The End of the Speed Arms Race: Why Are HFT Firms Moving Upmarket?

Since the turn of the century, high-frequency trading (HFT) has been the most visible expression of quantitative sophistication in financial markets. Firms competing in the race for minimal latency did so through intense infrastructure spending far beyond the confines of traditional software optimisation. The edge was in hardware architecture and for the winners, turned processing speeds into steady profits.

This business model however has since reached its limits. The evidence is not only visible in academic literature but also suggested by the strategic decisions of the firms involved.

The Saturation of Latency Arbitrage

A very instructive research published in the Quarterly Journal of Economics by Aquilina and Budish estimated the profitability of pure latency arbitrage at 0.42 basis points of the trade notional. A modest return for the trouble but when captured across a significant fraction of the daily traded volume, it becomes a coveted profit margin. The bounty is however concentrated among a small number of winners. The dynamic of this arms race is also self-limiting by design: the proceeds of this "latency tax" are dictated by the daily trading volume and shared amongst HFT firms based on technological performance, making its economics less attractive for newcomers.

Nowadays around 200 firms worldwide compete in the low-latency space. A staggering number for a market where the infrastructure moat still exists but where the incremental return to further investment in processing speed is declining. Firms that have built foundational cashflow from that moat are now actively looking for their next source of durable edge: for most it is quantitative research.

The Firms That Are Already Pivoting

Hudson River Trading is the clearest public example. In this October 2025 episode of Bloomberg's Odd Lots podcast we learn that while the public still views them through a "pre-2020 perception" of pure high-frequency trading, they now consider themselves both a high-frequency and a medium-frequency trading firm. The firm now generates a large pecentage of their yearly revenue from medium-frequency strategies and spends around $1 Billion annually on AI research to support this pivot.

Jane Street's strategic positioning tells a similar story from a different lense. Approximately 70% of their activity is proprietary trading at time horizons measured in minutes or hours rather than milliseconds. Their revenue growth — nearly doubling to $20.5bn in 2024 — has come from this research-driven, longer-horizon approach. The contrast with Virtu Financial, which has remained closer to its pure market-making roots and seen more modest growth, tells an interesting story.

In Europe, Optiver and IMC, the Amsterdam-based market makers, have diversified into asset classes where the competition is less about latency and more about understanding market dynamics like FICC, energy and metals. Flow Traders, which maintained a narrower focus on European ETF market making, has grown as a result more slowly than its peers in recent years.

What This Convergence Signals

The pattern is consistent: the firms with the resources to spare are keenly moving away from pure execution advantage and toward longer-horizon, research-intensive strategies. This is not a reaction to poor performance but a forward-looking repositioning by organisations that understand where durable edge will live in the next decade.

The implication is structural. Pure execution advantage has a ceiling set by physics and infrastructure costs. Research advantage on the other hand should not. A firm that can identify regime shifts faster than its competitors, validate new approaches before the window closes, and deploy with the same rigour that characterised the best HFT operations is competing on a different playing field. One where incumbent advantage is weaker and barriers to entry harder to grasp.

Quantitative research capabilities become the primary competitive metric. The question for any new entrant is not whether this is the right space but what are the prerequisites and the path forward to truly compete within it.

---