How to Use AI Assistant to Create a Trading Strategy

What the AI Can Do

The assistant understands trading concepts, indicator names, strategy logic, and market terminology. It can:

• Set up entry conditions with specific indicators and parameters
• Configure exit rules (stop loss, take profit, trailing stop, breakeven)
• Select instruments and timeframes
• Combine multiple conditions with AND logic
• Adjust existing strategy parameters based on your feedback
• Explain what each indicator does and suggest reasonable parameters

What the AI Cannot Do

By design, the AI never provides trading recommendations. It will not:

• Tell you what to trade or when to trade
• Generate strategies on its own without your input
• Recommend specific parameter values as “optimal”
• Interpret backtest results as buy/sell signals
• Predict future market movements

This boundary exists by design. The AI is a translator, not an advisor. You bring the trading hypothesis; the AI converts it into a machine-testable format.

Prompt Engineering for Trading AI

The quality of the AI's output depends directly on the clarity of your input. Vague prompts produce generic setups; precise prompts produce exactly what you envision. Three principles make your prompts more effective:

Be specific about indicators and parameters. Instead of “add a moving average,” say “add EMA(50) crossover above EMA(200).” Including the period, type, and comparison target eliminates guesswork.

State the logic explicitly. “Buy when RSI is oversold and MACD turns positive” translates cleanly into two conditions. “Make something that catches reversals” forces the AI to guess your indicators and risk tolerance.

Separate entry from exit. Describe entry conditions in one message and exit rules in the next. This lets you verify configurator updates at each step.

The Conversation Workflow

Phase 1: Initial Setup

Start with the big picture: “I want to test a swing trading strategy on ETH, daily timeframe, using momentum indicators.” The AI will set instrument, exchange, and timeframe, then ask about entry conditions.

Phase 2: Refinement

Iterate on specifics: “Change RSI threshold from 30 to 25,” or “Add Bollinger Band upper band touch as an additional exit condition.” Each message adjusts the live configurator — you see changes in real-time on the left panel.

Phase 3: Exit Rules

Define your risk management: “Set 4% stop loss, 12% take profit, and enable trailing stop at 2.5 ATR.” The AI handles the configuration and warns you about conflicts (like setting both trailing stop and breakeven simultaneously).

Iterating on AI Suggestions

The first configuration the AI produces is rarely the final version. Effective iteration follows a test–review–refine cycle:

Step 1 — Run the initial backtest. Let the AI set up your idea, then run it. Examine win rate, profit factor, and drawdown.

Step 2 — Identify the weakness. Low win rate? Ask the AI to add a trend filter. Excessive drawdown? Tell it to tighten the stop loss. Too few trades? Request a lower indicator threshold.

Step 3 — Change one variable at a time. Asking “lower RSI from 30 to 25 and switch to 1H and add volume filter” changes three things simultaneously. If results improve, you won't know which change mattered.

Step 4 — Compare versions. Each backtest is saved with its full configuration. After 3–5 iterations, compare results side by side to identify which adjustments improved performance.

Step-by-Step Example

Here is a complete conversation flow that builds a strategy from scratch:

You: “I want to trade BTC/USDT on the 4H timeframe. Buy when price is above EMA(200) and RSI(14) drops below 35.”
AI: Sets instrument to BTC/USDT, timeframe to 4H, adds two entry conditions. The configurator updates in real time.

You: “Add a third condition: MACD histogram must be positive.”
AI: Adds MACD histogram > 0 as condition 3. Three conditions now visible in the left panel.

You: “Set stop loss at 3% and take profit at 9%. Enable trailing stop at 2%.”
AI: Configures exit rules, confirms that trailing stop and breakeven are mutually exclusive, applies trailing stop.

You run the backtest, review results, then return to chat: “Win rate is low. Change RSI threshold from 35 to 30.” The AI updates the condition. You run again and compare.

Pro Tips

One change at a time: Don't try to describe your entire strategy in one message. Start with basics, run a backtest, then iterate. The AI maintains context across the conversation.

Use indicator names: Say “RSI(14)” not “that overbought indicator.” Specific names with parameters get exact results.

Reference the configurator: “Change the second entry condition to use EMA instead of SMA” — the AI understands references to existing configuration.

FAQ

Further Reading

• RSI on Investopedia
• MACD on Investopedia
• Bollinger Bands on Investopedia