A new dawn for Chinese quants
DeepSeek has reignited optimism in China’s quantitative fund industry
The emergence of Chinese artificial intelligence (AI) start-up DeepSeek has taken the world by storm with its cost-effective large language models, challenging Western dominance of the AI sector despite US efforts to curb China’s advancements in this strategic field.
Founder Liang Wenfeng has become a national folk hero overnight, and was invited to meet with President Xi Jinping and other senior government officials in February alongside an exclusive group of top Chinese tech entrepreneurs from firms like Alibaba, BYD, Huawei, Tencent, and Xiaomi.
Before establishing DeepSeek in 2023, Liang co-founded High-Flyer Quant, a leading Hangzhou-based quantitative investment firm, with two engineering classmates from Zhejiang University in 2016. Like many of its peers in China’s quant sector, High-Flyer’s journey has not been without its challenges.
Quant funds began to emerge onshore in China in 2010 following the launch of the Shanghai Shenzhen CSI 300 Futures, a crucial hedging tool for quant trading. This attracted a wave of overseas-trained Chinese quant managers returning home to establish their own firms.
However, when A-shares markets crashed in the summer of 2015, the Shanghai and Shenzhen stock exchanges imposed penalties on 42 trading accounts suspected of market distortion through algorithmic trading.
Regulatory changes
In October that year, the China Securities Regulatory Commission (CSRC) introduced new operational and risk management requirements for quantitative trading in the securities and futures markets to protect small investors, and to mitigate market volatility resulting from high-frequency trading.
This regulatory clarification, coupled with the introduction of Stock Connect schemes established between Shanghai and Hong Kong in 2014 and Shenzhen and Hong Kong in 2016, spurred a surge of domestic and foreign quant managers seeking to enter the market.
Domestic start-ups like High-Flyer, Lingjun, Minghong, and Ubiquant - now recognised as the “Big Four” quants in China – alongside global players such as Two Sigma, Winton Capital, Man Investment, D.E. Shaw, and Millburn Ridgefield, emerged around this time. Competition intensified, leading to improved returns for investors.
As quant funds consistently outperformed traditional mutual funds, substantial capital inflows followed. According to data from CITIC Securities, the industry’s assets under management soared from 62 billion RMB (US$8.5 billion) in 2015 to 1.08 trillion RMB by end of 2021, reaching the historic trillion-RMB mark milestone. By that time, 30 quant firms had surpassed 10 billion RMB in AUM, including High-Flyer.
China downturn
However, China’s quant funds faced a perfect storm of challenges in subsequent years. In 2022 China’s stock market experienced a sharp downturn, dropping over 20% within three months due to economic challenges and global market volatility, wiping out prior gains. Volatile market conditions continued in 2023 and 2024, marked by significant swings with periods of extreme bullishness followed by sharp corrections. In February 2024, a crash in Chinese small-cap stocks triggered a panicked sell-off by quant funds, only for major indices to rebound as government-led funds intervened. Then in September, after a prolonged doldrum, the CSI 300 Index soared 27% following the government’s announcement of its largest post-pandemic stimulus measures, but not all quant funds were able to capitalise on this resurgence.
Traditional quant strategies, such as market neutral, small-cap and micro-cap equity, and leveraged multi-strategy, struggled to perform in these volatile conditions, leading investors to grow wary of their effectiveness.
Meanwhile, the rapid rise of quantitative trading drew increased regulatory scrutiny. CSRC data revealed that programme trading represented close to 30% of the trading value of the A-share market as of April 2024.
In May, the CSRC published new “Administrative Rules for Program Trading in the Securities Market”, imposing stricter limits on algorithmic trading, requiring funds to disclose strategies, lower leverage ratios, and reduce reliance on high-frequency models.
Market volatility
The securities watchdog acknowledged programme trading’s role in enhancing market liquidity and price discovery. However, given that the A-share market is predominantly retail-driven, it warned that activities like high-frequency trading have “significant technological, informational, and speed advantages over small and medium-sized investors, leading to issues such as strategy convergence and trading resonance and increased market volatility”.
