07 November 2012

AUTHORs Trading Around the Close

Jeff Bacidore
Managing Director,
Head of Algorithmic Trading ABSTRACT
[email protected]
In this paper, we discuss the issues a trader should consider when trading in and
Ben Polidore around the closing auction. Specifically, we describe the auction mechanisms used
Director, Algorithmic Trading by the exchanges, including the order types offered and pre-auction information that
[email protected] the exchanges publish. We then provide an empirical analysis of trading volume,
auction imbalances, and price reactions to imbalances, and discuss how a trader
Wenjie Xu
Quantitative Analyst, should use these findings to execute orders “optimally”. Our results suggest that
Algorithmic Trading algorithms meant to target the close should focus much of their open market trading
[email protected] to the period prior to the imbalance announcement, as opposed to just prior to the
close itself. We also show that these algorithms should trade differently depending
Cindy Y. Yang on whether the trade is a rebalance trade or a flow trade. As an example, we discuss
Quantitative Analyst, our new ITG Dynamic Close algorithm, which utilizes the research contained in this
Algorithmic Trading
paper to implement rebalance and flow trades more effectively.
[email protected]@itg.com

Introduction
The close of trading is generally the most actively traded period of the day.
contact Institutional traders in particular often trade around the close either because they
are benchmarked to the close (e.g., indexers) or because they are drawn to the
Asia Pacific increased liquidity around this time. Traders have the option of trading in the closing
+852.2846.3500 auctions to ensure they receive the closing price. Alternatively, traders may opt to
trade part of their order in the open market to limit their impact on the closing price
Canada and potentially capture some of the price impact caused by their own auction
+1.416.874.0900 trading. The actual trading strategy a trader should follow depends on both the
EMEA market dynamics (e.g., how his trading impacts prices) as well as on the motivation
+44.20.7670.4000 behind the trade (e.g., implementation of a flow trade).
United States In this paper, we discuss the issues a trader should consider when trading in and
+1.212.588.4000 around the closing auction. Specifically, we describe the auction mechanisms used
by the exchanges, including the order types offered and pre-auction information that
[email protected]
the exchanges publish. We then provide an empirical analysis of trading volume,
www.itg.com
auction imbalances, and price reactions to imbalances, and discuss how a trader
should use these findings to execute orders “optimally”. And lastly, we discuss our
new ITG Dynamic Close algorithm, which utilizes the research contained in this paper
to implement rebalance and flow trades most effectively.
Our results show that prices react to imbalances at the time of the first imbalance
announcement. Traders and algorithms executing large orders in the closing auction
should consider trading a portion of their order in advance of the imbalance
announcement to capture the price impact caused by their own closing auction
orders. We also show that the optimal trading strategy in light of our findings
depends on the objective behind the trade, i.e., rebalance trade versus flow trade.
For both large rebalance and flow orders, the optimal strategy involves executing a
07 November 2012 2

fraction of the order in the closing auction and a fraction in the open market prior to
the initial imbalance announcement. The key strategy difference between rebalance
and flow trades is that rebalance trades will generally continue to trade in the open
market up until the close. Flow trades, on the other hand, will often do little or no
open market trading after the initial imbalance announcement. This finding is in
stark contrast to how most “close algorithms” work, as they typically concentrate
their open market trading to just prior to the close. Our research suggests that
Close Algorithms should concentrate more of their open market trading prior to the
initial imbalance announcement and should trade flow trades differently than
rebalance trades.

NYSE & NASDAQ Close Auction Mechanics

1. A Brief Review
NASDAQ
NASDAQ provides two special order types to participate in the closing auction:
“On-Close” orders, i.e., Market-on-Close (MOC) and Limit-On-Close (LOC), and
“Imbalance-Only” (IO) orders. On-Close orders allow traders to participate in the
closing auction directly, regardless of whether any imbalance exists at the time of
the auction. Imbalance-Only orders, on the other hand, execute only when there is an
opposite side imbalance and serve as a tool for traders to specifically offset
imbalances without amplifying the imbalance when trading on the same side as the
imbalance. MOC/LOC and Imbalance-Only orders can be cancelled and modified until
3:50 PM. After that only Imbalance Only (IO) orders can be submitted, but neither
on-close orders nor IO orders can be canceled or modified.1
From the 3:50 PM cut off until 4 PM, NASDAQ publishes imbalance information, which
is updated every 5 seconds. Imbalance direction, quantity, reference price, near price
and far price are included in the feed.2 As such, traders must decide whether to send
their MOC/LOC orders prior to receiving any of imbalance feed information.
At 4 PM, regular session trading on NASDAQ ends, and the continuous book and the
close book are merged for the final closing cross. The official closing price is
determined by maximizing the number of shares executed.3 Better-priced orders
will be executed at the official closing price. Orders priced at the official closing
price will be executed based on order type and time priority, which implies that LOC
orders priced at the closing price may not fill. Figure 1 summarizes the Nasdaq
auction mechanism.

