Latest / How Canadian Markets Work / Episode 7: Alternative Trading Systems and Dark Pools | Canadian Investing

Transcript

Episode 7: Alternative Trading Systems and Dark Pools | Canadian Investing · How Canadian Markets Work

0:00
Welcome to today's Deep Dive. We're taking a
0:03
look at a really comprehensive stack of research
0:06
today, all about the hidden machinery of Canadian
0:09
stock markets. Yeah, it's a fascinating look
0:11
at the underlying plumbing, basically. We're
0:13
covering market fragmentation, the order protection
0:17
rule, and even the complex routing mechanics
0:20
that institutions use. Right. And our mission
0:22
here is to kind of shatter this illusion of how
0:25
buying a stock actually works. We're going to
0:28
explore alternative trading systems or ATSs,
0:32
dig into the mechanics of dark pools, and figure
0:35
out if you should actually be worried about where
0:38
your trades are executing. Exactly. Because the
0:40
source material gives us this brilliant breakdown
0:43
of a system that is, well... it's infinitely
0:45
more complex than it looks on the surface. Oh,
0:47
for sure. We're looking at the tech processing
0:49
billions of dollars, the fail safes for retail
0:52
investors. And like you said, the specific algorithms
0:55
institutions use to move capital completely unnoticed.
0:59
Which I think brings us to a really fundamental
1:01
illusion that we just need to break right away.
1:04
Let's do it. So when you walk down the street
1:06
and see this massive, imposing building with
1:09
grand pillars and a big sign that says bank,
1:11
you trust that your physical money is in that
1:14
physical building, right? Right. It gives you
1:16
that sense of tangible reality. Exactly. And
1:19
when you log into your brokerage account to buy
1:21
a stock on the Toronto Stock Exchange, that TSX,
1:25
you kind of intuitively imagine a digital version
1:28
of that grand building. You assume your order
1:30
marches right up to the TSX server, finds a seller
1:33
on the TSX, and the deal is done. Yeah, exactly.
1:36
But that's not actually what happens, is it?
1:39
No, not anymore. I mean, that intuition made
1:41
perfect sense historically. Decades ago, the
1:43
structure of the market actually mirrored that
1:45
exact physical analogy. It was a fully centralized
1:48
system. Exactly. One stock, one exchange, one
1:51
order book. If a company was listed on the TSX,
1:54
you had to trade it on the TSX. It operated as
1:57
a functional monopoly. Which, I mean, from a
2:00
pure efficiency standpoint, a single centralized
2:03
order book sounds incredibly streamlined, right?
2:06
All the buyers and sellers are meeting in one
2:09
highly visible place. Centralization definitely
2:11
offers simplicity. But as the research points
2:14
out, monopolies inherently lack any incentive
2:16
to reduce costs. Because they have no competition.
2:19
Right. Regulators recognize that having a single
2:22
unchallenged exchange basically gave that entity
2:24
total pricing power over the whole market. They
2:27
could just charge whatever they wanted. Exactly.
2:29
Trading fees, data fees, listing fees. They could
2:32
dictate all of it because participants literally
2:34
had nowhere else to go. Wow. OK. So. How did
2:38
they fix that? To introduce market forces and
2:41
drive down those frictional costs. Regulators
2:44
open the doors to competing marketplaces. In
2:47
Canada, we call these alternative trading systems,
2:49
or ATSs. Okay, so that monolithic structure just
2:52
gets broken up. Instead of one giant building
2:55
processing every trade, we suddenly have this
2:57
distributed network of different venues. Right.
3:00
Places like Omega, Alpha, or Match Now. All of
3:03
them are competing to execute your trade. And
3:05
so the landscape completely changes. Completely.
