance of any participant or
any category of participant over others. The Exchange believes the
models underlying the SMI are objective and designed to avoid bias and
discrimination. Use of the SeMi Order (like use of the DPO) remains
voluntary for all market participants. Accordingly, if any market
participant feels that the SeMi Order does not meet their needs, they
are free to pursue other trading strategies.
B. Self-Regulatory Organization's Statement on Burden on Competition
The Exchange does not believe that the proposed rule change will
impose any burden on competition that is not necessary or appropriate
in furtherance of the purposes of the Act. The Exchange believes that
the proposed change would promote competition by offering market
participants the optional use of an order type designed to protect
against potentially undesirable executions by preventing trading during
periods of market instability as identified by the SMI.
C. Self-Regulatory Organization's Statement on Comments on the Proposed
Rule Change Received From Members, Participants, or Others
No written comments were solicited or received with respect to the
proposed rule change.
III. Date of Effectiveness of the Proposed Rule Change and Timing for
Commission Action
Within 45 days of the date of publication of this notice in the
Federal Register or within such longer period up to 90 days (i) as the
Commission may designate if it finds such longer period to be
appropriate and publishes its reasons for so finding or (ii) as to
which the Exchange consents, the Commission will:
[[Page 106635]]
A. by order approve or disapprove such proposed rule change, or
B. institute proceedings to determine whether the proposed rule
change should be disapproved.
IV. Solicitation of Comments
Interested persons are invited to submit written data, views and
arguments concerning the foregoing, including whether the proposed rule
change is consistent with the Act. Comments may be submitted by any of
the following methods:
Electronic Comments
<bullet> Use the Commission's internet comment form (<a href="https://www.sec.gov/rules/sro.shtml">https://www.sec.gov/rules/sro.shtml</a>); or
<bullet> Send an email to <a href="/cdn-cgi/l/email-protection#99ebecf5fcb4faf6f4f4fcf7edead9eafcfab7fef6ef"><span class="__cf_email__" data-cfemail="2654534a430b45494b4b434852556655434508414950">[email protected]</span></a>. Please include
file number SR-NYSEARCA-2024-112 on the subject line.
Paper Comments
<bullet> Send paper comments in triplicate to Secretary, Securities
and Exchange Commission, 100 F Street NE, Washington, DC 20549-1090.
All submissions should refer to file number SR-NYSEARCA-2024-112. This
file number should be included on the subject line if email is used. To
help the Commission process and review your comments more efficiently,
please use only one method. The Commission will post all comments on
the Commission's internet website (<a href="https://www.sec.gov/rules/sro.shtml">https://www.sec.gov/rules/sro.shtml</a>). Copies of the submission, all subsequent amendments, all
written statements with respect to the proposed rule change that are
filed with the Commission, and all written communications relating to
the proposed rule change between the Commission and any person, other
than those that may be withheld from the public in accordance with the
provisions of 5 U.S.C. 552, will be available for website viewing and
printing in the Commission's Public Reference Room, 100 F Street NE,
Washington, DC 20549, on official business days between the hours of 10
a.m. and 3 p.m. Copies of the filing also will be available for
inspection and copying at the principal office of the Exchange. Do not
include personal identifiable information in submissions; you should
submit only information that you wish to make available publicly. We
may redact in part or withhold entirely from publication submitted
material that is obscene or subject to copyright protection. All
submissions should refer to file number SR-NYSEARCA-2024-112 and should
be submitted on or before January 21, 2025.
For the Commission, by the Division of Trading and Markets,
pursuant to delegated authority.\26\
---------------------------------------------------------------------------
\26\ 17 CFR 200.30-3(a)(12).
---------------------------------------------------------------------------
Vanessa A. Countryman,
Secretary.
[FR Doc. 2024-30916 Filed 12-27-24; 8:45 am]
BILLING CODE 8011-01-P
</pre><script data-cfasync="false" src="/cdn-cgi/scripts/5c5dd728/cloudflare-static/email-decode.min.js"></script></body>
</html>Indexed from Federal Register on December 30, 2024.
This is legal information, not legal advice. Laws vary by jurisdiction and change frequently. Always verify current law with official sources and consult a licensed attorney in your jurisdiction for advice on your specific situation.