FIGURE 1:
NASDAQ Auction Mechanics
Imbalance on Close Book every 5 seconds
MOC/LOC, IO entry start

7:00 am 3:50 pm 4:00 pm

MOC/LOC: Entry Not Permitted; No Cancellations Close Crossing
IO: Entry Permitted; No Cancellations
Source: ITG

1
IO orders can be entered anytime until 3:59:59 p.m., but they cannot be cancelled or modified after 3:50:00 except to increase the number
of shares or to increase (decrease) the buy (sell) limit price. See NASDAQ Rule 4709 (a)(2)(3)(4).
2
Far price is the hypothetical auction clearing price on the close book which includes On-Close and Imbalance Only orders. Near price on
the other hand is the hypothetical auction price based upon the liquidity available in both the close book and the continuous book. Current
reference price and Near Price change along with the market.
3
If multiple prices meet this criteria, secondary criteria are to first choose the price that reduces minimizes the imbalance and then
minimize the distance from the 4 PM NASDAQ quote. If the resulting price is outside the benchmarks established by NASDAQ, the cross will
occur at a price within the threshold that best meets these optimization objectives (NASDAQ Rule 4709 (c)(2) (A) (B)(C)(D)).
07 November 2012 3

NYSE
The NYSE closing auction permits submission of MOC and LOC orders, as well as a
special “Closing Offset” (CO) order that can offset any imbalances at the closing
price, similar in spirit to NASDAQ’s IO order. MOC and LOC orders can be submitted
and canceled until 3:45 PM without restriction. After 3:45 PM, MOC/LOC orders will
only be accepted if they are on the opposite side of any Regulatory Imbalance
(discussed below). In the case where there is no Regulatory Imbalance, submissions
of MOC/LOC after 3:45 PM will be rejected. CO orders, however, may be entered on
both sides of the market till 4:00PM, and are not restricted to only offsetting
Regulatory Imbalances.4
Starting at 3:45 PM5, the exchange publishes informational order imbalance
messages every 5 seconds, which provide reference price, paired quantity, imbalance
size, and indicative clearing prices.6 In addition, if a large imbalance exists at 3:45
PM, the exchange is required to publish a specific “Regulatory Imbalance” message.7
This message is important as it governs the entry of offsetting interests after the
cut-off. As noted above, MOC and LOC orders are permitted after 3:45 PM only if they
offset a Regulatory Imbalance. If an imbalance exists but is not large enough to
trigger a Regulatory Imbalance, a trader may not submit an MOC or LOC order after
3:45 PM, even if the imbalance feed shows that an imbalance exists. Once a
Regulatory Imbalance is disseminated, traders can send orders on the opposite side
of the imbalance until 4PM regardless of whether future imbalance messages show
that the imbalance is eventually eliminated or even reversed. For example, if there
were a Regulatory sell imbalance of 50,000 shares, traders could send buy MOC/LOC
orders until 4PM even if the sell imbalance turned into a buy imbalance. It is
interesting to note that while the Regulatory Imbalances provide an opportunity for
traders to offset imbalances, these occur only about 11% of the time.
Floor brokers and DMMs have several advantages over other market participants in
NYSE’s closing auction. From 2 PM until 3:45 PM, floor brokers get a “sneak peek” of
the close book every 15 seconds, including the MOC/LOC interests, imbalances, and
even the CO interest that is not available in any other feed.8 Closing D-quotes used by
Floor brokers can add to an imbalance or create an imbalance anytime until 10
seconds before the market close. Closing D-quote interest from Floor brokers are
included in the imbalance feeds starting at 3:55 PM.9
Unlike that in the automated process in NASDAQ, NYSE’s closing price is manually
set by the DMM so that MOC/LOC orders, floor broker interests, orders in the Display
Book, and the DMMs own interest are paired up.10 Limit and/or LOC orders trading
against the imbalance amount are not guaranteed an execution in the closing
transaction even if the price of such order is the same as the closing price.11
Figure 2 summarizes the NYSE Auction mechanism.

