MCMC Options Trading System

Monte Carlo Markov Chain trading system for options. Uses regime-switching simulations to generate probabilistic price forecasts and high-probability options signals.

Quick Start

# Install dependencies
pip install -r requirements.txt

# Get today's options signals
python scripts/options_now.py

# Run options backtest
python scripts/backtest_options_v4.py

---

How to Use the Strategy

The System

1. MCMC simulates 25,000 price paths to predict direction
2. Filters for high-probability setups (strength, slope, momentum, regime)
3. Outputs ranked options signals with confidence scores

Your Weekly Routine

Day	Action
Monday AM	Run python scripts/options_now.py
If signal 70+	Buy ATM option, ~35 DTE
Set alerts	+80% take profit, -35% stop loss
Wait	Let it play out

Position Sizing

Rule	Value
Position size	10% of portfolio per trade
Max positions	3-4 open at once
Strike	ATM (at the money)
Expiration	30-40 DTE

Exit Rules

Condition	Action
Option up +80%	Sell - Take Profit
Option down -35%	Sell - Stop Loss
30+ days held	Sell - Time Exit

Example ($10,000 Account)

Monday:
  1. Run: python scripts/options_now.py
  2. Output: "HD $339 PUT - Confidence 72/100"
  3. Buy 2x HD $339 PUT @ $4.80 = $960
  4. Set alerts: TP at $8.64 (+80%), SL at $3.12 (-35%)
  5. Done. Check again next Monday.

Outcome A: HD drops, option hits $9.00 → Sell → +87% win
Outcome B: HD rises, option hits $3.00 → Sell → -37% loss
Outcome C: 30 days pass → Sell at market → Variable

Key Rules

• Only buy on Monday/Tuesday (avoid weekend theta decay)
• ATM strikes only (higher probability than OTM)
• Same dollar amount per trade (not same # of contracts)
• Max 4 positions at once
• Different tickers (don't stack same stock)

---

Commands

Command	Description
python scripts/options_now.py	Get today's signals with confidence scores
python scripts/options_signal_v4.py	Conservative - only shows signals in clear regimes
python scripts/backtest_options_v4.py	Run full backtest (2022-2025)
python scripts/get_signals.py	Stock signals (not options)
python scripts/run_backtest.py	Stock backtest

---

Understanding the Output

#1 HD PUT | Confidence: 72/100 (HIGH)
   Strike: $339 ATM | Exp: ~Apr 17
   Price: $338.93 | Vol: 23%
   Strength: 0.78 | Slope: -60.1
   5d: -4.7% | 20d: -12.6%

Field	Meaning
Confidence	70+ = HIGH (take it), 50-69 = MEDIUM (caution), <50 = LOW (skip)
Strength	MCMC signal strength (0.72+ required)
Slope	Trend steepness in degrees (20+ required)
Vol	Stock's volatility (20-85% range required)
5d/20d	Recent momentum confirmation

---

System Architecture

scripts/
  options_now.py         # Main signal generator (use this)
  options_signal_v4.py   # Conservative signal generator
  backtest_options_v4.py # Options backtester
  get_signals.py         # Stock signals

trading/
  indicator.py           # MCMCIndicator - core signal engine

config/
  tickers.yaml           # Watchlist (50+ stocks)
  best_params.json       # Optimized parameters

---

How It Works (Technical)

Signal Generation

1. Monte Carlo Simulation - 25,000 price paths using historical volatility
2. Slope Calculation - Forecast median vs current price, normalized
3. Regime Detection - SPY determines bull/bear/neutral market
4. Multi-Filter - Strength >= 0.72, Slope >= 20°, momentum confirmation

Regime Filter

SPY Condition	Regime	Allowed Trades
Above 50 & 200 MA + positive momentum	BULL	CALLs only
Below 50 & 200 MA + negative momentum	BEAR	PUTs only
Mixed signals	NEUTRAL	Lower confidence

Why It Works

• Momentum continuation - We bet trends continue, not reverse
• Multiple filters - Only ~1-2 trades per week qualify
• Asymmetric payoff - +80% wins vs -35% losses = profitable at 45% win rate
• Regime alignment - Trade with the market, not against it

---

Backtest Results by Period

Period	Trades	Win Rate	Profit Factor	Return
2022 Bear	7	29%	1.34	+5%
2023 Bull	7	29%	1.70	+12%
2024 Full	12	67%	6.84	+119%
2025 YTD	14	86%	23.65	+336%

---

Risk Warning

• Past backtest results do not guarantee future performance
• Options can expire worthless (100% loss on position)
• Only trade money you can afford to lose
• Consider paper trading first to learn the system

---

Configuration

Key Parameters (config/best_params.json)

Parameter	Value	Description
n_simulations	25,000	Monte Carlo paths
slope_threshold	15.0	Degrees for signal
signal_strength_min	0.72	Minimum to trade
position_size_pct	0.10	10% per trade

Ticker Universe (config/tickers.yaml)

50+ liquid stocks including:

• Mega caps: AAPL, MSFT, GOOGL, AMZN, META, NVDA, TSLA
• Tech: AMD, NFLX, CRM, ADBE, AVGO
• Finance: JPM, BAC, GS, V, MA
• Consumer: HD, COST, NKE, DIS
• And more...

---

GPU Acceleration

Uses CuPy for CUDA-accelerated simulations when available:

• CPU: ~1,000 simulations (NumPy)
• GPU: ~50,000 simulations (CuPy/CUDA)

Falls back to NumPy automatically if no GPU detected.

---

License

MIT License