3:09
Today, a major Canadian bank stock listed on
3:11
the TSX isn't just trading there. It is actively
3:14
trading across several different Canadian venues
3:16
simultaneously all day long. That is wild to
3:20
think about. And it worked. The initial goal
3:23
was achieved. Competition forced these venues
3:25
to lower their transaction fees to attract order
3:27
flow and it tightened the spreads across the
3:30
board. Let me try to translate this to like a
3:32
normal consumer experience. Sure. It's essentially
3:35
like if you built a software program to buy a
3:37
very specific TV. Instead of being forced to
3:41
buy it at one major retailer and, you know, paying
3:44
their proprietary shipping fees. The software
3:46
just scans everywhere. Right. It scans Best Buy,
3:48
Amazon, Walmart, and maybe three regional warehouses
3:51
I've never even heard of, and instantly routes
3:53
my purchase to whoever is offering the absolute
3:56
best price. The underlying logic is identical.
4:00
Competition lowers the barrier to entry and just
4:02
reduces the operational friction of the whole
4:04
thing. Okay, lower fees are obviously a massive
4:06
benefit, but doesn't taking a single stock and
4:10
allowing it to trade in six or seven different
4:12
places at the exact same millisecond create a
4:16
highly chaotic, fragmented environment? That
4:19
is a big question, yeah. Because if my liquidity
4:21
is suddenly dispersed across all these multiple
4:24
venues, I have to imagine it gets much harder
4:27
to actually locate the other side of my trade.
4:29
You are identifying the primary structural trade
4:32
-off that regulators had to accept here, which
4:33
is market fragmentation. Fragmentation, right.
4:36
When you introduce competition, you take the
4:39
available liquidity and you slice it across multiple
4:41
venues, rather than keeping it pooled in one
4:43
deep centralized order book. It's like, OK, think
4:48
of physical infrastructure. If you have a massive
4:51
10 -lane highway, a huge convoy of trucks can
4:54
travel down it pretty easily. But... If you replace
4:58
that one big highway with five separate two -lane
5:01
side streets, the total capacity might technically
5:04
be the exact same, but when that convoy tries
5:07
to move through, you instantly get traffic jams.
5:10
That captures the mechanical problem perfectly.
5:12
And the data in our sources actually points out
5:14
how this dynamic uniquely impacts Canada. Because
5:18
we're a much smaller market than the U .S. Exactly.
5:21
Canada operates with a fraction of the total
5:23
daily volume and way fewer overall participants
5:26
compared to the states. Right. So when you take
5:29
a significantly smaller baseline of liquidity
5:31
and mandate that it gets fragmented across competing
5:34
ATSs, that fragmentation bites much, much harder.
5:39
Because dividing a stock with relatively low
5:41
volume across multiple venues means some of those
5:44
venues might literally only have like a handful
5:47
of shares sitting there at any given second.
5:49
Exactly. The math becomes really problematic
5:52
for smaller companies. I can imagine. It leads
5:54
to a very specific Canadian market structure
5:56
struggle. If you look at a mid -cap US company,
6:00
the spread remains razor thin because the sheer
6:02
volume of algorithmic trading fills in the gaps
6:04
across their venues. But in Canada? In Canada,
6:08
because the liquidity for a small -cap stock
6:10
is chopped up, you might have wide gaps between
6:13
the buyers on ATSA and the sellers on ATSB. Right.
6:17
It makes executing large trades in Canadian small
6:21
cap stocks mechanically difficult without accidentally
6:24
moving the price. And the technological burden
6:26
to navigate this must be absolutely massive.
6:28
I mean, managing trades across all these different
6:30
side streets requires some serious infrastructure.
6:34
It really does. Fragmentation basically shifted
6:36
the cost burden from exchange monopolies down
6:39
to the broker dealers. Because the brokers have
6:41
to figure out how to navigate it all. Right.
6:43
To find you the best price, brokers had to develop
6:46
complex smart order routers or SORs. Smart order
6:51
routers? SORs. These algorithms have to maintain
6:55
constant microsecond connectivity to every single
6:58
ATS. They're paying for colocation, fiber optic
7:02
lines, raw computing power, all just to synthesize
7:05
a consolidated view of the market. Wow. And I
7:08
assume those costs eventually trickle down to
7:11
the client. Yeah, exactly. The cost of building
7:13
and maintaining that routing tech gets factored
7:15
into the broader ecosystem, which offsets a portion
7:18
of the savings generated by the lower fees. Let's
7:21
dig into those smart order routers for a second.