4
vvNYSE Rule 123C(2)(b)(iii)
5
NYSE may publish informational imbalances before 3:45 PM with Floor Official’s approval. NYSE Rule 123C (5)(b)
6
NYSE Rule 123C (4) and (5)
7
NYSE Rule 123C(1)(d) when disparity between MOC and marketable LOC is larger than 50,000 shares, or is significant in relation to the
average daily trading volume in the security with prior approval
8
NYSE Rule 123C(6)(b)
9
http://www.nyse.com/pdfs/fact_sheet_nyse_orders.pdf for more about d-quotes and e-quotes
10
NYSE Rule 123C(7)
11
These orders may be included in full or in part at the discretion of the DMM, based on market conditions and the availability of offsetting
 07 November 2012 4

FIGURE 2:
NYSE Auction Mechanics
MOC/LOC : Entry Not Permitted*; No Cancellations
Informational Imbalance CO: Entry Permitted; No Cancellations
MOC/LOC, CO d-quotes, publication - Not Madatory Order Imbalance on Close Book every 5 seconds
e-quotes entry start

2:00 pm 3:55 pm 3:59:50 pm

7:30 am 3:00 pm 3:45 pm 4:00 pm

Floor Broker’s Last time Close

Regulatory MOC/LOC quotes included for brokers Crossing
Floor Broker’s to enter via
Sneak Peek Imbalance publication in the Imbalance
(One Single Msg) calculations handheld

* Unless offsetting Regulatory Imbalance if exists

Source: ITG

An Empirical Study of the Close

1. Volume Profile around the Close

While most traders are aware that volume generally increases substantially as the
close nears, the fact that there is a “mini spike” around the time the exchanges begin
publishing imbalance information is not as well-known. As shown in Figure 3, volume
increases right after the initial imbalance on both exchanges, i.e., at 3:45 PM on the
NYSE and 3:50 PM on NASDAQ. This is indicative of market participants reacting to
the first imbalance publication. For the NYSE, a second, less pronounced “bump”
occurs at 3:55 PM, when the floor broker interest is included in the imbalance
calculation, suggesting that market participants respond to this incremental
information as well.

FIGURE 3:
Volume Profile 3:30 PM to 4:00 PM
NASDAQ Common Stocks NYSE Common Stocks
1 Minute Volume Profile on Normal Days 1 Minute Volume Profile on Normal Days
3:30 pm - 4:00 pm 3:30 pm - 4:00 pm

3.0% 3.0%
1 Min Bin Volume as % Day Volumes

1 Min Bin Volume as % Day Volumes

2.5% 2.5%
First Imbalance Floor brokers’
message revealed interests included
2.0% 2.0%
First Imbalance
message revealed
1.5% 1.5%

1.0% 1.0%

0.5% 0.5%

0% 0%
3:30
3:32
3:33
3:36
3:38
3:40
3:42
3:44
3:46
3:48
3:50
3:52
3:54
3:56
3:58
3:60

3:30
3:32
3:33
3:36
3:38
3:40
3:42
3:44
3:46
3:48
3:50
3:52
3:54
3:56
3:58
3:60

Time Time
* Sample includes all NYSE- and Nasdaq-listed stocks for all full days in 2011.
Source: ITG
07 November 2012 5

2. Evolution of imbalances over time

Since imbalance information is meant to draw liquidity into the market, the
imbalance situation generally evolves over time, via price adjustments and/or
offsetting liquidity. Recall that the imbalance is calculated relative to the current
reference price, so a change in imbalance not only reflects additional liquidity
submitted to the close book (e.g., offsetting MOC/LOC orders on the NYSE) but also
reflects any changes in the reference price.
To investigate how imbalances change over time, we determined the frequency with
which initial imbalances were reduced, eliminated, or reversed for all NYSE and
NASDAQ common stocks for 2011. The results are given in Tables 2 and 3. For
NASDAQ, initial buy (sell) imbalances are neutralized 46% (49%) of the time, while
only 1% of these initial imbalances actually change direction. Recall that NASDAQ
only allows IO orders to be submitted after the initial imbalance is published, and IO
orders will not affect the reference price formation. A shrinking buy (sell) imbalance,
therefore, indicates that market participants reduce imbalances by pushing regular
session trading prices in the direction of the imbalance, or submitting more
offsetting IO orders at or better than current reference price. 12

