Prompt Engineering for Trading: Getting Better Results from AI

import requests

API_KEY = "YOUR_API_KEY"
BASE_URL = "https://tickatlas.com/v1"

def gather_market_context(symbol: str) -> str:
    """Fetch multi-indicator data and format as prompt context."""
    indicators = ["RSI_14", "MACD_hist", "EMA_50", "SMA_200", "ATR_14", "Stochastic_K"]
    results = {}

    for ind in indicators:
        resp = requests.get(
            f"{BASE_URL}/indicator",
            headers={"X-API-Key": API_KEY},
            params={"symbol": symbol, "indicator": ind, "timeframe": "H4"},
        )
        if resp.status_code == 200:
            results[ind] = resp.json()["data"]

    # Format as structured text for the LLM
    context = f"## Market Data for {symbol} (H4 Timeframe)\n\n"
    for ind, data in results.items():
        context += f"- {ind}: {data['value']}\n"

    quote = results.get("RSI_14", {})
    if quote:
        context += f"\nCurrent Bid: {quote.get('bid')}\n"
        context += f"Current Ask: {quote.get('ask')}\n"

    return context

context = gather_market_context("EURUSD")
print(context)

ANALYSIS_PROMPT = """
You are a senior technical analyst with 20 years of forex experience.
You make decisions based solely on the indicator data provided.
You never fabricate data or reference information not in the context.

{market_data}

## Task
Analyze the above data and provide:
1. Current trend direction (bullish/bearish/neutral) with reasoning
2. Key support and resistance levels implied by the indicators
3. Trade setup if one exists (entry, stop loss, take profit)
4. Confidence level (low/medium/high) with justification
5. What would change your analysis (invalidation conditions)

## Output Format
Respond in JSON with keys: trend, support_resistance, trade_setup,
confidence, invalidation.
If no trade setup exists, set trade_setup to null.
"""

import json
from openai import OpenAI

client = OpenAI()

def analyze_market(symbol: str) -> dict:
    """Get AI analysis based on real indicator data."""
    context = gather_market_context(symbol)

    response = client.chat.completions.create(
        model="gpt-4o",
        messages=[
            {"role": "system", "content": "You are a technical analyst. Respond only in valid JSON."},
            {"role": "user", "content": ANALYSIS_PROMPT.format(market_data=context)},
        ],
        temperature=0.2,  # Low temperature for consistent analysis
        response_format={"type": "json_object"},
    )

    result = json.loads(response.choices[0].message.content)

    # Validate required keys
    required_keys = ["trend", "confidence", "trade_setup"]
    for key in required_keys:
        if key not in result:
            raise ValueError(f"Missing required key: {key}")

    return result

analysis = analyze_market("EURUSD")
print(json.dumps(analysis, indent=2))

def multi_timeframe_context(symbol: str) -> str:
    timeframes = {"D1": "Daily trend", "H4": "Swing context", "H1": "Entry timing"}
    context = f"# Multi-Timeframe Analysis: {symbol}\n\n"

    for tf, purpose in timeframes.items():
        resp = requests.get(
            f"{BASE_URL}/indicators",
            headers={"X-API-Key": API_KEY},
            params={
                "symbol": symbol,
                "timeframe": tf,
            },
        )
        data = resp.json()["data"]
        context += f"## {tf} ({purpose})\n"
        for ind_name, ind_value in data["indicators"].items():
            context += f"- {ind_name}: {ind_value}\n"
        context += "\n"

    return context