7:24
If my order for, say, 500 shares of a Canadian
7:28
telecom company is flying out of my broker and
7:31
hitting this fragmented web of ATSs, what is
7:34
actually governing that router? How does the
7:36
system mathematically prevent some random low
7:40
-volume ATS from executing my trade at a terrible
7:43
price. The safeguard governing those routers
7:46
is called the Order Protection Rule, or OPR.
7:49
Order Protection Rule. Yeah. When regulators
7:51
fragmented the market, they had to implement
7:54
a technological patch to prevent the exact chaos
7:57
you're describing. Because otherwise it would
7:59
just be the Wild West out there. Exactly. So
8:01
the Order Protection Rule dictates that a broker's
8:04
smart order router cannot execute a trade at
8:07
an inferior price. if a better price is visibly
8:10
available on any other recognized venue. Oh,
8:13
okay. So the router is required by law to clear
8:16
out the best prices first, regardless of where
8:19
they are physically located. Mechanically, the
8:21
router looks at the Canadian best bid and offer,
8:23
the CBBO. Right. Let's say you want to buy 500
8:26
shares. The router sees that venue A is offering
8:29
100 shares at $20, and venue B is offering 400
8:33
shares at $20 .01. Okay. the router is legally
8:37
obligated to ping venue A first, capture those
8:40
100 shares at $20, and only then route the remaining
8:44
400 shares to venue B at 20 .01. And it happens
8:48
in fractions of a millisecond. The technological
8:51
coordination required to actually execute that
8:54
is staggering. It's incredible. I mean, you hit
8:57
the buy button and within a millisecond, an algorithm
9:00
queries a distributed network of exchanges across
9:02
the country, slices your order based on available
9:05
liquidity at each price point, captures the shares,
9:09
and aggregates them back into your account. All
9:11
in the blink of an eye. And because of this order
9:13
protection rule, you are mathematically guaranteed
9:16
to receive a price at least as good as the best
9:19
one displayed anywhere in that network. And it's
9:21
entirely invisible to you. invisible, highly
9:24
regulated, and incredibly effective for the standard
9:26
retail order. It ensures best execution on what
9:29
we call the visible quotes. Now, wait. I want
9:33
to focus on that specific terminology right there.
9:35
OK. The order protection rule guarantees the
9:38
best visible price, which kind of structurally
9:41
implies there is a segment of the market operating
9:44
with prices that are entirely invisible to that
9:46
routing algorithm. You caught that. Yeah. The
9:49
venues we have discussed so far, the TSX and
9:52
the visible ATSs, they operate as lit markets.
9:57
Lit markets. So fully public. Yes. In a lit market,
10:00
the pre -trade information is public. The limit
10:02
order book is broadcast to the world. Any participant
10:05
can see exactly how many shares are bidding and
10:08
asking at every price level. But there's another
10:11
side to this. Right. The ecosystem also contains
10:14
dark venues. I have to say, the nomenclature
10:17
alone, Dark Pools, it carries such a heavy connotation.
10:21
It really does. It immediately suggests this
10:23
environment lacking transparency, which runs
10:26
completely counter to the fundamental premise
10:28
of a fair market. It sounds rigged for insiders.
10:32
The terminology has certainly fueled a lot of
10:34
regulatory and public scrutiny, for sure. But
10:37
let's separate the rhetoric from the actual mechanics.
10:40
OK. What is it mechanically? A dark venue is
10:42
simply an alternative trading system that does
10:45
not publish its order book. The pre -trade intention
10:48
is completely hidden. So nobody sees what you
10:50
want to do. Exactly. You can place an order to
10:52
buy 10 ,000 shares and no one in the broader
10:55
market will see that demand. However, the moment
10:58
that order matches with a seller and executes,
11:01
the trade is immediately reported to the consolidated
11:03
tape. Just like a trade on the TSX. Exactly like
11:06
a trade on the TSX. The post -trade transparency
11:09
is identical. It's just the pre -trade part that's
11:11
hidden. OK, I understand the mechanical difference,
11:15
but from an analytical perspective, why allow
11:18
a segmented opaque market to exist at all? It's
11:21
a fair question. In a purely efficient system,
11:24
shouldn't maximum transparency lead to the most
11:26
accurate price discovery? Allowing massive institutional
11:29
players to hide their intentions feels like an
11:31
informational asymmetry against the retail investor.
