#Gate广场AI测评官

Did People Make Money Trading Stocks with "Lobster"?

Recently, after OpenClaw ("Lobster") went viral, many "smart people" have applied "Lobster" to the stock market for backtesting and stock selection, using it as a "stock trading assistant." So, is using "Lobster" for stock trading actually reliable?

😂 "Lobster Stock Trading, Lost 7000 in Three Days"

After losing 5840 yuan on March 17, "Lobster" comforted Li Yue during the final backtesting: "Today's operations didn't go smoothly, we'll try again tomorrow."

From March 13 to 17, Li Yue's "Lobster" recommended three stocks in total. "Every one of them lost money; the three stocks lost 7000 yuan combined over three trading days," said Li Yue, showing his trading activity on March 17. At 10:05 on March 17, "Lobster's" suggestion was to stop loss on an electric power equipment stock, which resulted in a 6.33% loss; three minutes later, "Lobster" "guided" him to buy a solar concept stock, which showed a floating loss of 3.27% by market close that day.

Li Yue stated that currently his requirement for "Lobster" is to report the day's trading plan before market open, report on holdings every 20 minutes during trading hours, and summarize the day's experience and lessons after market close, allowing "Lobster" to self-improve.

However, after reviewing the operations of these few days, Li Yue discovered that his "Lobster's" stock selection logic primarily relies on the previous trading day's Dragon and Tiger boards. Additionally, what troubles Li Yue is not just the account losses; in his view, "Lobster's" memory is somewhat poor—there are things it keeps repeating that "Lobster" still doesn't remember.

For Mr. Wang from Dalian, Liaoning Province, "Lobster" is more like a "regularly scheduled assistant." Mr. Wang stated that he had "Lobster" combine Python and financial APIs (Application Programming Interfaces) to automatically conduct market backtesting according to his preset logic. During the backtesting process, "Lobster" will score different stocks and then generate reports to send to him.

"I came into contact with 'Lobster' in January this year, and overall the experience has been quite good," Mr. Wang said. Ultimately, his stock selection will still combine macroeconomic conditions and market environment to determine which stocks to buy; "Lobster" is just an auxiliary tool for judgment.

👋 "It Can Improve Efficiency, But Not Win Rate"

In the view of industry insider Timi, using "Lobster" for stock trading can only improve efficiency, but cannot improve win rate.

"The capability boundaries of large language models are determined by their technical architecture. Due to limited context window, it can only process information within a certain volume, not the comprehensive data analysis one might imagine."

The true value of "Lobster" lies in improving information processing efficiency, such as organizing listed company data and building richer information networks in a short time. It is essentially an extension and leverage of personal ability, not a disruptive trading tool.

For ordinary investors, what needs to be guarded against is cognitive misalignment. If one has not mastered basic trading logic and enters the market solely with the idea of "let AI make money for me," the probability of success is nearly zero. "It's like quantitative trading—many trading strategies operate at millisecond levels, not relying on advice from 'Lobster' that comes 10+ seconds later."

💡 What Risks Exist?

While efficiency improves, unknown risks often accompany it.

On March 11, the Industrial and Information Technology Department's Cyber Security Threat and Vulnerability Information Sharing Platform released "Advice on Six Do's and Six Don'ts to Prevent OpenClaw Open-Source AI Agent Security Risks," pointing out the risks posed by this tool and recommending that users deploy on isolated network segments when deploying and applying it, strengthen permission management, establish manual review and circuit-breaker emergency mechanisms, and add double confirmation and other security measures for critical operations.

"Lobster" and other AI agents are entering the daily lives of ordinary people with low barriers to entry and high accessibility. However, behind the convenience of technology lie hidden risks. In daily use, users often need to grant the tool access to accounts, passwords, and even system permissions, creating data security hazards. When this logic extends to the investment field, risks escalate from information leakage to direct threats to fund safety.