Lobsters cannot cross the "moat" of quantitative investment; leading quantitative institutions with over ten billion poured cold water on the AI investment craze.

```

In the spring of 2026, the financial world was completely stirred by a red "lobster," presenting an unprecedented scenario of "fire and ice."

This open-source AI agent, named OpenClaw, instantly transformed from a geek toy into the focal point of the finance industry.

This frenzy also brought about caution.

This is not merely a technological iteration but an extreme stress test for human nature and risk control boundaries.

As algorithms gain “hands and eyes,” where does the future of investing lead? Is it towards endless leaps in net value curves, or the eve of losing control?

“Lobster” = The Biggest Enemy of Excess Returns?

On the evening of March 12, 2026, while the internet was still celebrating OpenClaw’s “revolutionary” impact, a quantitative hedge fund named “Starwide Investment” poured cold water on the hype at this hot moment.

This institution published an article on its official WeChat account, directly addressing the core issue in the opening:

“Let me pour some cold water: the ‘lobster’ you’re busy raising right now may actually be the biggest enemy to your excess returns in quantitative investing.”

Why would a quantitative investment firm be so wary of this open-source tool that’s been revered?

Starwide Investment offered a piercing logical analysis in the article, directly quoting classic financial theory:

“Wall Street figured something out decades ago: all information obtainable from public sources has already been reflected in stock prices. In finance, this is called the ‘semi-strong efficient market.’ Simply put: public information doesn't generate excess returns.”

This statement may be somewhat obscure for ordinary investors, so Zishi Tang (the column) uses a simple example to explain:

Imagine at an auction, an antique’s true value is one million. When details (age, maker, condition) are published in the newspaper (public information), all the experts immediately rush to the auction. By the time you read the newspaper, spend three days researching, and go to the auction, the price has already been bid up to one million or even 1.05 million.

At this point, your judgment of “this antique is worth one million” based on the newspaper is the “consensus.” You can’t profit from this information because the price already includes it.

Market Pricing Gets Faster

Starwide Investment pointed out that AI tools like OpenClaw essentially just reduce the time for you to “read public information” from three days to three minutes:

“AI lets you process public information in three minutes instead of three days, but the value level of that information doesn’t change. What’s more, institutions already have more advanced tools integrated into their processes, so public information will only be priced faster and more thoroughly.”

In other words, while investors are amazed that AI can read ten thousand financial reports in a minute, top institutional algorithms have already digested and traded on the data within milliseconds.

From Starwide Investment’s perspective, the costly “lobster” investors are cultivating ultimately produces the least valuable thing in investing:

“The ‘lobster’ you spend heavily on only produces the least valuable thing in investments: consensus.”

“Using consensus to quickly understand an unfamiliar company is fine. But real Alpha (excess returns) always comes from ‘non-consensus’—which is something the ‘lobster’ can't give you.” So wrote Starwide Investment.

So-called “non-consensus” refers to unique insights not yet known by the public or reflected in prices, such as unpublished supply chain research, exclusive data sources, or deep behavioral insights. These are blind spots that OpenClaw, relying only on public web data, can never reach.

Starwide Investment, issuing this warning, has its own story.

Public information shows that Starwide Investment was founded in 2020. Its founder, Deng Jian, previously worked at established quant firms like Jiukun Investment and Innova Assets, with a strong technical background.

At its inception, Starwide quickly entered the “one billion club” thanks to outstanding performance, once becoming a favorite among top brokers like CITIC Securities.

However, as market styles shifted, its asset management size fell back to the 1-2 billion RMB range. After dropping below one billion, the company became low-profile, rarely speaking in the public eye.

This time, at the peak of OpenClaw’s popularity, this hedge fund decided to make a high-profile “return” by pouring cold water.

Why Did OpenClaw Suddenly Explode in Popularity?

Before Starwide Investment’s warning, OpenClaw had already sparked a frenzy on social media.

For many investors unfamiliar with technology, this name rapidly spread alongside tags like “disruptive,” “free,” and “automated.”

The software’s logo, reminiscent of a red crustacean, led social media to nickname “OpenClaw” as “Lobster.”

But beyond marketing filters, what exactly is OpenClaw? Why has it attracted such huge attention?

Simply put, OpenClaw is a recently developed open-source AI agent by an Austrian engineer.

The “open-source” attribute means its core code is fully public—anyone can download, view, and even modify it. This enabled it to quickly gather global developer intelligence and iterate fast.

OpenClaw became popular recently due to its positioning: a “super employee with hands and eyes.”

