Prediction Market OpenClaw Practical Manual Part 2: From Configuration to Trading in Action

Summary: Step-by-step guide to connecting OpenClaw with Hyperliquid and Polymarket, letting AI place trades for you.
Author: Changan I Biteye Content Team
When was the last time you kept an eye on the market?
The longest I’ve stayed watching the market was from the US stock market open until Asian market close, with a two-hour nap in between. When I woke up, I realized I missed a perfect entry point.
That made me wonder: why must humans do this?
After integrating OpenClaw with Hyperliquid and Polymarket, I tested the answer. Some things went better than expected, some pitfalls were deeper than anticipated. Everything is detailed inside.
Since I’ve experienced rain before, Biteye also wants to hold up an umbrella for you.

1. Installation Steps
   To let AI trade on your behalf, simply installing OpenClaw isn’t enough. It only handles scheduling; the actual trading is done by skills. To run skills, you need to grant the appropriate trading permissions.
   This chapter covers installation, configuration, and notification setup—all at once. Once configured, you can jump straight into the practical part in Chapter 2.

1.1 Install Skills: Find the Capabilities You Need
Any task you want to accomplish can be searched and installed via ClawHub using relevant skill keywords, such as:
hyperliquid skill
polymarket skill
After installation, confirm that the skill has been loaded by visiting the local control panel: left sidebar → click on skill.
If you see the recently installed skill, the installation was successful.
Some skills will show “Save key,” indicating that API or private key configuration is needed—this is normal. The next step is to fill in the corresponding permission information.

1.2 Configure Trading API: Grant AI Order Placement Permissions
To enable AI to place orders, it must have trading permissions.
Avoid using your main account directly; it’s recommended to create a separate test account with only a small amount of funds.
Different exchanges have different methods; here are two examples:

1️⃣ Hyperliquid
Open Hyperliquid → More → API
Generate an API wallet that only has trading permissions, no withdrawal rights.
Enter a wallet name → click Generate to get an API key, with optional authorization duration.
After generating the API wallet, a tip from experience: no need to transfer funds into the API wallet; keep funds in the main account. (Transferring funds to the API wallet will result in loss of those funds.)
During configuration, fill in two pieces of information: the main account’s wallet address (API Key) and the private key shown when generating the API wallet (API Secret).
AI signs transactions via the API wallet, operating on the main account’s funds.

Test method: ask OpenClaw to check your account balance by entering: “Check my current balance on Hyperliquid”
If it returns a specific number, the API setup is successful.

2️⃣ Polymarket
Open the Polymarket settings page:
Settings → Relayer API Keys → API key → click Create New
After creation, you get:
RELAYER_API_KEY, RELAYER_API_KEY_ADDRESS, Signer Address
To trade, you also need to export the signer’s private key and fill it into the configuration.
Using OKX wallet as an example: click the avatar in the top left → Manage Wallet → View Private Key
Then, input the private key along with the API key into OpenClaw’s openclaw.json file.
Open openclaw.json (usually located at C:\Users\YourUsername.openclaw), and fill in three items under the env field.
⚠ The private key is a high-privilege credential—do not screenshot, share, or store it in chat logs.
After configuring the API, you must also perform an on-chain authorization. The simplest way is to manually make a trade on Polymarket, such as buying a 1 USDC position.

1.3 Connect Telegram: Use it on the go
Once the API is set up, Little Lobster can trade, but you probably don’t want to stare at your screen all day, right?
By integrating Telegram, you can receive transaction notifications, error messages, or even place quick orders via your phone. Whether out for a meal or lying in bed, Little Lobster will keep you informed of what it’s doing.

Create a Telegram bot:
Open Telegram, search for BotFather, and create a new bot: send /start → /newbot → follow prompts → after creation, you get a BOT Token.
To get your Chat ID: send a message to your bot, then open: (replace with your token) /getUpdates
In the returned JSON, find chat → id; that number is your Chat ID.
Fill in the two parameters:
TELEGRAM_BOT_TOKEN
TELEGRAM_CHAT_ID
Test by sending: “Send me a test message”
If Telegram receives it, push notifications are working.
If not, common issues include proxy or network problems. You may need to specify an HTTPS proxy port in the configuration.

