Author: Max.s
After the dramatic fluctuations of 2024 and the profound reshuffling of 2025, the quantitative finance industry is standing at a new crossroads. At last week's 2025/2026 China Quantitative Investment New Year's Summit, Dr. He Kang, Chief Strategist and Chief Financial Engineer of Huatai Securities Research Institute, delivered an in-depth speech entitled "Quantitative Industry Trends in 2025 and Outlook for 2026." This is not only a strategy report on the A-share market, but also a battlefield manual on how Alpha can find new space for survival in an increasingly crowded market.
For practitioners at the intersection of Web3 and traditional finance, this report reveals a clear signal: traditional alpha is waning, while new paradigms—whether it's "Order as Token" based on large models or alternative assets represented by cryptocurrencies—are becoming a battleground for institutional investors.
The following is an in-depth review and industry outlook based on Dr. He Kang's speech.
2025 will be a year of both high growth and high volatility for the quantitative finance industry. A significant change is that while the size of securities-focused private equity funds remains high, the growth of public quantitative funds is even more rapid. As of the third quarter of 2025, the size of public index-linked funds has exceeded 200 billion yuan, of which actively managed quantitative funds reached 120 billion yuan.
Behind this lies an interesting structural change: the top-ranked player has changed.
The previous dominance of top players has been disrupted, with institutions like Boda and Guojin emerging as dark horses thanks to their extremely flexible strategies. Dr. He's research revealed that these top-performing public quantitative funds are essentially "private funds disguised as public funds." They boast extremely high turnover rates, astonishingly rapid strategy iteration, and even rival top private funds in their use of intraday trading (T+0).
This phenomenon reveals the survival rule for 2025: due to the exponentially increasing difficulty in obtaining excess returns, only extreme flexibility can break through in a red ocean market. For investors, the past allocation logic of "choosing big names and lying low" is no longer applicable ; they must use more refined attribution analysis to identify those managers who truly possess "agile development" capabilities.
For the past five years, the mainstream narrative in the quantitative trading industry has been "fully invested in stock picking," using alpha from stock selection to cover market fluctuations. However, after the market education of 2025, "market timing" has returned to the center stage. Dr. He Kang categorizes market managers into five types: A, B, C, D, and E, with the most noteworthy being type E—logic-based market timing . Unlike black-box predictions, this type of strategy constructs an explicit logical chain of "If A then B."
The rise of sub-domain modeling.
As market efficiency improves, universally applicable factors become increasingly difficult to identify. Top managers are adopting a "divide and conquer" strategy: dividing the entire stock market into different "domains" such as growth, cyclical, small-cap, and micro-cap, and training models separately within each domain. This is similar to how in Web3, you can't use the same logic to trade Bitcoin and on-chain Meme coins—their pricing logic, liquidity characteristics, and participant structures are completely different. Through domain-specific modeling, quantitative strategies can extract higher excess returns in specific market segments.
If domain-specific modeling is a tactical optimization, then the introduction of Large Language Models (LLM) is a strategic dimensionality reduction attack. Dr. He Kang mentioned three levels of application of large models in quantitative finance, among which the third level is the most memorable: treating financial transactions as a language, namely "Order as Token".
In traditional NLP (Natural Language Processing), GPT predicts the next word (token); however, in large-scale financial models, the input is a price sequence, trading volume, and order flow over a past period, and the model predicts the next "price token." This is not just a technological shift, but a revolution in thinking.
Traditional quantitative models are often based on statistical linear or nonlinear regression, while the Transformer architecture allows models to capture dependencies over extremely long periods and complex nonlinear patterns. Imagine that future trading will not be based on a linear weighting of a few factors, but rather on a pre-trained large financial model that "generates" future price paths, much like generating text. This is strikingly similar to the intent-centric AI agent trading logic currently prevalent in the Crypto space—AI is no longer an auxiliary tool, but the direct executor.
The Blue Ocean of Alternative Data: The Institutionalization of the Cryptocurrency Market
When the excess returns of the A-share market are pushed to their limits, smart money begins to turn its attention to alternative markets with lower correlations through return swaps (TRS) or offshore entities.
Compared to the T+1 settlement system and daily price limits in the A-share market, the crypto market features 24/7 trading, T+0 settlement, high volatility, and fragmented liquidity. For quantitative institutions with high-frequency trading capabilities and risk control models, this is practically the A-share market before 2015—alpha is everywhere, and the competitive landscape is far from solidified.
This section specifically introduces the Funding Rate Arbitrage strategy. In the perpetual contract mechanism of the crypto market, both long and short positions need to pay funding fees to maintain price anchoring. During bull market cycles, long positions often have to pay high fees to short positions. This creates a fixed-income-like "market-neutral strategy": buy spot and short an equivalent value of perpetual contracts, hedging against price volatility risk while steadily earning funding fees. In this area, the 1Token Funding Rate Arbitrage Strategy Index has become an important industry indicator.
Industry data shows that this type of strategy has an annualized return far exceeding that of traditional fixed-income products under specific market cycles, and has extremely low correlation with traditional assets (stocks and bonds). 1Token, as a professional digital asset service provider, constructs indices that not only reflect the overall arbitrage opportunities in the market, but also embody the evolution of Crypto quantitative trading from a "handmade workshop" to "institutionalized and indexed" approaches.
For traditional financial professionals, the significance of paying attention to indices like 1Token lies in the fact that they provide a window into Web3 liquidity premiums. When funding rates remain high for an extended period, it signifies extremely euphoric market sentiment and serves as a warning of potential selling pressure in the spot market; conversely, it could present a good opportunity to buy on dips.
Looking ahead to 2026, Dr. He Kang's key words are "dynamic" and "antifragile".
From Static Allocation to Dynamic Game Theory: In the past, fund of funds (FOFs) or asset allocation often involved setting a static weight (such as a 60/40 portfolio). However, in the future, a dynamic adjustment mechanism must be introduced. For example, when a certain strategy (such as micro-cap index funds) becomes too crowded, due to the "stampede risk" caused by homogeneous trading, its weight must be proactively reduced, even if its historical performance is excellent.
The "airbag" structure of products: Having experienced the pain of drawdowns, investors' aversion to downside risk has reached its peak. Derivatives with "airbag" or "snowball" structures, as well as index-incrementing products protected by options, will become mainstream in 2026. This is exactly the same logic as DeFi structured products—sacrificing some potential upside potential for greater certainty and principal protection.
Whether seeking independent alpha within the A-share market or allocating to Hong Kong stocks, US stocks, or even crypto assets, the core objective of finding low-correlation assets is to reduce the overall correlation of the portfolio. Dr. He Kang specifically mentioned that although it is difficult to generate pure alpha from Hong Kong stocks (poor liquidity, expensive short-selling tools), their value still exists as part of a diversified portfolio. Meanwhile, the crypto market, with its unique driving logic, will become an important piece of the puzzle in hedging traditional financial risks.
Dr. He Kang's speech actually revealed the essence of financial engineering: the process of finding certainty in uncertainty.
In 2025, the traditional low-hanging fruits of the quantitative industry have been harvested. Practitioners are left with only two paths: either relentlessly hone their technical skills, using large models to uncover deeper nonlinear patterns; or venture overseas on the asset side, aiming for a breakthrough in a blue ocean market like Crypto.
For native Web3 users, this also serves as a warning: with top institutions like Huatai Securities beginning to conduct in-depth research and focus on this area, the entry of established players is only a matter of time. When traditional quantitative trading techniques are applied to decentralized trading markets, new opportunities and fierce competition will arrive simultaneously.
In 2026, whether it's TradFi or Crypto, only the evolved will survive.