TABLE 1:
NASDAQ Imbalance Direction Change

NASDAQ Last Imbalance Direction

First Imbalance Direction

B N O S

B: Buy Imbalnce 53% 46% 0% 1%

N: No Imbalance 8% 84% 0% 7%

O: No Cross Found 1% 0% 99% 0%

S: Sell Imbalance 1% 49% 0% 50%

Source: ITG

TABLE 2:
NYSE Imbalance Direction Change13

NYSE
First Imbalance Direction

B S N

B: Buy Imbalnce 83% 15% 2%

S: Sell Imbalance 14% 84% 3%

N: No Imbalance 9% 10% 81%

Source: ITG

For the NYSE, Table 2 shows that a relatively large number of NYSE imbalances
(approximately 15%) actually flip sign between the initial imbalance and the final
imbalance. To better understand these imbalance “flips”, we created a subsample of
our Regulatory Imbalance data containing only cases where the imbalance actually
flips direction (i.e., from buy to sell or from sell to buy) and track how the imbalance
evolves between 3:45 PM and the close.

Current reference price is bounded by NASDAQ BBO. Thus current reference price moves along with regular session trading price.
12

13
Only Order Imbalances published from 3:45pm to 4pm are included. Informational imbalances happened before 3:45pm are not included.
Regulatory Imbalance message if exists will be the same as order imbalance message at 3:45pm.
07 November 2012 6

Figure 4 shows the imbalance quantity in every message as a percentage of the first
imbalance message. The figure shows that imbalances are eroded gradually as time
goes by, with the most dramatic change happening at 3:55 PM. The median
imbalance as of 3:55 PM (i.e., the 50th percentile) is still in the direction of the
imbalance, but is only 20% of the initial size. This suggests that floor brokers tend to
offset imbalances, though not completely. It also indicates that the informational
imbalances published prior to 3:55 PM provide an incomplete picture of the true
imbalance since floor broker interest has yet to be incorporated.

FIGURE 4:
Imbalance Reversals: Imbalance Quantity Over Time

100%
80%
60%
40%
20%
0%
-20%
-40%
-60%
-80%
-100%
3:45:00
3:45:25
3:45:50
3:46:15
3:46:40
3:47:05
3:47:30
3:47:55
3:48:20
3:48:45
3:49:10
3:49:35
3:50:00
3:50:25
3:50:50
3:51:15
3:51:40
3:52:05
3:52:30
3:52:55
3:53:20
3:53:45
3:54:10
3:54:35
3:55:00
3:55:25
3:55:50
3:56:15
3:56:40
3:57:05
3:57:30
3:57:55
3:58:20
3:58:45
3:59:10
3:59:35
Percentile 10 Percentile 20 Percentile 30 Percentile 40 Percentile 50
Percentile 60 Percentile 70 Percentile 80 Percentile 90

Source: ITG

3. Impact of imbalances on prices

To determine how prices respond to imbalances, we look at returns over two horizons:
1) the period between the initial imbalance and the end of regular trading at 4PM and
2) the period between the end of trading and the closing auction itself. Figure 6
summarizes our results by showing the typical price path from the time of the initial
imbalance publication through the closing auction.
With regard to the first horizon, we find that the initial imbalance message results in
both an immediate price reaction and excess market volume. For example, for an
actively traded NYSE stock with an imbalance equal to 1% of ADV, the price impact is
about 4 bps. This price effect occurs almost immediately and persists throughout
the end of (non-auction) trading. More precisely, prices react upon the first
imbalance information within seconds, and the average return from that point on is
not statistically different from zero on average, even for relatively large imbalances.
This implies that signals based on imbalance information are valuable, but capturing
that alpha requires traders to react immediately.
Turning to the effect of imbalance information on the second period, Figure 5
provides the median 4PM-to-closing returns for different imbalance levels using a
sample of liquid common stocks. Specifically, we bucketed observations by
imbalance (as % of ADV), computed the median return for that subsample (as a
fraction of spread), and plotted each median as a point in Figure 5. For example, the
right most dot in the Nasdaq panel provides the median return (the y-axis) for all
observations with an imbalance equal to roughly 4 percent of ADV (the x-axis).
07 November 2012 7