Open openclaw.json and add:
{“HTTPS_PROXY”: “”, “HTTP_PROXY”: “”}
Fill in the port according to your proxy software.

2. Executing Orders: Turning Judgments into Actions (More Details)
   Since AlphaGo’s match against South Korean Go master Lee Sedol, humans have been especially keen on competing with AI.
   Last October, an AI research group called Nof1 held a competition called Alpha Arena. The rules were simple: give a model $10,000, let it trade perpetual contracts on Hyperliquid, and whoever ends with the highest account balance wins.
   Recently, Aster held two live “Humans vs AI” trading competitions, both won by AI.
   After seeing this, I started to wonder: can I set up a similar pipeline using OpenClaw on Hyperliquid?
   This led to the creation of two skills:

2.1 Hyperliquid: Let AI Open Contracts
I searched ClawHub and found the relevant skill; installation follows the previous deployment tutorial.
Once configured, you can place orders with natural language: “Go long ETH on Hyperliquid with 10x leverage, 30% position.”
But in practice, it’s not that smooth.
The first issue is leverage. When asking for 10x, it often opens at 20x, and repeated adjustments don’t help.
The phrase “10x leverage” is ambiguous for AI; it adjusts based on its judgment of “reasonable leverage.”
The solution was to give clearer instructions, such as having it echo the specific parameters before executing, ensuring stability.
The second issue is position size: telling it “use 30% of balance” often results in full position or an absurdly small one.
The fix was to specify exact amounts, e.g., “Open a 200 USDC position,” which resolved the problem.

My initial goal was to replicate Alpha Arena: make OpenClaw run fully automatically, 24/7, without manual oversight.
This is theoretically feasible, but requires writing strategies in natural language—specifying the asset, entry conditions, position size, stop-loss, and exit conditions.
Once understood, it would monitor and execute based on that logic without manual triggers.
However, the current immature skill ecosystem is a real barrier.
Even with a good strategy, execution issues can cause deviations—similar to AI models running natively on Hyperliquid versus through third-party skills, where stability differs greatly.
This part will be revisited once the skill ecosystem matures.

2.2 Polymarket: Applying the Same Logic to Prediction Markets
After experimenting with Hyperliquid, I became curious: how would OpenClaw perform on prediction markets?
The underlying logic of both markets is quite similar—pricing the probability of an event.
They share the same information sources and similar decision frameworks, differing mainly in execution.
In the previous chapter, I only analyzed the market with PolyClaw without placing actual orders.
Analysis and order placement involve a psychological threshold: misanalysis costs nothing, but wrong orders cost real money.
This time, I wanted to take a step further.
After filling out the three keys and completing on-chain authorization, I asked OpenClaw to find a good entry opportunity.
It presented a market: “Will the Federal Reserve keep interest rates unchanged after the March meeting?”
Current YES price: $0.99, 24-hour volume: $2.4M, market consensus: 99% believe rates will stay the same.
Fundamentals support this view: stable inflation data, moderate economic growth, and a dovish Fed stance without rushing to cut rates.

Before placing an order, I usually ask: “What’s the order book depth? What’s the spread? How much should I buy?”
OpenClaw analyzes the current bid-ask spread; a large spread suggests limiting orders to save unnecessary costs.
Once confirmed, tell OpenClaw whether and how much to buy: “Buy a YES position of 5 USDC.”
After trading, checking positions is straightforward: “OpenClaw, what positions do I currently hold?”
From spotting the opportunity to executing the order, everything is done within OpenClaw—no need to switch pages or tolerate laggy Polymarket frontend.

3. Conclusion
   Those AI models in Alpha Arena, with $10,000 on Hyperliquid, run independently—never sleep, emotionless, unaffected by a single loss.
   Previously, only those with technical expertise could set up such infrastructure. Now, OpenClaw lowers the barrier to entry for ordinary users.
   You don’t need to code or build your own system—just give natural language commands, and AI will execute for you.
   The tools are here.
   The only question remaining: when will you start?