11:34
The transparency argument is valid, but we have
11:37
to look at the market impact costs associated
11:39
with extreme size. The source material outlines
11:42
a scenario that perfectly illustrates why institutional
11:44
players actually require an alternative to the
11:47
lit order book. Okay, what's the scenario? Consider
11:51
a major Canadian pension fund that has made a
11:53
portfolio decision to exit a position. That requires
11:57
them to sell two million shares of a specific
11:59
company. Two million shares. That is a massive
12:02
block of equity. It is, especially when you factor
12:05
in the liquidity profile of the underlying asset.
12:08
In this specific example, the company only averages
12:11
about 300 ,000 shares of total trading volume
12:14
per day. Oh, wow. So the fund needs to liquidate
12:17
a position that represents more than six times
12:20
the average daily volume of the entire market
12:22
for that stock. Exactly. Now, apply the rules
12:26
of the lit market to that problem. OK. If the
12:29
pension fund routes a two million share sell
12:31
order to the TSX limit order book, the pre -trade
12:34
transparency dictates that every participant,
12:37
every high frequency trading firm and every algorithmic
12:40
router immediately sees a massive wall of supply
12:43
enter the market. Oh, I see. The mechanics of
12:46
supply and demand would just trigger an immediate
12:48
reaction. Instantly. Any natural buyers would
12:51
instantly cancel their bids and step back knowing
12:54
that much supply is about to completely overwhelm
12:57
the market. Right. The order itself becomes a
13:00
market moving event before a single share is
13:02
even traded. The price plummets simply because
13:05
the intention was broadcast. So by the time they
13:07
actually sell, the price is tanked? By the time
13:09
the pension fund actually begins executing the
13:12
trade, they are selling at severely depressed
13:14
prices. The transparency completely destroyed
13:17
their execution quality. And if we trace the
13:20
capital back to its source, the entity absorbing
13:22
that terrible execution isn't some proprietary
13:25
trading desk, it's the underlying beneficiaries
13:28
of the pension fund. Exactly. It's ordinary people,
13:31
retirees whose capital is being eroded by the
13:34
structural mechanics of market impact. And that
13:36
is the core economic justification for dark pools.
13:39
They provide a mechanism for large institutional
13:42
orders to interact without broadcasting a signal
13:44
that allows everyone else to front -run or game
13:46
their intentions. Okay. That makes sense. But
13:50
how did the institutions mathematically execute
13:52
a trade of that magnitude in the dark? They aren't
13:54
just sending a single two million share block
13:56
into a dark venue and hoping another institution
13:58
happens to be waiting on the other side at that
14:01
exact second. No, definitely not. They rely on
14:04
execution algorithms designed to minimize their
14:06
footprint. How does that work? A really common
14:08
strategy involves using a VWAP algorithm, volume
14:11
weighted average price. The algorithm analyzes
14:14
the historical trading patterns of the stock.
14:16
So it looks at when the stock normally trades.
14:19
Right. It identifies that, for example, 20 %
14:21
of the volume typically trades in the first hour,
14:25
15 % in the last hour, and so on. The algorithm
14:28
then slices the 2 million -chair order into hundreds
14:31
of thousands of micro orders. Oh, wow. And it
14:35
distributes them over several days or even weeks
14:37
to perfectly match that historical volume curve.
