"High-Frequency Trading: Myths, Realities, and What the Research Says"

Few topics in markets generate more heat per unit of evidence than high-frequency trading. In the popular telling, HFT is either parasitic front-running or the benevolent force that made trading nearly free — depending on the book you read last. The academic literature, now two decades deep, supports a duller and more useful conclusion: HFT is a heterogeneous set of strategies, some of which demonstrably improved markets, some of which impose real costs, all of which are best understood as the industrialization of roles markets have always contained. This article sorts the claims by the strength of their evidence.

First, define the thing

HFT is not a strategy; it is a technology posture — colocation, direct feeds, microsecond decision loops, positions held for seconds to minutes, ending the day near flat — applied to several distinct economic activities:

• Market making: quoting both sides continuously, earning the spread, managing inventory — the electronic descendant of the floor specialist (Menkveld, 2013, documents a modern HFT market maker in exactly these terms).
• Arbitrage: enforcing price consistency between related instruments — index futures versus constituents, ETFs versus baskets, the same asset across venues.
• Directional/event strategies: reacting to public information — data releases, order-flow signals — faster than others.

Conflating these is how most bad HFT arguments begin, because their welfare effects differ sharply.

What the evidence supports

Spreads collapsed, and automation gets substantial credit. The transition to electronic, automated market making coincided with historically dramatic declines in quoted and effective spreads across equities and futures. Hendershott, Jones, and Menkveld (2011) exploited a natural experiment (NYSE autoquote) and found algorithmic trading causally improved liquidity for large-cap stocks. The immediacy an ordinary participant buys today is, by any historical standard, extraordinarily cheap — and HFT market making is a major reason.

Price efficiency at short horizons improved. Brogaard, Hendershott, and Riordan (2014) found HFT trades move prices toward efficient values and impound information faster. Cross-instrument consistency — futures tracking cash, ETFs tracking baskets — is tighter than at any point in market history because someone's business model is correcting deviations within microseconds.

The arms race has a real deadweight cost. Budish, Cramton, and Shim (2015) formalized what practitioners knew: under continuous-time trading, tiny speed advantages capture stale-quote profits, so firms rationally spend enormous sums on speed that produces no social value — and market makers widen spreads to cover being sniped. Their point is subtle and important: this is a market design flaw (continuous limit order books), not a moral failing of participants, and design changes (batch auctions, speed bumps) are the coherent response.

Liquidity provision is genuinely less committed. Modern market makers bear no affirmative obligation to quote through stress. In the May 6, 2010 Flash Crash, Kirilenko, Kyle, Samadi, and Tuzun (2017) found HFTs did not cause the crash but did not stop it either — many withdrew or traded with the pressure as inventories filled. The honest summary: everyday liquidity improved; tail-event liquidity became flightier. Both halves are true simultaneously.

What the evidence does not support

"HFT front-runs your orders." Front-running in the legal sense — trading ahead of a client order you were entrusted with — requires a client relationship HFTs don't have. What actually happens is anticipation: statistically inferring pending flow from public data and positioning ahead of it. That can be costly to predictable traders (as we discuss in our adverse-selection article, legible flow pays a tax), but it is the electronic form of behavior as old as markets — and its remedy is execution discipline, not indignation.

"HFT profits come from a hidden tax on retail." HFT revenues per trade are minuscule and have compressed for years as competition intensified; retail traders in the modern structure typically face the tightest effective spreads of any era. The participants most exposed to HFT costs are large institutional orders and slower professional traders — the ones whose flow is informative and detectable.

"Markets are rigged" (in the strong sense). Documented abuses exist — spoofing prosecutions are now routine, and the CFTC and exchanges actively police manipulative messaging — but the mainstream of HFT is registered, surveilled, and economically ordinary: intermediation and arbitrage at industrial speed.

What this means if you are not an HFT

For the systematic trader operating at human or middleware timescales — seconds, minutes, hours — three practical conclusions:

1. Don't race. Choose games speed doesn't decide. Edges based on horizon, analysis, or structural flows are inaccessible to microsecond competition; edges based on reacting to public events first are already owned.
2. Assume your flow is being read, and price it in. Randomize what can be randomized; avoid metronomic execution; measure your markouts. HFT is your counterparty, and it is very good at pattern recognition.
3. Take the free gift. Tight spreads and instant execution are subsidies to everyone whose strategy doesn't depend on speed. A disciplined operation — controlled order flow, measured costs, infrastructure that never sends an order it didn't mean to — captures the benefits of the modern structure while staying out of its crossfire. That, in one sentence, is the design philosophy GIDEON is built around: precision at sub-second timescales, without pretending to compete at microsecond ones.

The mature view of HFT is neither villain nor hero. It is weather — a structural feature of modern markets. You don't argue with weather; you dress for it.

References

• Hendershott, T., Jones, C. & Menkveld, A. (2011). "Does Algorithmic Trading Improve Liquidity?" Journal of Finance, 66(1).
• Brogaard, J., Hendershott, T. & Riordan, R. (2014). "High-Frequency Trading and Price Discovery." Review of Financial Studies, 27(8).
• Budish, E., Cramton, P. & Shim, J. (2015). "The High-Frequency Trading Arms Race." Quarterly Journal of Economics, 130(4).
• Kirilenko, A., Kyle, A., Samadi, M. & Tuzun, T. (2017). "The Flash Crash: High-Frequency Trading in an Electronic Market." Journal of Finance, 72(3).
• Menkveld, A. (2013). "High Frequency Trading and the New Market Makers." Journal of Financial Markets, 16(4).