High-Frequency Trading and Your Portfolio: What Long-Term Investors Need to Know

Editor's Note: This article was originally published in October 2011, when high-frequency trading was a newer and less well-understood phenomenon. I have substantially rewritten it to reflect the regulatory landscape that has since developed in Canada, the evidence accumulated over the past decade, and what I have learned about how these firms actually operate through first-hand due diligence. — Arthur Salzer, March 2026

By Arthur Salzer, CFA, CIM — Founder & CEO, Northland Wealth Management Inc.

Picture two office towers on opposite sides of Bay Street. In one, a portfolio manager is reviewing a quarterly earnings report, running valuation models, and working to build a diversified position for a family she advises. In the other, a server rack executes hundreds of thousands of buy and sell orders in the time it takes that manager to draw a single breath.

These are not competitors. They operate on entirely different time horizons and with entirely different objectives. Understanding that distinction is the starting point for thinking clearly about what high-frequency trading actually is, what it means for Canadian markets, and how it affects a long-term portfolio.

I first wrote about HFT in 2011, when it was generating alarm across the investment industry. Since then, the regulatory framework has evolved, the academic evidence has accumulated, and I have had the opportunity to travel to the United States to conduct first-hand due diligence on some of the largest electronic market-making operations in the world. What I came away with was more nuanced than what I had written in 2011 — and more directly relevant to how we actually build portfolios at Northland, where we rely extensively on exchange-traded funds.

What High-Frequency Trading Actually Is

High-frequency trading (HFT) is a category of algorithmic trading in which firms use co-located servers, ultra-low-latency connections, and proprietary order-routing technology to enter and exit positions in microseconds to milliseconds. They are not expressing views on company fundamentals. They are exploiting structural features of how markets process information and match orders.

HFT encompasses several distinct strategies. Electronic market making involves continuously posting buy and sell quotes and collecting the bid-ask spread while managing inventory risk at machine speed. Statistical arbitrage exploits price discrepancies between related securities across multiple venues simultaneously. Latency arbitrage — the most controversial form — uses speed advantages to detect and react to stale quotes before slower participants can update them.

The last strategy is where legitimate grievances are concentrated. When an algorithm can detect a price movement on one exchange and trade against resting orders on another venue before those orders can be cancelled, the person on the other side of that trade is systematically disadvantaged. They are not losing because they were wrong about the fundamentals. They are losing because they were slower by a few microseconds.

What is less commonly understood is that HFT firms are also the dominant liquidity providers for exchange-traded funds. That is not incidental to how ETFs work — it is structural. The tight spreads and reliable intraday pricing that make ETFs so useful as portfolio tools exist primarily because HFT market makers are on the other side of nearly every ETF trade. For investors who use ETFs extensively, HFT is not something happening to someone else's portfolio. It is happening in yours, every trading day.

Seeing It From the Inside

When I first raised concerns about HFT in 2011, I was doing what most commentators were doing: reacting to the theoretical risk and the drama of the May 2010 flash crash. What I had not done was actually look at these operations up close.

The process of conducting investment due diligence eventually took me to the offices of one of the largest electronic market-making firms in North America — a firm that is also a primary liquidity provider for the ETF market globally. What struck me first was the sheer scale of the technology infrastructure: the investment in low-latency networking, the co-location arrangements with exchanges, the teams of engineers maintaining and refining trading systems around the clock. These are not casual operations. They represent billions of dollars of capital deployment in technology infrastructure, pursued with an intensity that most asset managers simply cannot replicate.

What also became clear was the segmentation within HFT itself. The people I spoke with made a genuine argument that electronic market making had tightened spreads in liquid markets — and that argument is supported by the data. For an ETF investor buying a broadly diversified equity fund, the cost of execution today is lower in absolute terms than it was in 2005.

The honest tension, as I came to understand it, sits elsewhere: in what happens to that provided liquidity during genuine stress events. When markets sold off sharply during COVID in March 2020, the electronic market makers who had been reliably providing tight spreads pulled back significantly. Unlike the NYSE specialists they replaced, they have no legal obligation to stay. That asymmetry between fair-weather and stress conditions is real, and it has a specific expression in the ETF market that every investor in these instruments should understand.