Figure 5 shows that, even for relatively large imbalances, the median price change is
about 0.5 times spread. This implies that closing prints tend to occur at the 4PM bid
or offer on average regardless of imbalance size. Imbalances therefore predict the
direction of the price change between the last regular session quote and the closing
auction itself, but not the magnitude. This finding together with our earlier finding
that prices react immediately to the initial imbalance announcement suggests that
the market tends to push quotes in the direction of the imbalance to compensate
liquidity providers to such an extent that the closing trade itself executes at the 4PM
quoted prices.
The implication of our findings are profound given that many traders and algorithms
attempt to “beat the close” by trading in the minutes leading up to the close to
capture some of the price impact caused by the trader’s own on-close order. For
example, a trader may send 90% of the order into the closing auction as an MOC, but
trade 10% in the open market to capture the impact caused by their own MOC order.
Our results suggest that this strategy is not profitable since the price reaction to any
imbalance typically occurs at the time of the initial imbalance message. The only
benefit of trading into the close itself would be to avoid the (implicit) half-spread that
those in the auction typically pay when they are on the same side of the imbalance.

FIGURE 5:
Median return between 4:00 PM mid-quote and close price by imbalance size
NASDAQ Day Volume > 4 M shares NYSE Day Volume > 4 M shares
1.0

1.0
0.8

0.8
Price Change % Spread

Price Change % Spread

0.6

0.6
0.4

0.4
0.2

0.2
0.0

0.0

0 1 2 3 4 5 0 1 2 3 4 5
Initial Imbalance Size % Day Volume Initial Imbalance Size % Day Volume

Source: ITG

FIGURE 6:
Price Trajectory Illustrations

P Buy Imbalance Scenario

Immediate effect
triggered by Close Market adjusted return
Book initial Midquote @4:00 pm
imbalance vs. Closing price
No significant relationship
with Initial Imbalance
Proportion to
Imbalance size ~Half spread
liquidity
provision
cost

1st Imbalance Price impact 4:00 pm Closing price t

message dissemination stabilized trading ends announced
NYSE 3:45 pm seconds after
NASDAQ 3:50 pm

Source: ITG
07 November 2012 8

Trading Strategy Implications: Rebalance Trades vs. Flow Trades

Institutional trades done around the close typically fall into two categories:
rebalance trades and flow trades. Rebalance trades are those where a portfolio
manager is using existing positions to finance a trade. Flow trades, on the other
hand, are those where the trade is being done to invest inflows of cash from
investors or to generate cash to fund a redemption. An investor redeeming mutual
fund shares, for example, would receive the closing price on the date of the
redemption, as opposed to at the actual price received when the fund sold shares
to raise cash.

1 Rebalance trades
For rebalance trades, traders should aim to minimize implementation shortfall
costs, as these are consistent with “buying low and selling high” and are therefore
consistent with fund return maximization. To see why, suppose instead that
traders tried to minimize slippage relative to the closing price rather than to a
pre-trade benchmark. In this case, submitting large quantities to the closing
auction will, by design, have zero slippage, as the execution price will always equal
the closing price benchmark. But as our research documents, trading large
quantities at the close will tend to move prices adversely (following the imbalance
publication) and will lead to implicit spread costs in the auction itself. Since these
impacts are generally liquidity-induced and are therefore likely to revert, the fund
basically buys high and sells low.14
For rebalance trades, the goal of minimizing implementation shortfall involves
balancing the impact of trading in the closing auction against the impact of trading
in the open market. If the closing auction were sufficiently more liquid than the
open market, a trader should execute entirely in the closing auction. If, on the
other hand, the order is sufficiently large that it starts to have a significantly large
impact on the closing auction price, the trader can reduce trading costs by trading
a portion of his order in the open market.
Our research suggests that, for sufficiently small order sizes, the closing auction is
indeed cheaper than trading in the open market on an implementation shortfall
basis. But as order size increases, a trader could reduce implementation shortfall
by executing some of the order in the open market. Interestingly, for the open
market trading, our empirical results suggest that a trader should trade
disproportionately more before the imbalance announcement than just prior to the
close. Trading just prior to the close itself doesn’t really reduce costs since the
impact of any imbalance actually occurs well before the auction, specifically at the
time the initial imbalance is released.
Figure 7 shows a hypothetical “optimal” trade schedule for a relatively large
rebalance trade.15 In this example, the trader executes the majority of the order in
the close and the remainder in the open market. For the open market trading, the
trader executes a disproportionate amount prior to 3:45 PM, before the market
responds to the 3:45 PM imbalance publication. The front-loading captures some
of the price impact caused by the trader’s own closing trade. The trader will
continue to trade in the open market even after 3:45 PM, but this is done to reduce
total market impact.

