Brad Katsuyama - What Happens When You Hit “Buy” - [Invest Like the Best, EP.61]

Episode Summary

Executive Summary: Patrick O’Shaughnessy interviews Brad Katsuyama about discovering latency arbitrage, how exchanges monetize data and speed, and why IEX was built to reduce market friction and protect buy-side execution quality.

Main Topics: Discovery of latency arbitrage (Priority: 5/5): Katsuyama explains how orders stopped filling normally and began getting front-run across exchanges. How exchanges really make money (Priority: 5/5): Exchanges earn more from data, connectivity, and rebates than from matching trades. IEX’s speed bump model (Priority: 5/5): IEX was designed to neutralize speed advantages and improve fair price formation. Buy-side education and adoption (Priority: 4/5): IEX pitches institutions on lower information leakage and better execution quality. Maker-taker and agency conflicts (Priority: 4/5): The conversation criticizes rebates as kickbacks that distort routing and best execution. Entrepreneurship and resilience (Priority: 3/5): Katsuyama reflects on funding risk, conviction, and learning from the problem firsthand. Flash Boys and public scrutiny (Priority: 3/5): Michael Lewis’s book amplified the issue and helped validate the exchange’s mission.

Key Arguments: Orders became unfillable after 2007 because routing time exposed them to latency arbitrage. The market’s real action moved from the NYSE floor to New Jersey data centers. Exchanges monetize market data and technology more than trading itself. IEX’s speed bump removes the incentive to race faster orders ahead of slower ones. Maker-taker fees create conflicts because brokers may keep rebates instead of passing them on. High-frequency trading is not monolithic; market making and predatory arbitrage are different. The buy side benefits most when information leakage is reduced and orders rest in one protected venue.

Data Points: Year the problem became visible: 2007 - Katsuyama says fills changed after Reg NMS-era fragmentation and routing delays. Year of deeper internal discovery: 2009 - RBC electronic sales and trading team helped explain the latency issue. Latency between exchanges: 2 milliseconds - RBC order arrived at BATS first and NYSE last in early experiments. Ronan Ryan’s faster route: 476 microseconds - He described a network path from BATS to NYSE much faster than normal fiber. Speed bump delay: 350 microseconds - IEX’s coiled fiber creates an intentional delay for all incoming and outgoing traffic. Coiled fiber length: 38 miles - The speed bump is described as 38 miles of coiled fiber in a box. Thor product ranking change: from number 19 to number one - Greenwich Associates execution quality ranking after RBC’s solution. Initial self-funding: $425,000 - Founders seeded IEX with their own capital after quitting without funding secured. First round raised: $9.4 million - It took nine months to assemble the first meaningful financing round. Friends and family round: $1.5 million - Used to bridge the startup through its earliest phase. First bill from law firm: $275,000 - An early legal bill nearly exhausted the initial financing plans. Trading done on IEX in August record day: $10 billion - Katsuyama cites this as a major milestone for exchange scale. Largest exchange block trades on that day: 41 of 50 - Most of the day’s 50 largest exchange trades occurred on IEX. NYSE market share at IEX opening: 25 basis points - Katsuyama says NYSE’s copycat speed-bump venue has not gained traction. IEX market share: 2.3% - He notes IEX still has significant room to grow in U.S. equities. NYSE listed-volume share in 2004: 80% - He cites NYSE’s historical dominance in listing volume. NYSE listing volume decline: 75% - He says NYSE’s listing volume fell sharply since 2004. Exchange fees paid annually by companies: $700 million a year - He says companies pay NYSE and NASDAQ roughly this much to list. Rebates paid by major exchanges: $2.7 billion per year - NYSE, NASDAQ, and BATS pay rebates to attract order flow. IEX market data cost increase: 300% since 2014 - He argues exchange data costs have risen despite falling technology costs.

Pivotal Quotes: "you just could not buy or sell what you saw on your screen" — Brad Katsuyama: He describes the moment he realized displayed liquidity was not actually executable. "why are you so focused on sending orders to other exchanges? Why don't you focus on getting orders to your own exchange?" — David Einhorn: A meeting that pushed IEX toward building a true destination venue. "we want you to trust IEX" — Brad Katsuyama: He frames the core buy-side relationship as trust built through execution quality.

Implications: The fight over market structure is still open; investors should scrutinize routing, rebates, and execution quality rather than assume displayed liquidity is neutral.