The August 2015 ETF Flash Crash: The Clearest Cautionary Tale

The May 6, 2010 flash crash is the event most often cited in discussions of HFT risk. It is a meaningful data point, but it is a general equities story. For investors who hold ETFs, the more instructive episode happened on August 24, 2015.

That morning, U.S. equity markets opened into a volatile pre-market environment. In the first 45 minutes of trading, hundreds of ETFs — including very large, highly liquid ones tracking broad indices — temporarily traded 30 to 50 percent below their net asset values. The underlying holdings had barely moved. The disconnect was almost entirely a market-structure event: HFT market makers, who provide the overwhelming majority of ETF liquidity, had withdrawn their quotes in the pre-market disorder. When the open came, stop-loss orders and market orders from investors hit a market with essentially no bid. Many of those trades were subsequently broken by exchanges.

The mechanism that normally keeps ETF prices tethered to their NAV — the create/redeem arbitrage process in which authorized participants can exchange a basket of underlying securities for ETF shares, or vice versa — could not operate fast enough in those opening minutes to close the gap. The system worked as designed for the rest of the day. But the damage was done for investors whose stop-loss orders had converted to market orders at prices that were largely fictional.

The March 2020 bond market offers a similar, longer-duration lesson. Investment-grade bond ETFs briefly traded at discounts of five to ten percent to their stated NAVs as authorized participants stepped back from the create/redeem mechanism in a market where the underlying bonds were thinly traded and difficult to price. Here the underlying illiquidity was the root cause, and the ETF discount was the symptom. But the mechanism was the same: when the liquidity providers step away, the instrument that appeared highly liquid reveals its dependence on their participation.

I was not caught in either episode, but I was watching carefully. Both events shaped how we think about ETF execution at Northland.

The Canadian Regulatory Picture

Canada's equity market is a multi-venue system. Orders for TSX-listed securities and Canadian-listed ETFs can execute on the Toronto Stock Exchange, TSX Alpha Exchange, Nasdaq Canada, Cboe Canada, and several dark pools. This fragmentation is precisely the environment in which HFT speed advantages have the most value.

Canadian regulators did respond. The Canadian Securities Administrators published National Instrument 23-103 Electronic Trading and Direct Electronic Access to Marketplaces in 2012, establishing a framework for managing automated trading risks. The rules require marketplace participants to maintain pre-trade risk controls, position limits, and supervisory procedures designed to prevent automated systems from destabilizing markets. Since January 2023, the Canadian Investment Regulatory Organization (CIRO, successor to IIROC) has had oversight responsibility and runs active surveillance for manipulative HFT strategies — layering, spoofing, and quote stuffing are prohibited conduct under the Universal Market Integrity Rules (UMIR).

TSX Alpha Exchange offers order types specifically designed to reduce the cost of latency arbitrage for resting limit orders. Canada also produced one of the most visible participants in the post-flash-crash HFT debate: Brad Katsuyama, born in Brantford, Ontario and formerly a senior equities trader at RBC, who co-founded IEX in New York. IEX built a deliberate 350-microsecond delay on incoming orders — a speed bump — specifically to neutralize latency arbitrage. His work, detailed in Michael Lewis's Flash Boys, remains the clearest public explanation of how the adverse selection problem operates in modern markets.

What This Means for an ETF-Based Portfolio

Everything I observed during due diligence on HFT operations, and every market structure episode I have studied since, points toward the same set of practical disciplines for an ETF-using long-term investor.

Never use market orders at the open. This is the single most important rule to come out of August 24, 2015. In normal market conditions, a market order in a liquid ETF fills at a reasonable price. At the open in a volatile session, particularly in the first 30 minutes, spreads can be multiples of their intraday average and HFT market makers may be operating in a reduced capacity. A limit order costs you nothing when the market is functioning normally and protects you when it is not. We use limit orders for ETF execution as a matter of standard practice, not as a response to specific market conditions.

Stop-loss orders that convert to market orders are a particular hazard in ETFs. The August 2015 episode was largely a stop-loss disaster. An investor who had set a stop-loss on a broad equity ETF at, say, 10 percent below its price woke up to find the order had filled at 35 percent below NAV in the opening minutes of trading at a price that was later cancelled by the exchange — but not for everyone. Use stop-limits rather than stop-markets for ETF positions, or reconsider the role of automated stop orders entirely.