14
If trades are completely “informationless”, post-trade prices will revert to pre-trade prices once the market fully absorbs the order. In
the case where trades have permanent price impact or have alpha, post-trade prices on average will not equal pre-trade prices. But
even in these cases, it can be shown that in usual circumstances, minimizing shortfall is equivalent to maximizing fund performance.
15
We assume that the objective here is to minimize expected shortfall without regard to risk. If the trader were risk averse, the optimal
trading strategy would involve trading more in the closing auction, as well as back-loading more of the open market trading around the
4PM close.
07 November 2012 9

FIGURE 7:
“Optimal” NYSE rebalance trade: An illustration
BUY 5mm Rebalance
Target Quantity Cumulative Progress

90%

8%

Cumulative Progress
Target Quantity
6% 60%

4%
30%
2%

0% 0%
3:30

3:32

3:34

3:36

3:36

3:38

3:40

3:42

3:46

3:48

3:50

3:52

3:54

3:56

3:58

4:00
Time
Source: ITG

2 Optimal strategy for flow trades

For flow trades, the economics are different. Specifically, when a fund has a flow that
is priced relative to the closing price, the end investor receives the closing price
regardless of how the cash was actually raised.16 Any slippage from the actual
closing price incurred when selling a position to fund a flow will ultimately be borne
by the remaining fund investors, not by the redeeming investor. For example, if the
NAV at the close of trading were $20, but the net proceeds of the trade were only
$19.99, the net loss of $0.01 would be borne by the fund, since the investor will get
$20 regardless of the execution price. For flow trades, then, the objective should be
to get the closing price or better in order to avoid losses to the fund. Or put another
way, the trader should execute the trade in the closing auction, unless the trader can
somehow beat the closing price on average.
For flow trades, the optimal strategy looks somewhat similar to that of rebalance
trades, but with some important differences. For smaller orders, a flow trader should
execute all of his order in the closing auction since the order is unlikely to create
sufficiently large “capturable” impact to offset the open market trading costs. For
sufficiently large orders, though, a trader can benefit by trading some portion of the
order in the open market. But as with rebalance trades, capturing the price impact of
any imbalance requires trading prior to the imbalance announcement, not prior to
the close itself. In fact, trading between the initial imbalance announcement and the
closing auction is not a particularly good time to execute flow trades since the
impact of one’s own closing auction quantity is felt prior to this, so the trader is
actually paying the post-impact cost during this period.
To see why, suppose a trader puts 90% of a relatively large NYSE order into the
closing auction at 3:30 PM, prior to the initial imbalance publication. If the trader
executes the remaining 10% in the opening market prior to 3:45 PM, she is able to
execute prior to the adverse price movement that will occur when her demand is
revealed to the market at 3:45 PM as part of the imbalance message. But if she
waits until after 3:45PM to trade the 10% residual, it would be too late to capture the
price impact since the price reaction typically occurs just after the 3:45 PM
announcement.

16
More formally, the investor receives the net asset value (NAV) of the fund as of the close of trading, which is simply a weighted average of
each position’s closing price. A fund need not trade out of each individual position pro rata to fund a flow. But intuitively, if a trader funds
a flow by executing trades away from their 4PM values, any losses incurred relative to these 4PM values are incurred by the fund, not the
investor who is cashing out.
07 November 2012 10

Figure 8 shows a hypothetical optimal trade trajectory for a relatively large flow trade.
The trader here will execute most of the trade in the closing auction, which will have
zero slippage relative to the close since it is guaranteed the closing price. But since
that on-close order will lead to a price reaction when revealed as part of the 3:45 PM
imbalance announcement, the trader should execute a disproportionate amount of
the remaining 10% prior to the initial imbalance publication. Note that, unlike with an
identical rebalance trade, a trader would actually trade nothing between the 3:45 PM
imbalance publication and the 4PM close. Intuitively, a trader has no upside from
trading between 3:45 and 4 PM because the price impact of any imbalance has
already occurred. Trading during this period would only result in increased spread
costs and market impact. A trader therefore would prefer to simply trade in the
closing auction instead since he is assured of a zero slippage execution relative to
the closing price.

