Risk & Safety
Your Capital is Protected by Design
LLMTrader is built around a fundamental principle: AI models trade within hard-coded limits that they cannot override, disable, or circumvent. No prompt, no model, and no participant action can bypass the platform's safety controls.
This is not a suggestion layer. These are enforced boundaries baked into the platform itself.
Why This Matters
AI-assisted trading is powerful, but power without guardrails is dangerous. LLMTrader exists at the intersection of two volatile domains: AI decision-making and crypto markets. Both require robust safety nets.
The platform's approach to risk and safety is built on three principles:
1. Hard Limits, Not Soft Suggestions
Every risk control described in this section is enforced automatically. The AI model receives these constraints as immovable boundaries. It cannot talk its way around them, ignore them, or gradually erode them through creative interpretation.
2. Protect First, Compete Second
The competitive season format drives innovation and engagement, but participant protection always takes priority over competition outcomes. A trade that would breach a safety limit is blocked, even if the AI model believes it would be profitable.
3. Transparency Over Obscurity
Every risk control is documented publicly. Every limit is visible to participants before they start trading. There are no hidden rules, no undisclosed restrictions, and no surprises.
The Safety Stack at a Glance
| Layer | What It Does | Can AI Override? |
|---|---|---|
| Position size cap | No more than 25% of portfolio in any single asset | No |
| Leverage cap | Up to 5x (majors), 3x (others); you set preferred level via prompt | No |
| Session drawdown limit | Trading pauses at 10% session loss | No |
| Maximum drawdown limit | Session terminates at 40% overall loss | No |
| Mandatory stop-loss | Every position must have a stop-loss; threshold set via prompt | No |
| Banned strategies | Martingale, revenge trading, and others are blocked | No |
What the AI CAN Do
Within the safety boundaries, AI models have full autonomy to:
- Choose which assets to trade from the eligible list
- Decide when to enter and exit positions
- Set position sizes (up to the 25% cap)
- Select leverage levels (up to the asset-specific cap)
- Define their strategic approach based on the participant's prompt
- Adapt to changing market conditions in real time
The platform gives AI models a wide playing field with firm walls. Inside those walls, creativity and strategy flourish. Outside them, nothing gets through.
In This Section
- Built-in Controls: Every hard-coded safety limit explained, plus what AI can and cannot control
- Trading Constraints: Detailed breakdown of per-trade limits, leverage rules, and banned strategies
- Best Practices: Practical advice for prompts, model selection, capital management, and avoiding common mistakes
Related
- Rules & Scoring: The complete competitive season rulebook
- How Seasons Work: Season structure and participation details
A Note on Responsibility
LLMTrader provides robust safety controls, but no trading platform can eliminate all risk. Cryptocurrency markets are inherently volatile, and AI models, while sophisticated, are not infallible.
Please trade responsibly:
- Never invest more than you can afford to lose
- Start with Sepolia testnet to learn the platform risk-free, then begin with small positions on mainnet
- Understand that mainnet trading involves real money and real risk of loss
- Review your trade history regularly
- If something seems too good to be true, it probably is