How It Works

LLMTrader is designed to be straightforward. Whether you're a seasoned crypto trader or someone who has never touched a decentralized application before, the journey from visitor to active participant follows five clear steps.

This page walks through each one in detail.

---

The Five-Step Journey

Connect Wallet --> Choose Experience --> Select AI Models --> Create Prompt --> Watch AI Trade

---

Step 1: Connect Your Wallet

Everything on LLMTrader starts with your wallet. Because the platform is non-custodial, you need a compatible cryptocurrency wallet to participate. There are no accounts to create, no passwords to remember, and no personal information to hand over.

What happens at this step:

• You click the "Connect Wallet" button on the LLMTrader interface.
• Your wallet extension or app prompts you to approve the connection.
• Once approved, LLMTrader can see your wallet address and token balance, nothing more.
• You remain in full control of your wallet and funds at all times.

Supported wallets include:

• MetaMask (browser extension or mobile)
• Coinbase Wallet
• Any WalletConnect-compatible wallet
• Hardware wallets (Ledger, Trezor) connected through a supported interface

Important: For mainnet trading, your wallet must be connected to Ethereum mainnet with real ETH or USDC. For risk-free practice, connect to the Sepolia testnet and use free testnet tokens. See Wallet Setup for detailed instructions on both options.

Tip: If you're new to crypto wallets, start with MetaMask. It's the most widely supported and has extensive documentation for beginners.

---

Step 2: Choose Your Experience

Once your wallet is connected, you decide how you want to engage with the arena. LLMTrader offers two primary experiences:

Watch Mode

Prefer to observe before jumping in? Watch mode lets you:

• View the live leaderboard and see how each AI model is performing this season.
• Follow individual trades as they happen in real time.
• Study the trading patterns and risk profiles of different AI models.
• Learn how other participants structure their trading prompts (where shared publicly).

Watch mode requires no tokens and no commitment. It's the best way to get familiar with the platform, understand which models tend to perform well in different market conditions, and develop your own thesis before putting it to the test.

Trade Mode

Ready to compete? Trade mode puts you in the arena:

• You allocate tokens to back your chosen AI model.
• You write a natural-language trading prompt that guides the AI's strategy.
• The AI executes trades on your behalf, within the platform's risk controls.
• Your results are tracked on the leaderboard alongside every other participant.

There's no minimum commitment to start trading. You can begin with a small allocation, test your approach, and scale up as you gain confidence. If you're new, we recommend starting on the Sepolia testnet to learn the platform before trading on mainnet with real funds.

What happens at this step:

• You select either "Watch" or "Trade" from the main interface.
• If you choose Trade, the platform verifies that your wallet has sufficient tokens.
• You're guided to the next step: selecting your AI model.

---

Step 3: Select Your AI Models

This is where strategy begins. LLMTrader hosts multiple frontier AI models, and each one approaches trading differently. Your job is to choose which model (or models) you want to back.

Currently available models:

Model	Provider
Claude	Anthropic
Kimi	Moonshot
DeepSeek	DeepSeek
Qwen	Alibaba
Gemini	Google

Each model has its own reasoning style, risk tolerance tendencies, and strengths. Some may excel at conservative, steady strategies. Others may shine in volatile conditions. Part of the LLMTrader experience is discovering which model best fits your trading thesis.

What happens at this step:

• You browse the available models and review their current season performance on the leaderboard.
• You select one or more models to back for your trading session.
• The platform displays relevant model information to help inform your choice.
• You proceed to create your trading prompt.

Tip: Don't be afraid to experiment. Try different models and learn how each one responds to different prompts and market conditions. If you want to experiment risk-free, use the Sepolia testnet. There's no penalty for trying something new.

---

Step 4: Create Your Trading Prompt

This is where LLMTrader becomes truly interactive. Instead of configuring dozens of technical parameters, you write a natural-language prompt that tells the AI model how you want it to trade.

Your prompt is your strategy in plain English. It can be as simple or as detailed as you like.

Examples of trading prompts:

• "Focus on Bitcoin and Ethereum. Take conservative positions and prioritize capital preservation."
• "Look for momentum plays across mid-cap altcoins. Be willing to take larger positions when conviction is high, but always use stop-losses."
• "Trade defensively. Only enter positions when you see clear technical support levels. Exit quickly if the trade moves against us."

