LP taught me how to invest using Doubao: A private equity GP’s career transition story

Original | Odaily Planet Daily (@OdailyChina)

Author | Golem (@web3_golem)

As LPs learn to use AI, life is getting tougher for small private fund managers.

Ergou (@ryansoon777) was a general partner (GP) at a small offshore private USD fund focused on US stocks before the new year. But after the new year, he quit to join an AI startup.

"Raising capital for small private funds is inherently difficult these days. With the popularity of AI, many LPs would rather let Doubao assist them in stock trading than give us their money."

Ergou says his decision to switch careers largely stems from seeing AI's subtle impact on the relationship between LPs and GPs. Information and analytical capabilities are seemingly leveled by AI, making LPs more likely to question a GP's professional judgment, potentially increasing friction and even leading to capital withdrawal or fund liquidation in severe cases.

Small USD Private Funds Were Already Struggling

The private USD fund Ergou previously worked for wasn't in bad shape. It managed assets worth tens of millions of dollars, primarily invested in highly liquid US stocks, with a small allocation to crypto assets. Its annualized returns over the past three years significantly outperformed the Nasdaq.

Logically, with solid performance and increased investor demand for overseas wealth management in recent years, fundraising shouldn't be too hard. However, Ergou revealed it's nearly impossible for small USD funds like theirs to attract institutional LPs.

Currently, top-tier Chinese hundred-billion-dollar private USD funds (like Jilin, Hillhouse, and Boyu) typically use an "offshore + onshore" structure: the fund entity is domiciled in the Cayman Islands, often as a Cayman exempted company or Cayman SPC, while the management entity is based in Hong Kong or Singapore.

In recent years, due to regulatory and fundraising environment changes, more and more private USD funds are adopting pure onshore structures like Hong Kong LPF or Singapore VCC.

However, the small private USD fund Ergou worked for still used the most "primitive" structure: Cayman SPC + BVI (British Virgin Islands) fund manager structure.

A common saying in the fund industry is that LPs determine the structure. One reason top Chinese private USD funds stick with the "Cayman" model is that their overseas LPs include US university endowments, Middle Eastern sovereign wealth funds, and large European family offices. These top-tier "old money" institutions have been familiar with the Cayman structure for decades. Adhering to this convention helps reduce communication and trust costs between them.

But small Chinese private USD funds also domiciled in the Cayman Islands can't attract these top international capital sources. Their LPs are primarily based in Asia, putting them in an awkward position.

From an Asian perspective, the backers of USD private funds mainly come from private banks, mainland China (outbound capital), Hong Kong family offices, and Southeast Asian tycoons.

Even for similarly sized small private USD funds, these circles have a natural affinity and sense of security towards Hong Kong or Singapore, making them more willing to invest in Hong Kong LPF or Singapore VCC structures rather than Cayman SPC.

Besides the limitations imposed by fund structure and size, differences in investment strategy also made fundraising difficult for Ergou and his team.

Private fund investment strategies mainly fall into two categories: discretionary and quantitative. Discretionary strategies rely on the GP's research, experience, and judgment to decide what to buy and sell, with profitability hinging on the fund manager's market insight. Quantitative strategies involve translating investment logic into mathematical models and programs, executing trades automatically or semi-automatically at high frequencies, with profitability based on statistical patterns within the model.

"Currently, funds using quantitative strategies find it much easier to raise capital than those using discretionary strategies. Especially with AI empowerment, LPs are increasingly believers in quant." Ergou noted that DeepSeek (Odaily note: incubated by quant fund High-Flyer's team) exploded in popularity last year, further fueling market enthusiasm for quantitative strategies.

Furthermore, a key difference between quant and discretionary funds is that quant strategies can showcase data and algorithms to gain LP trust. Whether the fund is profitable or experiencing a drawdown, it remains within a controllable range. Excellent quant funds can even function as fixed-income products. Discretionary strategies, being more abstract, require GPs to expend more communication effort to gain LP trust fully. Especially during significant drawdowns, LPs can easily question the GP's investment ability.

Therefore, in summary, the living space for small USD private funds like the one Ergou worked for in China has been compressed by the broader environment, making fundraising increasingly difficult. Moreover, the few remaining large LPs within the fund are also questioning whether AI's "investment ability" far surpasses that of the GPs.

A Mixed Bag of LPs

"In the past, LPs generally listened to us because we were professionally trained. But now, they throw our reports into AI to translate into plain language and then come back to 'teach' us how to do it," Ergou said, noting that since AI became widespread, LPs, who previously only cared about final results, have become significantly more 'concerned' about his investment operations.

Ergou even had to remove an LP once. This was a 50-year-old real estate entrepreneur with a very "Chinese entrepreneurial" demeanor. He invested about $1 million into the fund Ergou worked for but didn't let go of control. He frequently argued with Ergou using fragmented market information and conclusions drawn from AI. "His attitude was terrible. He thought a young guy like me knew nothing. We couldn't build trust, so after some mediation, we eventually removed him."