14:40
So it's essentially structural camouflage. The
14:43
algorithm is just trying to blend the massive
14:45
institutional flow into the natural background
14:47
noise of the retail market? The routing logic
14:50
is highly sophisticated. The algorithm will send
14:53
small, dynamically sized slices to the lit markets,
14:56
specifically designed to look like standard 100
14:59
-share retail trades. Fascinating. And simultaneously,
15:03
it will route larger iceberg orders, or completely
15:06
hidden orders, into various dark pools. These
15:09
hidden orders just sit passively, waiting to
15:12
intercept opposing institutional flow or retail
15:15
orders that happened across the spread. The entire
15:18
objective is to execute the whole position over
15:20
time without ever alerting the broader market
15:23
algorithms that a massive directional seller
15:25
is present. You've got it. But wait, this introduces
15:28
a secondary structural problem. If the dark pools
15:31
are successfully siphoning off massive amounts
15:34
of volume into these hidden venues, doesn't that
15:36
inherently degrade the quality of the lit market?
15:39
You are hitting on the central tension in market
15:42
structure design right there. The degradation
15:44
of price discovery. Because a lit market is the
15:47
engine of price discovery, right? Exactly. Every
15:50
time a buyer and seller posts visible quotes,
15:52
they are establishing the baseline valuation
15:54
for that asset. They are taking on the risk of
15:57
displaying their hand to the public. And the
16:00
participants in the dark pools are avoiding that
16:01
risk entirely. They are avoiding the risk, but
16:04
they are still utilizing the data. Dark pools
16:06
do not generate their own pricing models. Oh,
16:08
they don't. No, they fundamentally rely on the
16:11
National Best Bid and Offer, or the CBBO in Canada,
16:15
generated by the lit markets to determine the
16:17
execution price of their hidden trades. Okay,
16:20
it's a classic free -rider problem. The dark
16:23
venues are using the visible exchanges as a reference
16:25
thermometer to price their own internal transactions
16:28
while actively withholding the liquidity that
16:30
makes the thermometer accurate in the first place.
16:32
Perfectly said. If a significant percentage of
16:35
total market volume migrates to the dark, the
16:38
lit price is being determined by a smaller and
16:40
smaller subset of actual trading activity. And
16:42
that's bad for everyone. Right. The spreads widen,
16:45
the visible order book thins out, and the reference
16:48
price becomes fragile. Regulators globally have
16:51
really wrestled with how to balance the institutional
16:54
need for minimal market impact against the systemic
16:57
need for robust public price discovery. So how
17:00
does the Canadian regulatory framework address
17:02
that free rider dynamic? Do they just cap the
17:06
amount of volume that can execute in the dark?
17:08
Rather than implementing strict volume caps,
17:11
Canadian regulators address the mechanics of
17:13
the execution itself. They implemented a rule
17:16
requiring meaningful price improvement. Meaningful
17:19
price improvement. What does that mean? Well,
17:22
in many jurisdictions, a dark pool can execute
17:24
a trade by simply matching the best price currently
17:27
available on the lit market. Canada determined
17:30
that was insufficient. So if the lit market bid
17:32
is $50 and the ask is $50 .05, a Canadian dark
17:37
pool cannot simply execute a cross at exactly
17:39
$50 or $50 .05? Exactly. The regulation dictates
17:44
that to bypass the lit order book, the dark execution
17:47
must provide a mathematically better price to
17:50
the participant on the other side of the trade.
17:52
Oh, interesting. If the spread is one cent, the
17:56
dark pool must execute at the half cent mark.
17:59
If the spread is larger, they must step inside
18:01
the spread by a full trading increment, or tick.
18:04
The regulatory logic there is actually fascinating.
18:07
It essentially imposes a structural toll on opacity.
