One year after the new regulations on quantitative trading were implemented, how are European and American markets regulating high-frequency quantitative trading?

Ask AI · How do U.S. and European regulators prevent algorithmic trading from misleading the market?

On April 3, 2025, the three major exchanges issued the “Implementation Rules for the Management of Programmed Trading,” and they took formal effect on July 7 of last year. By April 3, 2026, the “Implementation Rules” issued by the Shanghai, Shenzhen, and Beijing exchanges have reached their one-year term.

In the “Implementation Rules,” if an investor’s trading behavior on a single account reaches more than 300 submissions and cancellations per second, or reaches more than 20k submissions and cancellations per day, it is identified as high-frequency trading. At the same time, it further breaks down abnormal trading behaviors, implements differential regulation, and especially adopts differential fee arrangements to increase transaction costs.

One year after the launch of the new rules for algorithmic trading, has high-frequency algorithmic trading been effectively brought under control?

Among them, there is a set of data that we should pay attention to.

Based on public data, as of this first quarter, China’s “billion-yuan” algorithmic trading camp has expanded to 61 firms, with overall assets under management nearing 2 trillion yuan. In terms of management scale, the current management scale is close to 2 trillion yuan, but it has grown by nearly 400 billion yuan compared with the end of last quarter of last year, and has surged by about 800 billion yuan compared with the same period last year.

Judging by the investment scale of leading algorithmic trading firms, multiple top firms’ investment scales have already moved into the tier above 80 billion yuan, only one step away from the trillion-yuan investment scale.

On the one hand, the investment management scale of leading algorithmic trading institutions keeps getting larger. On the other hand, the pace of survival of the fittest among smaller and mid-sized algorithmic trading institutions has clearly accelerated, and the industry shows a pronounced Matthew effect.

Leading algorithmic trading institutions continue to obtain excess investment returns, using methods such as AI and big data to gain sustained investment advantages in the stock market, thereby increasing the difficulty for retail investors to make money. This is also what retail investors have felt most deeply in recent years.

For the stock market, at its core it is a place for wealth redistribution. If an institution in the stock market earns an excess rate of return, that would imply that some investors are losing money in the market.

China’s A-share market’s high-frequency algorithmic trading is highly controversial. The focus of investors is both the recognition criteria for high-frequency trading itself and the severity of penalties for high-frequency abnormal trading and non-compliant trading.

Under the impact of the new rules for algorithmic trading, if the maximum number of submissions and cancellations per second on a single account reaches 300 or more, it is recognized as high-frequency trading. However, for trading behaviors where the maximum number of submissions and cancellations per second on a single account reaches 299, how are their trading behaviors to be recognized? This is undoubtedly a key question the market cares about.

Algorithmic trading has a development history in the A-share market of only around 10 years, and the point when it truly experienced explosive growth was also around after 2020.

For the U.S. and European stock markets, algorithmic trading has been present for decades. Moreover, algorithmic trading currently contributes more than 80% of the liquidity to the local market, so its impact on the local market cannot be underestimated.

For the U.S. and European equity markets, how are high-frequency algorithmic trading activities regulated?

Taking the U.S. stock market as an example, regulation of algorithmic trading mainly reflects whether there is a suspicion that algorithmic trading behavior misleads the market.

In specific operations, common violations by high-frequency algorithmic trading include spoofing, layering to induce, front-running, market manipulation, and so on. Algorithmic trading institutions use algorithms, customer order information, and multiple layers of orders to achieve arbitrage. Any time there is an issue of induced trading or market misdirection, they will face heavy penalties.

In terms of the severity of penalties, light cases involve massive fines and forfeiture of illegal gains; severe cases involve permanent bans from the industry and criminal penalties, including jail sentences for relevant responsible parties.

In the European market, regarding behaviors associated with high-frequency trading, regulators focus more on abnormal trading behaviors, price manipulation behaviors, and behaviors that provide false liquidity, among others.

From the European market’s regulatory actions toward high-frequency algorithmic trading, the costs of ineffective high-frequency trading are increased significantly. Once the conduct touches on false trading, price manipulation, and similar behaviors, it will face risks such as huge penalties and criminal accountability. For algorithmic trading institutions, they must be especially cautious about any trading behavior.

Overall, when analyzing the U.S. and European stock markets, the regulatory focus on high-frequency algorithmic trading is not overly on “high-frequency” itself, but on whether high-frequency trading causes market misdirection.

At present, algorithmic trading provides a large amount of liquidity to the U.S. and European stock markets, but the market is more concerned about the authenticity of that liquidity. If an algorithmic trading institution provides fake liquidity to the market through false trades, induced trades, or similar actions—thereby constituting harm from misleading the market—then such trading behavior will face extremely severe penalties.

Compared with the A-share market, in terms of trading rules, the U.S. and European stock markets appear more comprehensive.

For example, the U.S. and European stock markets do not have daily price limit-up/limit-down rules. They open T+0 trading to the entire market, giving both individual and institutional investors multiple opportunities to correct errors. In addition, for retail investors, risk-hedging tools are more diversified, meeting investors’ needs for intraday risk hedging.

Against this backdrop, even if algorithmic trading institutions use algorithms and AI to gain a better competitive edge, the fact remains that with T+0 open to the entire market and the provision of robust risk-hedging tools, their trading costs are undoubtedly increased substantially.

Most importantly, institutionalization in the U.S. and European markets reaches 80% or more. The A-share market is still led by retail investors. This means that algorithmic trading institutions can use AI and algorithms to simulate retail investors’ trading behaviors, becoming the counterparty to retail investors’ trades. Retail investors cannot compete with algorithmic trading that has powerful technical tools, so they naturally become the weaker side in the stock market.

Algorithmic trading, in the major global stock markets, mainly plays the role of providing market liquidity and acting as a market maker. However, in the A-share market, algorithmic trading is particularly good at capturing the trading behavior of retail investors, and is even more likely to treat retail investors as the counterparty. To a certain extent, the main impact of domestic algorithmic trading is not only to provide liquidity to the market, but also to achieve arbitrage by capturing retail investors’ trading behaviors and investment sentiment. This is also the main reason why algorithmic trading has continued to make money in recent years, while retail investors cannot.

How should high-frequency algorithmic trading behavior be regulated?

For China’s A-share market, regulating algorithmic trading behavior primarily hinges on significantly increasing the difficulty of capturing retail investors’ trading behavior and investment sentiment. The most direct approach is to open T+0 trading for the entire market, while also providing retail investors with more risk-hedging tools.

In practical terms, to prevent overly speculative behavior in the market, you can impose appropriate limits on the number of T+0 trades. For example, a retail investor’s number of intraday trades should not exceed 2 times. In addition, for retail investors, they need to pass professional testing, and only if the test result is 80 points or higher can they open access to more risk-hedging tools.

Only by making it significantly harder for algorithmic institutions to capture retail investors’ trading behavior and investment sentiment can we maintain the trading fairness of retail investors in the stock market. The phenomenon of “algorithmic trading continues to earn excess returns, while retail investors can’t make money” can then be fundamentally changed.

Author’s statement: personal views, for reference only