"To be honest, our LPs are exceptionally talented in their own fields. They are authorities in their domains. But now, with AI as their aide, they believe they've also become authorities in investing," Ergou lamented.

Small USD private funds have narrow fundraising channels, often relying on the friends of the boss or referrals from acquaintances. Hence, their LPs are a "mixed bag." According to Ergou, his fund's LPs included high-net-worth individuals, entrepreneurs from traditional industries, and funds of funds (FOFs). "Our LPs included a Shanxi coal boss, a billionaire ranked in the Forbes top 300-400, and even a wealthy second-generation who introduced his father because we got along well."

Their relationship with LPs was also delicate. For some LPs, they wouldn't even charge the 2% management fee, only taking the 20% performance carry. The main characteristic of this LP base is enthusiasm for participating in financial markets and "taking capital offshore," but lacking the time and energy to rapidly learn and research market trends.

Therefore, in a sense, the GP's core value lies in shouldering information gathering, market research, opportunity screening, and investment judgment for the LP. They use their professional expertise to compensate for the LP's deficiencies in time, energy, and cognition, thereby completing the transformation from information to decision-making.

However, with the proliferation of AI tools, this information processing and research capability, once heavily reliant on professional institutions, is rapidly being democratized. Except for the final capital allocation and trade execution, much of the GP's traditional workload is being replaced by AI at lower cost and higher efficiency.

"It's not hard for our LPs to open a broker account with IBKR. With AI assistance, they can buy whatever industry or asset they like all by themselves." Ergou believes AI's impact is particularly severe for funds employing discretionary strategies because investing is ultimately results-oriented. If an LP catches a wave and their personal investment returns surpass the fund's, they'll naturally start questioning the fund's capabilities.

In contrast, the "information democratization" brought by AI has a smaller impact on quantitative funds and might even widen the gap between funds.

The parameters and algorithms within quantitative fund strategies are constantly evolving. AI accelerates this iteration, making it a field of efficiency and intelligence. An average person using ordinary AI to build a quantitative strategy can absolutely not compete with large quantitative funds that possess specialized knowledge in mathematics, finance, and other areas.

"Quantitative strategy inherently requires consistently staying ahead of market peers to generate alpha. If you think your basic AI has crafted a good strategy, it's likely already been discovered and iterated upon by most smart people," Ergou notes, highlighting the advantage of top-tier quantitative funds.

Will AI Replace GPs?

However, Ergou isn't worried that AI will completely replace professions like GPs or analysts. AI remains neutral and accessible to everyone; it's a lever. GPs can use AI to enhance their knowledge systems and investment strategies, generating more returns for LPs. What truly frustrates Ergou is that AI increases friction between GPs and LPs.

"Some LPs even question why we didn't invest in the currently hot assets, and they analyze it quite convincingly. They don't understand that GPs don't just invest in whatever is popular," Ergou finds this phenomenon a bit exasperating, especially after tech stocks like AI and semiconductors became hot themes in the US market this year, allowing retail investors to achieve excess returns by simply betting on sector leaders.

In a bull market, retail investors can indeed easily outperform funds. First, individual investing is more flexible, offers more room for error, and capital is more concentrated. Second, with AI-assisted research, retail investors' research efficiency is also significantly enhanced, like having an all-around expert on standby 24/7.

Especially in this year's US stock market, if retail investors bet on hot memory stocks like Sandisk, Micron, or SK Hynix, their return on investment might exceed most funds. "At this point, LPs might either consider putting more capital into their personal accounts and less into the fund, or they might directly withdraw from the discretionary fund," Ergou says, noting that during a bull market, everyone often feels like a "stock god."

But all this hinges on retail investors using AI correctly. Using inferior AI can be counterproductive. Ergou points to this as the biggest source of friction with his LPs. "High-net-worth individuals in China primarily use companion-style dialogue AIs like Doubao, while more analytical tools like ChatGPT and Claude are less common. This companion-style AI, designed to provide emotional value, is highly prone to hallucinations in professional fields."

Essentially, the problem isn't AI's capability, but that most people don't truly understand how to use AI. AI can integrate vast amounts of information in seconds, building a logically coherent analytical framework. But logical coherence doesn't equal factual accuracy. For LPs lacking professional backgrounds, it's often difficult to distinguish conclusions based on real data from probability-based inferences generated by the model.

Therefore, most investors aren't so much seeking analysis from AI as they are seeking validation. AI's ultimate goal isn't to help investors "distinguish truth from falsehood," but to complete a conversation.

So, will AI replace GPs? AI can cheaply generate ten thousand logically sound investment research reports. But the essence of asset management is an "ancient service industry" based on trust and the entrusting of one's mind. The GP-LP relationship is also a process of mutual selection.

In a future where all "tasks" will eventually be executed by AI to maximize "results," "human private funds" might also need to learn from AI and practice providing more emotional value.