Is Short-Timescale Algorithmic Trading Good or Bad for the Market?

This is a genuinely contested question, and reasonable people in finance and economics disagree. Here is the honest landscape of arguments.

The Case That It Is Good (Or At Least Net Positive)

Short-timescale algorithmic trading, especially high-frequency market making, has dramatically narrowed bid-ask spreads compared to the pre-electronic era. Retail investors today pay a fraction of what they used to in implicit transaction costs. Markets are also deeper and more continuously liquid — you can usually trade size without waiting for a human counterparty.

Price discovery is faster: information gets incorporated into prices in milliseconds rather than minutes, which arguably means prices are "more correct" more of the time. Arbitrage between related instruments — an ETF and its constituents, for example — keeps prices consistent in ways that benefit everyone holding either side.

The Case That It Is Bad (Or Has Serious Costs)

Much of HFT activity is a zero-sum race for speed. Billions get spent on microwave towers and colocation that do not produce real economic value — they just redistribute who captures the spread. Critics like Michael Lewis (Flash Boys) and economists like Eric Budish argue this is a textbook arms-race waste.

Liquidity provided by algos can be illusory: it evaporates exactly when markets are stressed. The 2010 Flash Crash, the 2015 Treasury flash event, and various ETF dislocations all show the same pattern — the bid disappears at the worst possible moment.

Some strategies are straightforwardly predatory and have led to prosecutions: spoofing, layering, momentum ignition. And there is a more subtle concern that ultra-fast trading taxes slower informed traders — the fundamental investors doing real research — discouraging exactly the kind of analysis that makes prices informative in the first place.

Where There Is More Consensus

Most economists agree that some automation and electronic trading has been clearly beneficial. The disagreement is mostly about the marginal returns to ever-faster speeds (nanosecond competition) and about specific predatory strategies.

Proposals like frequent batch auctions (Budish, Cramton, Shim) try to keep the benefits while killing the speed race — replace continuous trading with very short discrete auctions, so being first by a microsecond stops mattering. That is an active policy debate rather than a settled question.

My Honest Read

The first wave of electronification was almost certainly good for markets. Spreads narrowed, access widened, and the cost of trading for ordinary people collapsed. That part is not really in dispute.

Whether the current speed arms race adds anything beyond the first few milliseconds is much harder to defend on efficiency grounds. The systemic-stability concerns are real but hard to quantify — we know liquidity can vanish under stress, but predicting when and how much is a different problem.

If you forced me to pick: the existence of fast electronic markets is good, and the marginal nanosecond is probably not.