Day 86 of 90 | Thursday, January 22, 2026

4 days remaining in our journey to build a profitable AI trading system.

Today was crisis mode. We discovered a critical Alpaca API bug that prevented us from closing positions, combined with PDT restrictions that locked our $5K account. Here’s how we navigated it.

The Hard Lessons

These are the moments that test us. Critical issues that demanded immediate attention.

LL-291: Alpaca API Bug - Close Position Treated as New Short (CRITICAL)

When attempting to close a LONG put position (SPY260220P00658000, 8 contracts), Alpaca’s API returned:

  • Error: “insufficient options buying power for cash-secured put (required: $113,000, available: $2,607)”

The bug: Alpaca incorrectly treats SELL-to-close as a NEW short (cash-secured put), requiring massive collateral instead of simply closing the existing long position.

What we tried (ALL FAILED):

  1. Market order via Python SDK
  2. Limit order
  3. close_position() endpoint
  4. Direct REST API with position_intent=’sell_to_close’
  5. DELETE /v2/positions/{symbol}
  6. close_all_positions()
  7. Partial close (1 contract)
  8. Account config: closing_transactions_only=True
  9. Account config: pdt_check=’exit’

Resolution: Pivoted to $100K paper account (PDT-enabled, $268K buying power) and successfully placed iron condor:

  • Put spread: Sell 660, Buy 655 @ $0.43 credit
  • Call spread: Sell 720, Buy 725 @ $0.38 credit
  • Total credit: $81/contract, Max risk: $419

Key takeaway: Always have a backup account. PDT-enabled accounts (>$25K) avoid the day-trading trap.

SOFI Loss Realized - Jan 14, 2026

  1. SOFI stock + CSP opened Day 74 (Jan 13)

Key takeaway: System allowed trade despite CLAUDE.

SOFI Position Held Through Earnings Blackout

SOFI CSP (Feb 6 expiration) was held despite Jan 30 earnings date approaching.

Key takeaway: Put option loss: -$13.

Important Discoveries

Not emergencies, but insights that will shape how we trade going forward.

Trade Data Source Priority Bug - Webhook Missing Alpaca Data

Status: FIXED

Iron Condor Win Rate Improvement Research

Current win rate is 33.3% (2/6 trades) vs target 80%+. Need to improve.

Iron Condor Entry Signals & Timing

System not generating enough trade signals. Need clear entry criteria.

Quick Wins & Refinements

  • Memgraph Graph Database Evaluation - FLUFF - LL-267: Memgraph Graph Database Evaluation - FLUFF

Date: January 21, 2026 Category: RAG / Resource …

  • Deep Operational Integrity Audit - 14 Issues Found - LL-240: Deep Operational Integrity Audit - 14 Issues Found

Date January 16, 2026 (Friday, 6:00 PM …

  • Phil Town Valuations - December 2025 - This lesson documents Phil Town valuations generated on December 4, 2025 during the $100K paper trad…
  • Theta Scaling Plan - December 2025 - This lesson documents the theta scaling strategy from December 2, 2025 when account equity was $6,00…

Today’s Numbers

What Count
Lessons Learned 10
Critical Issues 3
High Priority 3
Improvements 4

Crisis Summary

  • $5K Account: LOCKED (PDT + API bug) - 4 positions trapped
  • $100K Account: Active - Iron condor placed, $81 credit collected
  • Lesson: PDT-enabled accounts (>$25K) are essential for options trading

Tech Stack Behind the Lessons

Every lesson we learn is captured, analyzed, and stored by our AI infrastructure:

flowchart LR subgraph Learning["Learning Pipeline"] ERROR["Error/Insight
Detected"] --> CLAUDE["Claude Opus
(Analysis)"] CLAUDE --> RAG["Vertex AI RAG
(Storage)"] RAG --> BLOG["GitHub Pages
(Publishing)"] BLOG --> DEVTO["Dev.to
(Distribution)"] end

How We Learn Autonomously

Component Role in Learning
Claude Opus 4.5 Analyzes errors, extracts insights, determines severity
Vertex AI RAG Stores lessons with 768D embeddings for semantic search
Gemini 2.0 Flash Retrieves relevant past lessons before new trades
OpenRouter (DeepSeek) Cost-effective sentiment analysis and research

Why This Matters

  1. No Lesson Lost: Every insight persists in our RAG corpus
  2. Contextual Recall: Before each trade, we query similar past situations
  3. Continuous Improvement: 200+ lessons shape every decision
  4. Transparent Journey: All learnings published publicly

Full Tech Stack Documentation

The Journey So Far

We’re building an autonomous AI trading system that learns from every mistake. This isn’t about getting rich quick - it’s about building a system that can consistently generate income through disciplined options trading.

Our approach:

  • Paper trade for 90 days to validate the strategy
  • Document every lesson, every failure, every win
  • Use AI (Claude) as CTO to automate and improve
  • Follow Phil Town’s Rule #1: Don’t lose money

Want to follow along? Check out the full project on GitHub.

Day 86/90 complete. 4 to go.