In layman’s terms: regular AI is like an “advisor who can only speak,” so when you ask how to analyze a company, it can only give you written advice. But OpenClaw can not only think but also directly control your computer or mobile devices, performing many practical tasks.

Fire and Ice: While Some Pour Cold Water, Others Rush In

Starwide Investment’s warning was a bucket of cold water trying to dampen the blind enthusiasm for OpenClaw.

But on the other side of the investment community, the story reached a totally opposite climax.

Despite the reminder “public information ≠ excess returns,” many top domestic investment institutions and brokers have quietly moved, or even openly embraced the “lobster.”

EFund is a prime example.

As the “aircraft carrier” of the mutual fund industry, EFund’s moves are often seen as an industry barometer.

According to reports, EFund quickly formed dedicated teams to test and explore OpenClaw in isolated network environments.

EFund’s fintech division representatives reported that their use focuses on automated collection and analysis of market information, data governance, and more.

If mutual funds are “verifying,” then some “aggressive” quant hedge funds are already in “practice.”

A quant firm managing about 1-2 billion RMB—Diewei Hedge Fund—recently posted a notable roadshow announcement.

Unlike traditional roadshows pitching products, this time they thoroughly dissected a 7*24 hour quant research agent group.

The topics revealed that this system is not just a “chatbot,” but aims to build a full-process “unmanned research factory.”

Roadshow details: 1) Discussion of why IMO (International Math Olympiad) level reasoning is the entry threshold for quant agents, and how LLMs (large language models) transition from “knowing more” to “doing more.” 2) Special study on why generic agents fail in quant scenarios. 3) Simulate a hedge fund office where multiple AI agents collaborate to achieve “1+1 > 100” efficiency.

This marks that, for some buy-side institutions, OpenClaw is no longer just an auxiliary tool.

Furthermore, institutional activity couldn’t happen without the push from sell-side research.

Zishi Tang’s review of research reports reveals that an “OpenClaw report arms race” began earlier this year.

Since February 4, many top brokers—including Shenwan Hongyuan, GF Securities, Huachuang Securities, Founder Securities, Guojin Securities, Zheshang Securities—have had their financial engineering teams issue in-depth reports on “OpenClaw enabling financial research.”

These reports are not superficial.

From how to install Skills (skill plugins), to configuring Tavily Search (precise search tool), Office-Automation (office automation skills), and even specifics like connecting Agents with Wind’s database for US stock market analysis.

Some reports directly provide tutorials on “four practical work modes”: workflow planning → stock picking strategy → individual stock analysis → quant strategy building.

Brokers’ Internal Networks Ban ‘Raising Lobster’

While investment institutions celebrate the “efficiency revolution,” a stricter “ban” had quietly descended.

According to public reports, at least 20 domestic brokers have issued internal bans to strictly control or prohibit the use of OpenClaw.

This list covers nearly half the industry, including CITIC Securities, Guotai Junan, Huatai Securities, China Merchants Securities, GF Securities, Dongwu Securities, Zheshang Securities, Guojin Securities, and other top-plus mid-sized brokers.

The core rule is “simple”:

Absolutely forbidden to install or run OpenClaw on company devices, business networks, or information systems.

Any devices with OpenClaw must be uninstalled immediately; personal devices with the program may not connect to corporate networks.

This “physical isolation” red line bars the “lobster” from financial giants’ systems.

National regulators have also issued top-level risk alerts.

On March 10, 2026, the National Internet Emergency Center issued a “Risk Alert Regarding OpenClaw Safe Usage,” strictly warning about four major risks:

“Prompt injection, accidental deletion of core data, plugin (skills) poisoning, and public high/medium-risk vulnerabilities that are easily exploited.”

On March 13, the National Cyber/Information Security Notification Center’s monitoring data showed:

“Many OpenClaw assets exposed to the Internet are at high security risk and are likely targets of cyber attacks. Among these, agent behavior is uncontrollable and difficult to regulate. While executing commands, OpenClaw agents are prone to permission loss, execute tasks beyond authorization, ignore user instructions, and may delete user data, steal user information, or take over terminal devices, causing major financial losses.”

OpenClaw’s popularity acts as a mirror, reflecting the reality of society in an era of technological transformation.

Risk Alert and DisclaimerThe market has risks; investment requires caution. This article does not constitute personal investment advice and does not consider the special investment objectives, financial situation, or needs of individual users. Users should assess whether any opinions, viewpoints, or conclusions in this article suit their specific circumstances. Investment based on this article is at your own risk. ```