High-Flyer is a Chinese investment-management business founded in 2015 and based in Hangzhou, China. It specializes in quantitative investing, using mathematical models, large datasets, and computer systems to guide securities trading. Co-founded by Liang Wenfeng, it became one of China’s prominent quantitative fund managers. It is also known for financing and providing the technological foundations for DeepSeek, the artificial-intelligence company established in 2023.
Business and investment approach
High-Flyer manages privately offered investment funds rather than publicly offered mutual funds. Although English-language reporting often describes it as a hedge fund, its operations sit within China’s domestic private securities-fund industry. The High-Flyer name is used for a business encompassing related investment-management and technology operations, rather than a single investment product.
Its core approach replaces much discretionary security selection with systematic analysis. Quantitative investment systems process market information, identify statistical relationships, construct portfolios, and execute trades according to model-based rules. High-Flyer has placed particular emphasis on machine learning, which allows models to learn patterns from data rather than relying exclusively on relationships explicitly specified by researchers.
Such methods do not eliminate investment uncertainty. Historical patterns may disappear, competing investors may identify similar opportunities, and trading expenses may reduce apparent returns. Backtesting can assess how a strategy would have behaved on historical data, but it cannot establish how that strategy will perform in future markets. Portfolio outcomes also depend on liquidity, execution conditions, and transaction costs.
Development and computing infrastructure
High-Flyer’s development coincided with the expansion of quantitative investing in China’s securities markets. Its investment research increasingly incorporated artificial intelligence, making computing infrastructure an important part of its business alongside financial data and investment expertise.
In 2019, High-Flyer established a dedicated AI operation to support research into AI algorithms and their applications. It invested in computing clusters built around graphics processing units, hardware suited to the parallel calculations used in modern machine learning.
This infrastructure had uses extending beyond financial prediction. Training large neural networks requires substantial computing capacity, specialized engineering, and systems for handling large datasets. High-Flyer’s investment in those capabilities helped create the technical base from which DeepSeek emerged.
The connection does not mean that financial forecasting and general-purpose AI are the same activity. Investment models are evaluated partly by trading outcomes under changing market conditions; language models are trained and assessed for tasks involving language, reasoning, coding, and related capabilities.
Relationship with DeepSeek
DeepSeek was established in 2023 under Liang Wenfeng’s leadership, with funding from High-Flyer. It pursued general-purpose AI research and subsequently became known for its large language models.
The relationship between the two businesses is central to High-Flyer’s wider prominence. Capital accumulated through investment management, together with existing computing resources and technical experience, supported an AI research organization outside the conventional model of a start-up initially financed by external venture-capital investors.
High-Flyer and DeepSeek nevertheless refer to distinct activities. High-Flyer is associated principally with quantitative investment management, while DeepSeek develops AI models and related services. DeepSeek’s model capabilities or adoption figures are not measures of High-Flyer’s fund performance.
Disclosure and interpretation
As a privately held investment business, High-Flyer does not publish the same standardized public disclosures required of a listed company. Reported assets under management therefore need to be read with attention to their date, scope, and the entities included. Assets managed for clients are also different from the firm’s own capital or the personal wealth of its founders.
Its proprietary models and detailed trading processes are not fully public. Descriptions of High-Flyer as AI-driven establish the importance of machine learning to its research approach, but do not reveal the complete design of individual strategies or demonstrate that every investment decision is made autonomously.