18:11
That's a great way to put it. If an institution
18:13
wants the privilege of hiding their pre -trade
18:16
intentions and avoiding the risk of public price
18:18
discovery, they are legally required to subsidize
18:21
the execution quality of the participant they
18:24
are trading against. It creates an economic friction
18:26
that prevents dark pools from merely skimming
18:29
flow without providing tangible value. By mandating
18:32
price improvement, the regulation ensures that
18:35
retail orders interacting with dark liquidity
18:37
actually receive a superior execution than they
18:41
would have on the public exchange. That is pretty
18:44
clever. It really is. It's widely considered
18:46
a highly effective piece of market structure
18:49
engineering. Synthesizing all of this mechanical
18:52
complexity, the smart order routers navigating
18:55
fragmented ATSs, the order protection rule enforcing
18:58
best execution, and the algorithms optimizing
19:01
for meaningful price improvement in the dark,
19:04
we really have to contextualize what this means
19:07
for you, the listener. Yeah, the distinction
19:09
we need to emphasize here is the difference between
19:11
understanding market mechanics and attempting
19:14
to optimize against them. Because when you pull
19:17
back the curtain on latency arbitrage and algorithmic
19:20
slicing, the initial instinct is to feel, you
19:23
know, structurally disadvantaged. It's natural
19:26
to feel that way. There is this temptation to
19:29
alter your behavior, to try and outmaneuver the
19:31
high frequency systems or to worry about which
19:34
specific ATS your broker is routing your ETF
19:37
purchase through. But the data consistently shows
19:39
that for the long -term allocator, the individual
19:42
purchasing broad index ETFs, dollar cost averaging
19:45
into positions, and operating on a multi -year
19:48
time horizon, this underlying market plumbing
19:51
is entirely benign. That technological friction
19:54
has already been optimized on your behalf. The
19:56
order protection rule mathematically guarantees
19:59
your router is sweeping the best visible prices.
20:01
Exactly. And the Canadian price improvement regulations
20:04
ensure that if your order does interact with
20:07
a dark pool, you are capturing a better price
20:09
than the public spread. The system is designed
20:12
to process your execution efficiently. The engineering
20:15
is incredibly complex, so the user experience
20:18
doesn't have to be. The friction costs for retail
20:21
participants are literally lower than at any
20:23
point in financial history. So don't sweat the
20:26
small stuff. Right. Wasting cognitive bandwidth
20:29
worrying about microsecond routing inefficiencies
20:31
distracts from the actual variables that determine
20:34
long -term outcomes. The variables that actually
20:36
require optimization are your structural fees,
20:40
your asset allocation relative to your duration
20:42
risk, and honestly, your behavioral discipline
20:45
during periods of high market volatility. Spot
20:47
on. The algorithms fighting over fractional pennies
20:50
in the dark pools are just not a threat to a
20:53
properly constructed long -term portfolio. The
20:55
objective of exploring these sources today is
20:57
to provide clarity on the environment you are
20:59
deploying capital into. Understanding the resilience
21:03
and regulation of the execution plumbing should
21:05
provide confidence, not anxiety. Just let the
21:09
institutional routers manage the fragmentation.
21:12
Focus your analytical energy on the long game.
21:14
We have successfully dismantled the physical
21:17
storefront illusion and explored the digital
21:19
infrastructure routing the modern Canadian equity
21:21
market. Thank you for diving into the mechanics
21:24
of fragmentation and price discovery with us
21:26
today. It's been a great deep dive. But before
21:29
we sign off, I want to leave you with one final
21:32
thought regarding those execution algorithms
21:34
we talked about earlier. The camouflage tactics.
21:38
Exactly. We established that institutional algorithms
21:41
deliberately slice massive block trades into
21:44
tiny algorithmic increments. And they dynamically
21:47
route these micro orders to lit venues, specifically
21:50
sizing them to perfectly mimic the standard hundred
21:53
share lots of ordinary retail participants. To
21:56
avoid triggering a market impact. They engineer
21:59
the order flow to look entirely natural. So the
22:02
next time you log into your charting software
22:04
and monitor the intraday tape, and you observe
22:06
a rapid sequence of thousands of small, seemingly
22:09
independent trades printing across the bid and
22:11
the ask, ask yourself something. What's that?
22:14
How much of that visible volume is actually organic
22:17
retail sentiment? And how much of it is just
22:20
a massive pension fund algorithm successfully
22:23
playing dress up? Keep questioning the data you
22:25
are presented with, keep looking beneath the
22:27
surface of the systems around you, and we will
22:30
catch you on the next Deep Dive.