FT Alphacast
FT Alphacast

Sal Arnuk on high frequency trading

In this edition, Cardiff Garcia talks with Sal Arnuk, co-head of Themis Trading, about high frequency trading and the fragmented trading infrastructure in the US. Among the topics covered are the regulatory and technological reasons for the dramatic changes of the last decade, high frequency traders

Featured Speakers

Financial Times HostSal Arnuk Guest

Topics Discussed

Episode Summary

Executive Summary: Cardiff Garcia interviews Sal Arnuk of Themis Trading on the rise of high-frequency trading (HFT), arguing it has distorted market structure by worsening liquidity in stress, enabling latency arbitrage and information leakage, increasing flash-crash risk, and undermining capital formation. Arnuk blames regulation, technology, and for-profit exchanges for fragmenting markets and incentivizing speed over fair execution.

Main Topics: What high-frequency trading is (Priority: 5/5): Arnuk defines HFT as automated, high-speed trading using different strategies, especially market-making, statistical arbitrage, and more predatory tactics like momentum ignition and order detection. Market fragmentation and regulatory drivers (Priority: 5/5): He argues Reg ATS, decimalization, and Reg NMS, combined with technology and for-profit exchanges, created a fragmented market with many exchanges and dark pools. Liquidity, flash crashes, and stress behavior (Priority: 5/5): Arnuk contends HFT supplies liquidity in calm markets but withdraws or consumes it in periods of stress, worsening flash crashes and mini-flash crashes. Latency arbitrage and data leakage (Priority: 5/5): He describes enriched exchange feeds and co-location as giving HFT firms faster or more detailed information than the public tape, allowing them to trade ahead of institutional orders. Impact on IPOs and capital formation (Priority: 4/5): Arnuk says compressed spreads reduced the economics of brokerage firms supporting small-company IPOs, hurting capital formation and regional broker-dealers. Regulatory responses and reform ideas (Priority: 4/5): He reviews circuit breakers, limit-up/limit-down, stub-quote changes, sponsored access limits, and proposes excess cancellation fees rather than a blanket financial transactions tax. Evidence, critiques, and motivation (Priority: 3/5): Arnuk defends his view against academic studies and the 'who cares' argument, saying many studies are conflicted and that long-term investors still suffer from reduced trust and worse market quality.

Key Arguments: HFT is not one thing; it includes market-making, statistical arbitrage, and predatory strategies that can exploit order flow and predictability. Exchanges have become for-profit data centers that sell speed and information advantages, creating conflicts of interest. Reg ATS, decimalization, and Reg NMS unintentionally accelerated speed, fragmented the market, and weakened the old exchange model. Co-location and enriched data feeds let HFT firms see or infer institutional activity before the public tape, enabling latency arbitrage. In normal times HFT can appear to add liquidity, but in stressed markets it can rapidly withdraw, widen quotes, or turn into liquidity demanders. The flash crash showed how thin books, HFT front-running, internalizers, and delayed public data can create a vacuum of liquidity and extreme price dislocations. Compressed spreads reduced the profitability of regional broker-dealers, contributing to fewer small-company IPOs and weaker capital formation. Broad academic claims that HFT lowers costs may miss hidden costs, conflicted funding, and the effects on larger institutional orders and market confidence. A financial transactions tax would punish all market participants; excess cancellation fees would better target abusive order traffic and bandwidth consumption. HFT may persist, but awareness, consolidation, and alternative market designs could reduce the most damaging behavior.

Data Points: NYSE order execution speed: from about 20 seconds to around 1 second - Average speed of order execution on the New York Stock Exchange over roughly a decade, cited from Andy Haldane's paper. U.S. equity trading volume share by HFT: less than 20% in 2005 to roughly 66%-75% now - Garcia cites estimates showing HFT became the dominant share of U.S. equity market volume. Alternative estimate for HFT share: 50%-75% - Arnuk notes the exact U.S. share is ambiguous, but much higher than a decade earlier. European HFT share: about 35% - HFT penetration in Europe, up from near zero about a decade earlier. NYSE market share in NYSE-listed securities: 80% in 2005 to about 24% now - Illustrates migration of trading away from the primary exchange. LSE market share: two-thirds in June 2008 to less than one-third today - Shows similar fragmentation in the U.K. market. Fastest execution speed: around 10 microseconds - Current fastest trade execution speed mentioned in the discussion. Quote cancellation rate: 90%-95% canceled before reaction - Arnuk says automated orders are often canceled before investors can respond to them. Flash crash timing: May 6 - Date of the flash crash discussed in relation to HFT and market structure. Limit-up/limit-down pause discussion: 5, 15, or 60 seconds (varied proposals) - Arnuk criticizes the short pause windows being considered by exchanges and regulators. IPO underwriters' small-company economics: under $10 million - Intel and Microsoft are cited as IPOs that were under $10 million when brought public. Single-stock circuit breaker threshold: 10% band - Current circuit breaker threshold for the most liquid stocks. VWAP/market impact example: 40 cents - Arnuk argues stepping out of a VWAP algorithm could move a stock by much more than a penny or two. Internalizer retail commission example: $8 per trade - Retail brokers charge low commissions because they monetize order flow through wholesalers/internalizers. Bad order example: 5,000 shares at 18 versus 118 by mistake - Used to explain how a fat-finger error can snowball in a fast, fragmented market. Mini-flash-crash example: Accenture and Sam Adams Beer traded down to a penny - Examples cited as extreme price dislocations during the flash crash.

Pivotal Quotes: "high-frequency traders provide liquidity in a monsoon, and they consume it in a drought." — Cardiff Garcia (attributing Andy Haldane): Used to summarize the claim that HFT behaves differently in calm versus stressed markets. "the market belongs to the owners, not the renters." — Sal Arnuk: Arnuk argues market structure should serve long-term investors and capital formation rather than rent-seeking speed traders. "it was a perfect plan to create this fragmented mess, this broken vase, which is our market structure today." — Sal Arnuk: His strongest description of how regulation, technology, and exchange incentives shaped the current market.

Implications: Listeners are left with a skeptical view of HFT’s benefits: lower displayed spreads may hide worse execution, weaker resilience, and poorer capital formation. The debate points toward stricter data/access rules, better fee structures, and alternative market designs.

🔓 Sign Up for Unlimited Episode Search

About FT Alphacast

Alphachat is the conversational podcast about business and economics produced by the Financial Times in New York. Each week, FT hosts and guests delve into a new theme, with more wonkiness, humour and irreverence than you'll find anywhere else Hosted on Acast. See acast.com/privacy for more information.

View all episodes from FT Alphacast