FIGURE 8:
“Optimal” NYSE flow trade: An illustration
BUY 5mm Rebalance
Target Quantity Cumulative Progress

90%

8%

Cumulative Progress
Target Quantity

6% 60%

4%
30%
2%

0% 0%
3:30

3:32

3:34

3:36

3:36

3:38

3:40

3:42

3:46

3:48

3:50

3:52

3:54

3:56

3:58

4:00
Time
Source: ITG

ITG Dynamic Close Algorithm

The new ITG Dynamic Close algorithm incorporates this research directly by taking
into consideration the fact that 1) imbalance impact tends to occur immediately
after the announcement of the imbalance and 2) the objective for rebalance trades is
different than flow trades. As such, the ITG Dynamic Close Algorithm takes a
dramatically different approach relative to other “Into the Close” algorithms. The
algorithm has two modes, Rebalance and Flow, to accommodate the two different
objectives traders typically have when trading around the close. The algorithm
automatically adapts its strategy to either reduce implementation shortfall costs
(Rebalance mode) or to outperform the closing price (Flow mode). The algorithm
can also offset Regulatory Imbalances, though it may opt to continue trading in the
open market if it thinks it is advantageous to do so.
07 November 2012 11

Conclusion
A deep understanding of the closing auction mechanics and market dynamics are
essential for traders to implement trades efficiently around the closing auction. This
paper reviews auction mechanics in NYSE and NASDAQ, discusses empirical patterns
around the close, and provides suggestions on how to use this information most
effectively when trading. Specifically, our results indicate that a trader wishing to
“beat the close” needs to consider trading before the announcement of imbalances,
as opposed to just prior to the closing auction. Our analysis also suggests that
traders and algorithms that target the close of trading should trade differently
depending on whether they are implementing rebalance trades or flow trades. By
taking into account the price dynamics around the close and tailoring their trading
strategy to fit the motivation behind the trade, traders who execute around the close
can improve their performance significantly relative to the more conventional
approach of trading into the close.

References
NYSE Regulation Rule 123C .The Closing Procedures
http://rules.nyse.com/NYSETools/PlatformViewer.asp?selectednode=chp_1_3_8_13&
manual=%2Fnyse%2Frules%2Fnyse-rules%2F
NASDAQ Rule 4709 Closing Cross
http://www.gpo.gov/fdsys/pkg/FR-2004-03-18/html/04-6068.htm
The authors would like to thank Bob Airo of the NYSE, Jamie Selway, and Laura Tuttle,
for their comments and suggestion.

© 2012 Investment Technology Group, Inc. All rights reserved. Not to be reproduced or retransmitted without permission. 110712-19961

These materials are for informational purposes only and are not intended to be used for trading or investment purposes or as an offer to sell
or the solicitation of an offer to buy any security or financial product. The information contained herein has been taken from trade and
statistical services and other sources we deem reliable but we do not represent that such information is accurate or complete and it should
not be relied upon as such. No guarantee or warranty is made as to the reasonableness of the assumptions or the accuracy of the models or
market data used by ITG. These materials do not provide any form of advice (investment, tax or legal). All trademarks, service marks, and
trade names not owned by ITG are the property of their respective owners. The positions taken in this document reflect the judgement of the
individual author(s) and are not necessarily those of ITG.

Broker-dealer products and services are offered by: in the U.S., ITG Inc., member FINRA, SIPC; in Canada, ITG Canada Corp., member Canadian
Investor Protection Fund (“CIPF”) and Investment Industry Regulatory Organization of Canada (“IIROC”); in Europe, Investment Technology
Group Limited, registered in Ireland No. 283940 (“ITGL”) and/or Investment Technology Group Europe Limited, registered in Ireland No. 283939
(“ITGEL”) (the registered office of ITGL and ITGEL is First Floor, Block A, Georges Quay, Dublin 2, Ireland and ITGL is a member of the London
Stock Exchange, Euronext and Deutsche Börse). ITGL and ITGEL are authorised and regulated by the Central Bank of Ireland; in Asia, ITG Hong
Kong Limited, licensed with the SFC (License No. AHD810), ITG Singapore Pte Limited (CMS Licence No. 100138-1), and ITG Australia Limited
(AFS License No. 219582). All of the above entities are subsidiaries of Investment Technology Group, Inc. MATCH NowSM is a product offering
of TriAct Canada Marketplace LP (“TriAct”), member CIPF and IIROC. TriAct is a wholly owned subsidiary of ITG Canada Corp.