Yes, but only under specific conditions. Automated strategies can respond to price, volume, and liquidity changes that other automated trading helped create. When those responses point the same way and the order book is thin, they can reinforce one another and push prices further than any single order would. Participants on the wrong side of that move can face worse execution prices, slippage, or losses. The effect is not a law of automated markets. The same activity can also supply liquidity and steady prices, and which outcome occurs depends on direction, crowding, and available depth.
The best-documented case concerns foreign-exchange (FX) execution algorithms. The sections below explain how the loop forms, what the evidence shows, where it stops, and which controls regulators expect firms to have in place.
As an Amazon Associate I earn from qualifying purchases.
What “playing against themselves” means here
The phrase is a metaphor. Independent strategies do not share a goal, coordinate, or “fight” one another. Each one reads the same market data and responds according to its own rules. A feedback loop forms when one strategy’s orders change the prices or depth that another strategy is watching, and that second strategy’s reaction then changes what the first one sees.
That is different from a single firm’s system malfunctioning and trading against its own positions. Nothing in the evidence discussed here describes that scenario. It is a question of operational failure, which the control framework later in this article addresses.
#1 Best Overall
How a feedback loop forms
The sequence below is a mechanism that can arise under particular conditions. It is not a description of every trade in every market.
- A market event changes price or turnover. A large order, a news shock, or a sudden sell-off can cause prices to fall and trading volume to jump.
- Reactive algorithms adjust. Execution algorithms that set their pace from observed volume, and quoting systems that move prices, change how much they trade and in which direction.
- Their orders change the order book. The resulting orders shift execution prices and use up resting liquidity.
- Other systems read those changes as signals. They respond in turn, so the original move is extended by new activity.
- Liquidity providers pull back. Under stress, market makers may widen quotes, reduce size, or withdraw. That leaves less depth to absorb the next wave, so the next move is larger.
Each step can be reasonable on its own. The problem arises when the steps stack in the same direction at the same time.
Why the same logic can push in either direction
A selling algorithm during a flash crash
A participation-of-volume (POV) algorithm tries to trade a set share of market turnover. When turnover rises, its pace rises with it. During a sharp sell-off, volume surges, so the algorithm sells faster and adds to the selling pressure that produced the surge. The Bank for International Settlements (BIS) Markets Committee report, published 30 October 2020, uses this POV example to illustrate how a reactive execution strategy can intensify a fall.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →A buying algorithm during a decline
The same design can work the other way. An execution algorithm that is buying into a drop absorbs selling pressure, can support prices, and can help start a rebound. The direction of the order flow and the surrounding market context decide whether a reactive algorithm steadies or destabilises the market. Reactivity is not inherently harmful.
When small actions add up
Each execution algorithm may be designed to limit its own market impact. The BIS report warns that when many similar strategies respond to the same external event, their combined effect can resemble the impact of a much larger order, because the individual designs were built to minimise only their own footprint. Correlated logic is the core of the problem.
The BIS report and related evidence point to several conditions that make feedback risk more likely:
- Outsized orders relative to the depth resting at current prices.
- Crowded trades, where many participants hold similar positions or use similar logic and therefore tend to exit on the same side.
- Thin liquidity, with few resting orders to absorb a push, particularly in stress.
- Fragmented markets, where the same instrument trades across several venues and a move on one can spread to others.
Feedback risk is associated with these conditions. It is not an inevitable property of automated trading, and the evidence does not support a claim that it occurs in every market move.
The other side: liquidity and the observed severity
The same BIS report says execution algorithms can improve matching efficiency and provide liquidity, while also creating self-reinforcing loops that can exacerbate sharp moves. Both effects belong in the picture.
The report also notes that initial observations from the Covid-19 pandemic suggested these risks might not have been as acute as previously believed. That is an early reading from a single period in the FX market. It is not a measured frequency, and it does not show the channel is harmless.
