At a Glance
Brad Katsuyama (founder of IEX exchange and protagonist of Flash Boys) engages in an in-depth discussion with Patrick O'Shaughnessy on the impact of high-frequency trading (HFT) on market structure, the founding journey of IEX, and its fundamentally different business model from traditional exchanges. Core thesis: Traditional exchanges (NASDAQ, NYSE) extract rents by deliberately fragmenting markets and selling speed advantages (e.g., co-location, microwave connectivity, tiered data feeds), with revenue primarily derived from data and technology sales rather than trade matching itself — IEX, by contrast, fundamentally reverses this incentive structure through a 350-microsecond speedbump and a zero maker-taker fee model.
Theme 1: The Business Model of Traditional Exchanges — Selling Speed, Not Matching Trades
Brad Katsuyama argues that the core revenue source of traditional exchanges has shifted from trade matching to selling speed advantages and data, a classic form of rent-seeking.
- Revenue Structure Shift: Katsuyama points out that the share of trading revenue in exchange net income has been declining, and market data and technology sales have become the largest revenue source. Taking NASDAQ as an example, its revenue structure is as follows: trading fees of approximately $1.2 billion (but with about $1 billion in rebates, netting roughly $200 million), co-location and market data exceeding $500 million, and listing fees of around $350 million. Exchanges do not create market data — the data is generated by industry participants, and exchanges merely organize and sell it in tiers.
- Tiered Data and Speed Advantages: Exchanges classify market data by speed and detail, and the more one pays, the faster and more detailed the data received. IEX’s own market data bills have risen by 300% since 2014, while technology costs in other industries have continued to decline during the same period. Katsuyama emphasizes: "Exchanges sell advantages, they sell tiered access rights" — for example, a 10G LX cable is 2 microseconds faster than a standard 10G cable, and exchanges charge a premium for this to high-frequency traders (HFTs).
- Fragmentation as a Business Model: NYSE operates 3 exchanges, NASDAQ operates 3, and BATS operates 4 — fragmentation itself is intentional. Katsuyama explains: "If the goal is to bring buyers and sellers together, why separate them? Because fragmentation creates more dependencies — I can sell 4 sets of market data and 4 sets of cable connections." Exchanges maximize industry revenue by increasing friction.
- Arbitrage Between Microwave and Fiber: Exchanges sell microwave connections to HFTs, allowing them to be several microseconds faster than participants using fiber. Katsuyama questions: "Microwave connections create an arbitrage between microwave and fiber — what purpose does this arbitrage serve? None. It simply allows the exchange to sell a faster version, knowing that once one HFT buys it, all HFTs will have to buy it."
Theme 2: IEX’s Solution — Speedbump and Zero Maker-Taker Model
Katsuyama positions IEX as the antithesis of traditional exchanges: eliminating speed advantages through a 350-microsecond speedbump and removing agency conflicts through zero maker-taker fees.
- Speedbump Mechanism: IEX places a delay box equivalent to 38 miles of coiled fiber optic cable in front of its matching engine, imposing a 350-microsecond (approximately 0.35 millisecond) delay on all orders entering and leaving IEX. Katsuyama explains: "The benefit of co-location is putting your server as close as possible to the exchange’s matching engine. By pushing everyone back 38 miles, we achieved the opposite of co-location." This mechanism prevents HFTs from front-running on other exchanges after receiving price signals from IEX.
- Price Protection Logic: The key to the speedbump is that while orders pass through the delay box, IEX has sufficient time to obtain the latest price information from other exchanges, ensuring trades are executed at fair prices. Katsuyama notes: "Other exchanges sell microwave connections to HFTs but use fiber optics to get information from other exchanges themselves — meaning they sell others the ability to know information earlier than they themselves do. How can they price fairly?"
