AI Agent OpenClaw Will Do Your Trades. Should You Try It? | 巴伦周刊

01 文章选材

【Para.1】I came for the lobster and stayed for the covered calls.If a crustacean and options trading mash-up has you vexed ,understand that I’m talking about OpenClaw, the red-hot artificial-intelligence agent that some hearty souls are using to manage their investments these days. And yes, it was OpenClaw’s lobster motif that first caught my eye.

【Para.2】Unlike chatbots such as ChatGPT (sooo 2025), OpenClaw can proactively execute your tasks, like make appointments or buy tickets to a show. It can also advise you on investment strategies and even trade securities while you’re sitting on a beach.What could possibly go wrong?

【Para.3】OpenClaw, which was created last November by Austrian vibe coder Peter Steinberger (who now works for OpenAI), reportedly has 38 million monthly users (many in China) sloshing around its lobster-nomenclatured ecosystem. Discord’s OpenClaw channel, Friends of the Crustacean, is a flurry of often deeply technical activity. (My handle is Andre the Seal.)

【Para.4】“I started this business as a computer programmer, so I’ve always been building algorithms,” says Thomas Peterffy, chairman of Interactive Brokers Group. “We’ve given algorithms to clients for many years, and they automatically execute orders. AI is just like a new, even higher level computer language. We are big users. Certainly, it’s the future.”

【Para.5】Interactive Brokers’ customers use AI either through external agents like OpenClaw or through the brokerage’s internal tools provided by partner company Reflexivity. “The problem was never a shortage of data. It was that it would take days to work through. Now it’s like eight minutes,” says Jan Szilagyi, CEO and co-founder of Reflexivity, a Harvard University Ph.D. in economics and finance who once helped manage global macro portfolios for Stan Druckenmiller.

【Para.6】On Wednesday morning, with the market roaring after the announcement of a two week cease-fire between the U.S. and Iran, I asked Szilagyi to help me set up a post-TACO trade (TACO, of course, stands for “Trump always chickens out”). The algo came back with an 11-page report concluding that “first day TACO trades worked TOO WELL,” and suggested (with entry points) strong long positions in crude oil and the Cboe Volatility Index, or VIX; a moderate long on the State Street Energy Select Sector SPDR exchange-traded fund; a moderate short on the S&P 500 index; and a weak short on Treasuries.

【Para.7】Interactive scrutinizes customers using external tools like OpenClaw. “We have a program to catch orders they send and examine them within a fraction of a second to ensure they comply with regulations and applicable margins,” says Petterfy.

【Para.8】What if OpenClaw sells my stocks against my wishes? “That’s not my problem,” he says. “That’s your problem. We check the orders and execute.”

【Para.9】That problem might be closer to home over at upstart stock-trading platform Public.com, which has gone whole hog (or is it whole shellfish) into AI. Public’s customers can connect OpenClaw directly to their accounts to analyze and, yes, trade securities.

【Para.10】Privately held and venture-backed by Accel and Tiger Global, New York–based Public has billions in assets under management and millions of customers, according to Jannick Malling, Public’s Danish-born co-founder and co-CEO. Public began exploring AI in 2023 to create “the world’s first agentic brokerage,” he says. OpenClaw is bringing that endeavor to a boil.

【Para.11】“Trading went from phone-based, to the web, to mobile apps, and now to agentic AI,” says Malling. “That you can just type in ‘Find me three different ways to make 10 grand this month selling covered calls across my portfolio’ is a massive shift.”

【Para.12】Setting up OpenClaw requires digital elbow grease. “It took me 1½ to two hours to set up,” says Dean Ahrens, a 19-year-old trader from Westminster, Colo., who typifies the customers Public is looking to net. Ahrens recently used OpenClaw to build and execute a put-option strategy on the Invesco QQQ Trust ETF and to help find and buy puts on Hims & Hers Health.

【Para.13】On the institutional side, Cliff Asness, who runs giant quant hedge fund AQR, was once cynical about AI. “I probably slowed us down a couple of years on this,” he says, “which is rational when you have something that’s worked for 30 years. All these decisions involve dollars, time, and risk.”

【Para.14】Asness, 59, who notes analysts at his funds have been sifting through credit-card data and positive and negative words in earnings statements for decades, became a convert. “It turns out machine learning”—a subset of AI—“is better than our old methods,” he says.

【Para.15】Historically, AQR’s positions were built conceptually 50% on historical data and 50% on economic story, Asness says. Now, because of AI, it’s more like two-thirds data, one-third story.

【Para.16】Have AQR’s returns improved? “Well, it’s been a good five years,” he says. Asness is less sanguine about AI for retail investors. “I’m leery about a world where machine learning is on everyone’s lips. This notion that you can pick just a few stocks with it may not be worse than any other method, but I don’t see any magic there.”When it comes to OpenClaw, bon appétit. Just don’t get pinched.

04 检验题 

Read the following text carefully and answer the questions below by choosing A, B, C or D.

21. According to Paragraphs 4 and 5, how does Jan Szilagyi characterize the primary advantage of AI-powered tools in investment analysis?

[A] They generate more accurate financial predictions than human analysts ever could.

[B] They drastically reduce the time required to process large volumes of market data.

[C] They enable ordinary investors to access the same strategies as large hedge funds.

[D] They eliminate the need for any manual intervention in securities trading.

22. Thomas Peterffy’s remarks in Paragraphs 4, 7, and 8 suggest that Interactive Brokers’ stance toward customers’ use of external AI agents is one of:

[A] cautious oversight — the firm monitors such orders for regulatory compliance but holds users responsible for their own trading decisions.

[B] enthusiastic endorsement — the firm actively encourages all clients to adopt AI-driven trading tools.

[C] reluctant tolerance — the firm allows external agents only because competitors already do so.

[D] outright opposition — the firm regards AI agents as too risky to permit on its platform.

23. The author mentions Dean Ahrens’ experience in Paragraph 12 primarily to:

[A] demonstrate that young, inexperienced traders are particularly vulnerable to AI-driven trading risks.

[B] illustrate how OpenClaw can be configured and deployed for practical options trading by individual investors.

[C] argue that Public.com should lower the technical barriers to using AI trading platforms.

[D] suggest that AI has made professional-level trading strategies accessible to users of all skill levels without effort.

24. The word “sanguine” in the last paragraph is closest in meaning to:

[A] hopeful.

[B] knowledgeable.

[C] indifferent.

[D] consistent.

25. Which of the following best summarizes the main idea of the text?

[A] OpenClaw, a popular AI agent, is poised to replace traditional stockbrokers and fundamentally transform retail investing.

[B] AI agents like OpenClaw are being embraced across the financial industry, but their adoption comes with both promise and significant caveats.

[C] Institutional investors have proven that AI-driven trading consistently outperforms human decision-making.

[D] Regulatory authorities should impose stricter controls on AI agents that execute trades on behalf of retail investors.