The 2010 Flash Crash as a carefully attributed example
The 6 May 2010 US Flash Crash is often cited in discussions of algorithmic trading. The US Securities and Exchange Commission (SEC) staff analysis of that day treats it as a complex event involving interacting liquidity and trading dynamics. Its review considers that liquidity withdrawal and algorithmic trading could have contributed to feedback effects.
The studies the SEC summarises are generally consistent with the view that high-frequency traders did not cause the crash, although their withdrawal of liquidity may have worsened the declines. The event therefore does not show that bots caused a market crash. It shows how interacting liquidity and trading behaviour can make a decline deeper.
What regulators expect from firms running these strategies
FCA Handbook, MAR 7A.3
The UK Financial Conduct Authority (FCA) Handbook section MAR 7A.3, “Requirements for algorithmic trading,” says firms engaging in algorithmic trading must have effective systems and controls. The section specifically lists:
- system resilience and capacity;
- appropriate trading thresholds and limits;
- prevention of erroneous orders, and of contributing to disorderly markets;
- business-continuity arrangements;
- testing; and
- monitoring.
The rule page shows 1 January 2021 as its last update. Check the live FCA Handbook wording before quoting it as current.
FCA multi-firm review, 21 August 2025
The FCA’s multi-firm review of algorithmic trading controls, published 21 August 2025, says algorithmic trading firms can materially affect price formation and liquidity because of their trading footprint and strategies, and because they link fragmented markets. The FCA stresses that controls and oversight need to keep pace with complexity, speed, and technological change.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Where losses come from, and what safeguards can and cannot do
“At a loss” is one possible outcome of adverse execution, not a universal result. The table maps the main loss channels discussed in the evidence to the controls that address them.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware match| Loss channel | What it looks like | Control lever | Basis in the evidence |
|---|---|---|---|
| Adverse execution | Orders fill at worse prices than the strategy expected | Thresholds and limits on order size and pace; monitoring of execution quality | FCA MAR 7A.3 (thresholds, limits, monitoring) |
| Slippage in fast markets | Price moves between decision and fill, often as turnover spikes | Pace and participation limits; testing of how logic behaves under high turnover | BIS 2020 POV example; FCA MAR 7A.3 (testing) |
| Crowded trades | Many participants exit the same side at once | Monitoring of concentration and similar-logic exposure | BIS 2020 (correlated strategies); FCA multi-firm review, 2025 (monitoring) |
| Thin liquidity under stress | Depth disappears, so the next orders move prices further | Order-size limits relative to available depth; business-continuity arrangements | BIS 2020 (liquidity conditions); FCA MAR 7A.3 (continuity) |
| Trading-system malfunction | Erroneous or unintended orders from a single firm’s own system | Prevention of erroneous orders; resilience and capacity controls | FCA MAR 7A.3 (erroneous orders, resilience) |
These safeguards reduce exposure. They are expectations of effective systems and controls, not a guarantee that a strategy will avoid losses when it trades against other participants’ reactions. The FCA requirements describe what a firm’s framework should contain, while outcomes in a fast, fragmented market depend on conditions no firm fully controls.
What the evidence does and does not establish
- The clearest findings concern FX execution algorithms in a fragmented over-the-counter market, from the BIS report published 30 October 2020. They are not measured rates for equities, futures, crypto, or retail trading bots.
- The Covid-19 observations in the BIS report are early and come from one period, so they do not establish how often feedback loops occur.
- The FCA material is UK supervisory context. The FCA’s 2025 review created no new requirements; its purpose was to help firms comply with existing ones.
- The SEC account of the 2010 Flash Crash is a historical analysis of one US event. It does not provide a general prevalence estimate.
- No loss figure, frequency, or quantified share of trading activity is established by these sources. A reader should not infer one from the mechanism alone.
In short, feedback loops are a conditional risk that interacts with liquidity, crowding, and the direction of order flow. Managing them means treating algorithmic activity as something that can move markets, testing it under stress, and building limits and monitoring that stop a single firm’s logic from feeding a wider sell-off.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