- Zero Maker-Taker Fees: Traditional exchanges pay rebates to "makers" and charge fees to "takers" — the industry pays approximately $2.7 billion in rebates annually. Katsuyama views this as an "agency problem": brokers route client orders to exchanges offering the highest rebates, even if those exchanges have the worst execution quality. "Public data proves that exchanges with the highest rebates not only have the longest queues but also the worst post-execution price reversals — you just bought at $10, and the stock is about to drop to $9. Why do brokers put clients in the longest queues? Because they are paid to."
- IEX’s Revenue Structure: Unlike exchanges, over 90% of IEX’s revenue comes from trading fees (two-way charging), with market data provided free of charge and no connection fees. Katsuyama draws an analogy: "It’s like discovering that an electric company makes more money from your home beyond selling electricity — the core function of an exchange is matching buyers and sellers."
Theme 3: From Discovery to Entrepreneurship — IEX’s Financing and Key Turning Points
Katsuyama details the entrepreneurial journey from RBC’s Thor system to IEX, emphasizing that "experiencing the problem firsthand" is a necessary condition for entrepreneurial success.
- Problem Discovery and the Thor System: In 2007, Katsuyama discovered that orders could not be filled at the displayed screen price — seeing 10,000 shares, he could only buy a small portion, and the stock price would immediately jump. In 2009, after taking over RBC’s electronic trading desk, the engineering team revealed the truth: RBC’s orders arrived at different exchanges with a time difference of up to 2 milliseconds, while HFTs used dedicated networks (476 microseconds) to reach other exchanges ahead of RBC. The Thor system developed by RBC sent orders based on latency differences, ensuring all orders arrived at each exchange simultaneously, boosting the fill rate to 100%.
- The Pivot from Thor to IEX: A buy-side client told Katsuyama: "You’ve only solved 3% of the problem — that’s the share of orders I send to RBC. What about the other 97%?" This prompted him to realize the problem had to be addressed at the exchange level. In 2009, Katsuyama convened a meeting of 10 buy-side firms at the Parker Meridien Hotel in New York — two top buy-side traders with over 25 years of experience met for the first time. The buy-side made it clear: this project could not be done within RBC, because other brokers would not use a competitor’s exchange.
- The Difficult Fundraising Journey: Katsuyama and his team resigned without funding, using $425,000 of their own money to get started. The first legal bill alone was $275,000. An initial attempt to raise $50 million failed, and lowering the target to $25 million still drew no interest — "VCs don’t write $25 million checks, and PE wants to invest larger amounts." It ultimately took nine months to scrape together $9.4 million, including a $1.5 million friends-and-family round (Katsuyama told his father: "You must treat this money as going to zero"). His monthly salary was just $2,000 for a long time.
- Key Turning-Point Meetings: Two meetings changed IEX’s fate. The first was with David Einhorn (Greenlight Capital) — he questioned IEX’s initial focus on a router strategy: "Why do you keep trying to send orders to other exchanges? Why not give us a reason to all come to the same exchange?" This prompted the team to develop the speedbump. Einhorn also refused to invest alone, insisting on more buy-side participation. The second was with Bill Ackman (Pershing Square) — he said, "This has to happen. We build large positions and exit large positions, and we’ve always been looking for ways to prevent others from manipulating our orders." Katsuyama recalls: "Ron and I started hugging in the elevator on the way down — the desperation of having no money and every meeting deciding life or death is beyond words."
Theme 4: The Evolution of the HFT Ecosystem and the Future of IEX
Katsuyama believes the HFT profit pool has shrunk but will expand again when volatility returns; IEX’s next focus is educating buy-side participants and increasing market share.
- Changes in the HFT profit pool: Katsuyama argues that three factors have compressed HFT profits: rising costs (exchanges continuously raising the technological bar), increased regulation (SEC fines and sanctions on dark pools), and declining volatility. "If the market becomes volatile again, the cash registers will ring." He warns that rising costs are driving consolidation in the HFT industry, but the survivors may capture a larger slice of the pie.