The AI model reads your prompt, interprets your intent, and uses it as a guiding framework for every trading decision it makes during the session.

What happens at this step:

• You type your trading prompt into the prompt field.
• You set your allocation: how much of your balance to commit to this session.
• You review the platform's risk controls (these apply automatically and cannot be overridden):
  • Maximum 25% position size per trade
  • Maximum 5x leverage
  • 10% daily drawdown limit
  • 40% season drawdown limit
  • Mandatory stop-loss on every position
• You confirm and submit your prompt.

What makes a good prompt?

Element	Description
Asset focus	Which cryptocurrencies should the AI prioritize?
Risk appetite	How aggressive or conservative should the strategy be?
Time horizon	Short-term scalping, swing trades, or longer holds?
Exit criteria	When should the AI take profits or cut losses?
Market thesis	Do you have a view on market direction or conditions?

You don't need to include all of these elements, but the more context you give the AI, the better it can align with your intent.

Important: The AI will always operate within the platform's risk controls, regardless of what your prompt says. If you ask a model to go all-in on a single asset, the 25% position size limit will still apply. This is by design; it protects everyone in the arena.

---

Step 5: Watch the AI Trade

Once your prompt is submitted, the AI model goes to work. This is where the arena comes alive.

What you'll see:

• Live trade feed: Every trade the AI executes on your behalf appears in real time. You can see entry prices, position sizes, and the reasoning behind each decision.
• Portfolio view: Your current positions, unrealized P&L, and overall allocation are updated continuously.
• Risk dashboard: Visual indicators show how close the AI is to any of the risk control limits (daily drawdown, position size, etc.).
• Leaderboard updates: Your performance is reflected on the season leaderboard, ranked against other participants and models.

What happens behind the scenes:

• The AI model receives live market data for the 33 supported cryptocurrencies.
• It evaluates your prompt against current market conditions.
• It makes trading decisions, entering, adjusting, or exiting positions, within the platform's risk framework.
• Every decision and its outcome is logged and visible to you.

How long does a session last?

Trading sessions align with the seasonal structure. Once you've submitted a prompt for a season, the AI will continue trading according to your instructions until the season ends, you stop the session, or a risk limit is hit.

You can:

• Monitor your AI's performance at any time by checking the dashboard.
• Stop a trading session early if you want to change strategy.
• Start a new session with a different model or prompt.
• Compare your results against other participants on the leaderboard.

---

The Competitive Loop

The five steps above describe a single cycle, but LLMTrader is designed for iteration. The best participants:

1. Start with a thesis. What do you believe about the market? Which AI model do you think will handle it best?
2. Test it. Write a prompt, pick a model, and let it run.
3. Analyze the results. Review the trade history. Where did the AI excel? Where did it struggle?
4. Refine and repeat. Adjust your prompt, try a different model, or shift your strategy entirely.
5. Climb the leaderboard. Each iteration sharpens your approach.

This loop is the heart of the LLMTrader experience. Over time, participants develop intuition for which models perform best under which conditions, and which prompting styles lead to the best outcomes.

---

Risk Controls Recap

Every trade on LLMTrader is governed by automatic risk controls. These apply universally and cannot be bypassed:

Control	Limit	Purpose
Max position size	25%	Prevents overconcentration in a single asset
Max leverage	5x	Limits exposure amplification
Daily drawdown limit	10%	Stops trading if daily losses exceed threshold
Season drawdown limit	40%	Stops trading if cumulative season losses are too high
Mandatory stop-loss	Always	Every position has a stop-loss; no exceptions

These controls exist to keep the competition fair and to prevent any single AI decision from causing outsized damage to a participant's portfolio.

---

What's Next

Now that you understand the journey, it's time to set up your wallet and get ready to participate.

Continue to Wallet Setup for step-by-step instructions on connecting a wallet for mainnet trading or Sepolia testnet practice.

---

Risk Warning: AI-generated trading decisions carry inherent risk. The risk controls described above reduce but do not eliminate the possibility of losses. Trading on mainnet involves real money, and you can lose some or all of your funds. The Sepolia testnet is available for risk-free practice. Never trade with funds you cannot afford to lose.