How I Setup AI Investment Bot | Copy My Prompt Here

The first script I wrote wasn’t an alert engine — it was a performance analyser. Using quantstats and yfinance, I generated an HTML tearsheet comparing my portfolio against SPY.

import quantstats as qs

qs.reports.html(portfolio_returns, benchmark='SPY', output='tearsheet.html')

Key metrics: Sharpe ratio, Max Drawdown, Calmar ratio, rolling beta. A portfolio that returned 46% but with −40% max drawdown is not beating the market — it’s leveraged to luck.

Run this baseline first. Everything else is built on top of it.

Riskfolio-Lib computes optimal weights across four methods. The difference is dramatic:

The unoptimised version had a 40% max drawdown — concentration masquerading as skill. The optimiser runs monthly on GitHub Actions and auto-commits fresh target weights.

Score = number of signals fired simultaneously:

• Price < 50-day MA by a configurable threshold (e.g. −5%)
• Price < 200-day MA — deeper correction signal
• RSI(14) < 40 — technically oversold
• Price > 15% below 6-month high — significant drawdown
• Price < your average cost basis — natural averaging-down trigger

3–5 signals = STRONG BUY  ·  2 = BUY  ·  1 = MILD DIP

Multi-signal filtering eliminates the false positives that plague single-indicator systems.

One sector. Multiple moats. Far lower concentration risk than going all-in on a single name.

Five rules checked every morning:

RULES = {
    "max_single_position_pct": 15.0,   # alert if any stock > 15%
    "max_sector_pct":          40.0,   # alert if any sector > 40%
    "stop_loss_pct":          -30.0,   # alert if any position < -30%
    "min_position_value":      200.0,  # alert if position < $200
    "drift_threshold_pct":      5.0,   # alert if drift from target > 5%
}

The Telegram digest every morning:

Portfolio Advisory — 2026-05-14

BUY-THE-DIP ALERTS
  *** MSFT $408 — Below 200d MA, RSI 38, -20% off 6m high
      Suggest +$500 (current 14% → cap 15%)
  **  META $603 — Below 200d MA, -18% off high

Critical Alerts
  [CONCENTRATION] TSLA is 62.1% — REDUCE

Rebalance Suggestions
  AAPL: 3.2% vs target 12.0% — ADD 8.8%

Arrives in Telegram every weekday morning before you open your broker.

Every alert is stored with the price at alert time:

date, ticker, alert_type, score, price_now, suggested,
notes, your_action, price_30d, return_30d

One command backfills the outcome column after 30 days:

python journal.py --backfill   # fetches price 30 days after alert
python journal.py --summary    # win rate, avg return by alert type

Without this, there is no feedback loop. The journal is what turns an alert tool into an advisory system that improves over time.

on:
  schedule:
    - cron: '30 13 * * 1-5'   # weekdays at market open
permissions:
  contents: write              # needed to commit the journal back

GitHub free tier = 2,000 mins/month. This system uses ~110.

• Keep your conviction positions
• Let their weight dilute naturally as you add to others on dips
• Only add when signals fire — not on emotion
• Set a hard position cap per stock (e.g. max 12%)

The portfolio diversifies over time without a single forced sell decision.