- IEX’s market position: As of the interview, IEX held a market share of approximately 2.3% and had been profitable for two years. In August 2017, it recorded a daily trading volume record of $10 billion. More strikingly, of the 50 largest exchange trades that day, 41 occurred on IEX — a result of the aggregation of buy-side and bank proprietary trading desks.
- NYSE’s imitation and failure: After fiercely lobbying against IEX’s speedbump, NYSE launched its own speedbump market, but its market share was only 25 basis points (0.25%) . Katsuyama notes: "You can’t just put up a sign saying 'we have a speedbump too' and expect the buy side to come — we fought alongside the buy side for nearly a decade."
- Future focus: Katsuyama views 2018 as the first year IEX "no longer needs to build specific infrastructure," shifting its focus to: educating the buy side (helping them take more active control of order routing) and expanding the listing business (currently, NYSE and NASDAQ collect approximately $700 million annually in listing fees from listed companies, yet those companies know little about market microstructure).
Mentioned Positions
| Position |
Guest Stance |
Key Data |
| IEX (Itself) |
Bullish (as a solution) |
Market share 2.3%, daily trading volume record $10 billion, 41 of the top 50 trades executed on IEX |
| NYSE |
Risk Warning (business model issues) |
Net trading revenue approximately $200 million ($1.2 billion revenue - $1 billion rebates), data/technology revenue over $500 million, listing fees around $350 million |
| NASDAQ |
Risk Warning (business model issues) |
Same revenue structure as above; Katsuyama believes its model is superior to NYSE (no human factors) |
| BATS |
Risk Warning (driver of fragmentation) |
Operates 4 exchanges |
| Capital Group |
Positive (early supporter) |
First buy-side institution to commit investment |
| Greenlight Capital (David Einhorn) |
Positive (key advice and support) |
Suggested IEX focus on its own exchange rather than the router |
| Pershing Square (Bill Ackman) |
Positive (key support) |
Committed to "help with whatever is needed" |
Judgments Worth Remembering
1. "Exchanges do not create liquidity; investors create liquidity. Exchanges should enable buyers and sellers to meet with minimal friction." (Katsuyama) — Traditional exchanges maximize friction through fragmentation, tiered data, and speed products, while IEX minimizes friction via speedbumps and a zero-rebate model.
2. "Microwave connections create arbitrage between fiber and microwave—what is the purpose of this arbitrage? None. It simply allows exchanges to sell a faster version." (Katsuyama) — Unlike ETF arbitrage (which has a purpose), speed arbitrage is purely a rent-seeking opportunity created by exchanges.
3. "The exchange with the highest rebates not only has the longest queue but also the most severe post-execution price reversal. Why do brokers place clients in the longest queue? Because they are paid to do so." (Katsuyama) — The maker-taker model is inherently a conflict of interest, with public data proving it harms client execution quality.
4. "You must have experienced the problem you aim to solve, or you cannot survive all the unforeseen lows." (Katsuyama) — An entrepreneur's "missionary mindset" stems from firsthand experience of the problem, not secondhand knowledge.
5. "How much money you raise is far less important than who you raise it from." (Katsuyama) — IEX raised capital from buy-side institutions such as Capital Group, Einhorn, and Ackman; these relationships are more valuable than the amount itself.
6. "A CEO has a responsibility to evolve as the company grows—traits that make you excellent with 10 people can become weaknesses at 70, 100, or 1,000 people." (Katsuyama) — Entrepreneurs must remain self-reflective and not let ego hinder their own evolution.
7. "IEX's market data bills have risen 300% since 2014, while technology costs in other industries have continued to decline over the same period." (Katsuyama) — Exchanges hold monopoly pricing power over market data, epitomizing a systemic issue in the industry.
8. "After lobbying against IEX's speedbump, NYSE launched its own speedbump market—but with only 25 basis points of market share." (Katsuyama) — Trust and relationships cannot be built through simple replication; IEX's decade-long partnership with the buy side is its moat.