Back to Projects

Project Artha — SPY Options Simulation

Live

Rules-based SPY options paper-trading pipeline that runs twice daily, recommends defined-risk setups, and maintains a structured learning journal. One contract, no real money.

Python yfinance launchd Black-Scholes pandas
❌

The Problem

❌ Learning SPY options by reading about them builds familiarity but not judgment. You need reps — but real money creates emotional noise that distorts learning.

💡

The Approach

💡 Simulate one contract daily. Rules-based recommender picks the setup. Human can override. Every decision and outcome is logged with a structured 4-step journal entry.

✅

The Solution

✅ Two automated runs daily (9 AM and 12:15 PM PT). Recommender handles IV rank, catalyst buffers, trend filters, and strike selection. Journal tracks system vs human decisions over time.

What

Project Artha (Sanskrit for wealth and purpose) is a paper-trading simulation for SPY options. It runs twice per day via launchd, fetches live market data, recommends one defined-risk options setup (or no trade), and records the decision in a structured learning journal.

One contract. No real money. Goal: build options intuition through systematic simulation and deliberate journaling.

The Recommender

A rules-based engine that gates every trade:

Filters (in order): 1. IV Rank < 25 → NO TRADE (options are cheap, not worth selling) 2. DTE outside 1–45 day window → NO TRADE 3. CPI / FOMC / NFP within 4 days → NO TRADE (catalyst buffer) 4. ATR elevated + SPY extended upward → Bear call spread 5. ATR elevated + SPY extended downward → Bull put spread 6. Default → Iron condor

Strike selection — the critical rule: - DTE > 7: use 16-delta (standard) - DTE ≤ 7: use 1.5× implied move from the ATM straddle price

At short DTE, 16-delta places strikes inside the expected range. The ATM straddle price (call_mid + put_mid) is the market’s honest assessment of the expected ±move. Multiplying by 1.5 places strikes safely outside it.

The Journal

Each trade gets a 4-step entry: 1. Evaluated setups — all candidates considered with metrics 2. Trade setup — legs, credit, max loss, visual range bar 3. Outcome — filled after expiry (actual P&L) 4. Lesson learned — manual entry (what this confirmed or challenged)

Human Override Tracker — records when the system recommended a trade but the human passed, and vice versa. Over time this surfaces where human judgment adds vs destroys value.

Technical Details

  • CPI calendar uses hardcoded BLS release dates (formula-based calculation was unreliable — BLS doesn’t always release on the 2nd Wednesday)
  • Same-day morning releases (CPI at 8:30 AM ET) are treated as already-past by the 9 AM PT pipeline run — no blocking
  • FOMC decisions (2 PM ET) still block on same-day runs since pipeline runs before them
  • Pre-close run (12:15 PM PT) evaluates positions expiring today against 50% profit target, breakeven breach, or gamma risk proximity
Back to Projects