10天賺10萬美元，OpenClaw在預測市場的實戰經驗訪談

Original | Odaily (@OdailyChina)

Author｜Golem (@web3_golem)

Currently, "raising a lobster" is no longer difficult, but how to master OpenClaw and truly use it to make money directly still puzzles many lobster raisers.

Last week, I spoke with some deep OpenClaw users in the crypto space. Some mentioned that OpenClaw's heartbeat mechanism and scheduled tasks can improve the efficiency of news trading. However, more interviewees felt that using OpenClaw to trade cryptocurrencies for profit remains challenging. (Related reading: Odaily Editorial Tea Party)

But where there's a will, there's a way. A trader named Kevin managed to quadruple his $30,000 principal in 10 days with the help of OpenClaw, achieving a net profit of $100,000 (currently slightly retraced to $82,000). Kevin himself said it started as an experiment, and he didn't expect to actually make money.

So how did Kevin transform OpenClaw from a toy that only burns tokens into a money-making machine? Odaily spoke with Kevin. He shared his personal crypto journey and how he leveraged OpenClaw in prediction markets, hoping readers can gain some inspiration.

The Transition from "Scientist" to "Oracle"

Kevin's early career primarily involved ERP architecture design for enterprises. He later joined a top 3 internet giant in China to build a sports event betting software system from scratch. This professional experience laid the foundation for Kevin's current achievements in prediction markets. After 2018, Kevin entered the Web3 investment sector, mainly incubating and accelerating startups.

However, Kevin's true first pot of gold came five years after entering the crypto space. In 2023, ordi emerged, ushering in the "Inscription Summer" for the crypto market. With a background in computers and code, Kevin became one of the highly sought-after "scientists" at that time (Odaily note: "Scientists" refer to those who can write programs and code to quickly participate in new asset deployments during inscription launches).

"The period when ordi was listed on Binance was when my account value peaked. I ultimately cashed out around over 2 million RMB," Kevin said, noting he was also among the first batch to participate in the ordi launch, with a cost of less than 1 RMB per token, subsequently riding a thousand-fold increase.

After inscriptions completely cooled down, Kevin began searching for other opportunities. Finally, in the summer of 2025, he started seriously researching and participating in the prediction market Polymarket. "I had played Polymarket before, but the liquidity was poor, so I ignored it," Kevin said. For someone who had worked in traditional sports betting, the early Polymarket's trading depth was completely insufficient.

However, after Polymarket successfully predicted Trump becoming the 47th US President in 2025, Kevin's attention returned to Polymarket. "After 2025, Polymarket's reputation grew. It could handle large orders in terms of liquidity, and more importantly, deposits and withdrawals were very convenient," Kevin explained. Therefore, he began experimenting with running algorithms on Polymarket, becoming an "oracle."

Kevin's prediction market journey is divided into two phases: before using OpenClaw assistance and after using OpenClaw assistance. For clarity, Odaily has condensed Kevin's sharing as follows, enjoy～

How to Play Prediction Markets Before Using OpenClaw

Odaily: In the summer of 2025 when you started playing Polymarket, how much did you invest, and what was the final profit?

Kevin: I invested a total of about $100,000 at that time. By this year, the total profit was roughly double the principal.

Odaily: What was your main strategy?

Kevin: I don't place bets myself; I earned profits through writing programs for automated arbitrage. When I worked at a Web3 internet company building sports event betting systems, I was also involved in order book design. This experience was very helpful for understanding Polymarket's order book. Therefore, I used programs to capture user spreads between different market makers, especially in sports events, where a lot of sentiment arbitrage can be done.

Odaily: Do you have a dedicated team, and is anyone providing you with capital?

Kevin: I'm doing this alone; having AI assistance is enough. Initially, I was afraid Polymarket might block withdrawals, so I ran dozens of accounts. But later, I found the deposit and withdrawal process was smooth, so I reduced the number of accounts. I mainly run strategies with my own money, but indeed, some people provide capital for me to run strategies for them. However, this is just one way of making money.

How to Play Prediction Markets After Using OpenClaw

Odaily: When did you start using OpenClaw to play prediction markets?

Kevin: At the end of February. This was also an experiment to see how much money OpenClaw could make at the trading level, but I didn't expect to actually profit. For example, with the account KevinChe202603, I turned a $30,000 cost into a peak profit of $100,000, taking only 10 days.

Odaily: So what is your specific strategy?

Kevin: Frankly, this account's strategy is hybrid. Currently, 60% is still running the previous automated arbitrage algorithm, and 40% is using the "lobster" for subjective betting. Compared to market-making arbitrage, betting is a complex decision-making process, requiring consideration of smart money in prediction markets, public sentiment, team lineups, player form, etc. OpenClaw's role here is to actively collect various factors that determine match outcomes and turn them into an indicator. After several training sessions, it can even find other influencing factors I might overlook, saving me a lot of time and mental effort.

Odaily: But isn't this just AI predicting matches? Conversational AI can do that too, and some developers have even created specialized AI match prediction tools. What's special about OpenClaw?

Kevin: Having an information advantage is just one of OpenClaw's strengths. It can also discover new strategies on its own, conduct its own backtesting, and perform automated betting in matches. If the strategy is good, we just need to give the money to OpenClaw; everything else is automated. This is something AI prediction tools cannot do. For example, it can proactively discover some smart money addresses and "dumb money" addresses, either following the smart money's bets or using the dumb money addresses as contrarian indicators.

Furthermore, this is why Polymarket integrates particularly well with OpenClaw among all prediction markets—because Polymarket's API is the most AI-friendly, making data access very convenient for AI.

Odaily: In which areas is OpenClaw primarily placing bets now? Is it fully automated already?

Kevin: Based on my areas of expertise, OpenClaw is currently also mainly experimenting in the sports competition field. However, once mature, I will consider letting OpenClaw expand to other areas. Currently, I give OpenClaw small amounts of capital for automated betting, around $1,000. I still don't dare to put too much money into a fully automated account.

Odaily: Is your strategy replicable? Or will you write a Skill for the market in the future?

Kevin: I'm also trying because there is indeed user demand, to see if I can combine my methodology to allow everyone to build a profitable lobster. Later, I also plan to package some Skills for the market to use, which will definitely be paid, of course.