Understand the difference between spread and NAV discount in ETFs. The bid-ask spread on an ETF reflects the market maker's cost and risk in the moment. The premium or discount to NAV reflects whether the create/redeem arbitrage mechanism is functioning. In normal markets these two numbers are small and converge quickly. In stressed markets they can diverge sharply. For a UHNW investor making a large ETF purchase, it is worth checking both before executing — particularly in fixed income ETFs, where the underlying holdings are less liquid than the ETF itself.

Avoid concentrated trading at the open and close for large ETF positions. HFT activity is highest at the open (widest spreads, most volatile price discovery) and at the close (index rebalancing, end-of-day positioning). For a patient long-term investor executing a large ETF block, midday execution in normal market conditions typically produces better outcomes. This is not always possible, but when it is, it matters.

Liquidity in the ETF is not the same as liquidity in the underlying. A Canadian equity ETF with $5 billion in assets and $100 million in daily volume is liquid in normal conditions. In a genuine market dislocation, the authorized participants who provide that liquidity by creating and redeeming shares against the underlying basket may step back, and the apparent liquidity may not be available at anything close to NAV. For a long-term buy-and-hold investor this is rarely relevant — it is primarily an issue for investors who might need to sell in a crisis. Knowing which ETFs in your portfolio have this vulnerability is useful, even if you never need to act on it.

The structural advantage still belongs to the long-term investor. The more clearly you understand what HFT firms are actually doing — providing liquidity at machine speed in exchange for the spread, and stepping back when risk spikes — the more clearly you see that their business model is orthogonal to long-term fundamental investing. They care about microseconds. The families we advise care about decades. That is not a disadvantage. It is a structural edge that no amount of co-location can replicate.

How Northland Approaches ETF Execution

As a registered Portfolio Manager, Northland has a best execution obligation that extends beyond finding the best quoted price. It encompasses execution speed, certainty of fill, overall transaction cost, and market impact — all assessed against the specific characteristics of the order, the ETF, and current market conditions.

Our execution policy for ETF positions distinguishes between normal and stressed market conditions, large and small orders, and time-sensitive versus patient trades. For large ETF allocations, we use limit orders as a default, work orders through the trading day rather than concentrating at the open, and monitor NAV premiums and discounts before executing. For rebalancing trades we plan execution to avoid periods of peak HFT activity where possible.

Portfolio construction upstream of execution matters as much as the execution itself. We size ETF positions relative to their underlying liquidity, not just the ETF's own average daily volume. For exposure to less liquid asset classes — small-cap equities, emerging market debt, Canadian corporate bonds — we are particularly attentive to the gap between apparent ETF liquidity and the liquidity of what the ETF actually owns. Alternative investments held as long-term illiquid allocations — private equity, private credit, real assets — are entirely outside this ecosystem.

The Bottom Line

High-frequency trading is not something that happens to other people's portfolios. For any investor who uses ETFs — and most sophisticated long-term investors do — HFT firms are the primary source of the intraday liquidity that makes those instruments function. In normal conditions, that relationship is mutually beneficial: HFT market makers provide tight spreads and reliable pricing, and ETF investors get a cost-effective, liquid exposure to virtually any asset class.

The asymmetry shows up in stress. When markets dislocate, the same firms that provide liquidity in calm conditions can withdraw it, and the ETFs that appeared highly liquid can gap sharply from their NAVs. August 24, 2015 and March 2020 are the clearest documented examples. Neither event required anything more than basic execution discipline to navigate — limit orders, no stop-market positions, attention to the open. But only if you had done the thinking in advance.

What I took away from seeing these operations first-hand was less alarm and more clarity. The machines are fast, well-resourced, and structurally committed to providing liquidity right up until the moment providing liquidity becomes unprofitable. Building a portfolio that accounts for that reality honestly is not difficult. Ignoring it, and discovering it at the worst possible moment, is.

Frequently Asked Questions

About the Author

Arthur Salzer, CFA, CIM is the Founder & CEO of Northland Wealth Management Inc., an independent multi-family office he founded in June 2011. Arthur is responsible for overall firm leadership and investment strategy, and has spent more than three decades advising ultra-high-net-worth families on portfolio construction, alternative investments, and wealth preservation across market cycles. He has written on investment themes for Financial Post Magazine, including the Curve Appeal column from 2016 to 2022, and has conducted first-hand due diligence on investment managers and market participants across North America, Europe, and Asia.