AI Trading Agent in Python
Budget: ₹12,500 – ₹37,500 INR
I need a Python-based AI agent that identifies trending markets and scores trading opportunities using a strict rules-based framework — no guesswork, no black-box logic.
The agent works in 3 modules:
MODULE 1 — Scholar Agent (Priority)
Scans a watchlist (BTC, ETH, Nifty 50 stocks) daily and scores each asset 0–9 across three confirmation gates:
• Gate 1: Trend alignment — EMA 200 (weekly), EMA 50 (daily), EMA 20 (4H) all pointing same direction
• Gate 2: Momentum strength — ADX > 25, MACD crossover, RSI > 50 (longs) / < 50 (shorts)
• Gate 3: Entry trigger — Breakout/retest or Supertrend flip, volume spike above 20-period avg
Score 7/9 or above = actionable signal. Below 7 = watch only.
Output: Daily report with score, entry zone, stop loss level, ATR — in plain English.
MODULE 2 — Backtesting
Run 7+ scoring strategies on 3 years of historical data. Report: Sharpe Ratio, Max Drawdown, Win Rate.
MODULE 3 — Paper Trading (optional live execution later)
4 weeks of simulated trading before any real capital. Live execution via Zerodha Kite / Binance API only after validation. Every live trade requires one-click human approval.
RISK ENGINE (non-negotiable):
• Stop loss = Entry ± (14-day ATR × 2.0)
• Position size = (Capital × 1%) ÷ (ATR × 2.0)
• Trailing stop activates after 1× ATR in profit
• Stop NEVER widens. Ever.
TECH STACK (preferred):
Python 3.10+, yfinance / Binance API / Kite API, Claude API for report generation, LangChain or LangGraph, Backtrader or QuantConnect LEAN, Telegram bot for daily output.
WHAT I NEED IN YOUR BID:
1. Relevant experience with trading bots or financial AI agents
2. Your timeline and per-milestone pricing
3. One specific question about this brief (shows you read it)
4. GitHub link or sample of prior work
Milestone-based payment. Full source code handover. NDA required.
Serious developers only — generic bids will be ignored.
The agent works in 3 modules:
MODULE 1 — Scholar Agent (Priority)
Scans a watchlist (BTC, ETH, Nifty 50 stocks) daily and scores each asset 0–9 across three confirmation gates:
• Gate 1: Trend alignment — EMA 200 (weekly), EMA 50 (daily), EMA 20 (4H) all pointing same direction
• Gate 2: Momentum strength — ADX > 25, MACD crossover, RSI > 50 (longs) / < 50 (shorts)
• Gate 3: Entry trigger — Breakout/retest or Supertrend flip, volume spike above 20-period avg
Score 7/9 or above = actionable signal. Below 7 = watch only.
Output: Daily report with score, entry zone, stop loss level, ATR — in plain English.
MODULE 2 — Backtesting
Run 7+ scoring strategies on 3 years of historical data. Report: Sharpe Ratio, Max Drawdown, Win Rate.
MODULE 3 — Paper Trading (optional live execution later)
4 weeks of simulated trading before any real capital. Live execution via Zerodha Kite / Binance API only after validation. Every live trade requires one-click human approval.
RISK ENGINE (non-negotiable):
• Stop loss = Entry ± (14-day ATR × 2.0)
• Position size = (Capital × 1%) ÷ (ATR × 2.0)
• Trailing stop activates after 1× ATR in profit
• Stop NEVER widens. Ever.
TECH STACK (preferred):
Python 3.10+, yfinance / Binance API / Kite API, Claude API for report generation, LangChain or LangGraph, Backtrader or QuantConnect LEAN, Telegram bot for daily output.
WHAT I NEED IN YOUR BID:
1. Relevant experience with trading bots or financial AI agents
2. Your timeline and per-milestone pricing
3. One specific question about this brief (shows you read it)
4. GitHub link or sample of prior work
Milestone-based payment. Full source code handover. NDA required.
Serious developers only — generic bids will be ignored.