These measures, aimed at maintaining market stability, put increased pressure on quant funds relying on high-frequency trading and leverage to generate returns. Several funds saw sharp declines in profitability under the new rules, further eroding investor confidence. The industry’s AUM shrank to 1.13 trillion RMB by the end of 2024, a significant decline from 1.5 trillion RMB in 2022 and 1.92 trillion RMB in 2023.
Leveraging data
Since its inception High-Flyer has leveraged real A-share market data to train AI models for quantitative analysis. In August 2021, it reached over 100 billion RMB in AUM, a remarkable achievement for a relatively young firm.
However, in November 2021, it chose to stop accepting new subscriptions and voluntarily reduced its AUM. The decision was prompted by many of its investment products facing losses of over 10% due to poor timing by its AI models and overcrowded trades from rapid asset growth. A reduction in AUM would allow the firm to focus on overhauling its AI models.
Despite this setback, High-Flyer adapted and recovered. Its AI-enhanced funds rebounded with an average return of 18% in 2023, according to Private Equity Ranking network.
It was against this backdrop that High-Flyer developed DeepSeek as an AI-driven trading tool in 2021 to replace previous models. Unlike traditional quant models that rely on predefined algorithms, DeepSeek’s AI analyses vast amounts of data in real-time, 24/7, adapting to changing market conditions. Its dynamic algorithms can identify patterns and opportunities that human traders might overlook, making it particularly valuable in the volatile Chinese market environment.
DeepSeek is also designed with regulatory compliance in mind. By focusing on lower-frequency, data-driven strategies, it minimises reliance on high-frequency trading and excessive leverage, aligning well with China’s tightened rules. The AI algorithms are transparent and auditable, addressing regulators’ concerns about traditional quant models’ opacity.
High-Flyer reported that DeepSeek-powered funds achieved annualised returns of 15-20% in 2023, significantly outperforming traditional quant strategies during the same period. Their Sharpe ratios were 2.1 compared to the industry average of 1.3 , highlighting their ability to generate consistent returns with lower volatility.
Similarly, Hangzhou Lingjun Investment reported that its quant funds using machine learning and AI yielded 12–15% annualised returns in 2022–2023, compared to 8–10% for traditional quant funds.
In July 2023, High-Flyer decided to spin off part of its team to focus on developing AI products under a new subsidiary called DeepSeek, which has now gained global recognition.
Investors are taking note, allocating increased capital into AI-powered funds. According to CITIC Securities, inflows into such funds from institutions such as pensions and insurers surged 35% in 2023.
The success of DeepSeek and other AI-powered funds has reignited confidence in quant strategies and redefined possibilities. As financial markets grow more complex and data-driven, traditional quant strategies are struggling to keep pace. AI-powered algorithms can process millions of data points in seconds, identifying trends that human analysts would take weeks to uncover. This capability not only facilitates alpha generation but also enhances risk management and transparency.
The mutual fund industry is also racing to adopt AI, developing AI agents to boost research and investment efficiency. DeepSeek’s open-sourced, cost-effective large language models have significantly lowered the barriers for AI applications.
Even retail investors are turning to DeepSeek for its ability to help evaluate companies, pick stocks, and even code trading strategies. This has sparked a surge in demand for education on AI trading.
Embracing innovation
However, the rise of AI does not come without its challenges. Critics warn against an over-reliance on AI, in the case of black swan events such as model failures or unforeseen market conditions. Ethical considerations also arise, such as the potential for AI-driven strategies to exacerbate market inequality or create systemic risks.
Despite these concerns, the transformative potential of AI in finance is undeniable. DeepSeek is just the beginning. The pressing question now is not if AI will transform the industry, but how quickly and profoundly that transformation will unfold. For investors, regulators, and fund managers alike, the message is clear: the future belongs to those who embrace innovation.
*This article was published in Asia Asset Management’s April 2025 magazine under the same title.
