# K2J AI Quantitative Lab

> K2J AI Quantitative Lab is an independent, institutional-standard daily quantitative research system for US equities. Designed, built, and operated end-to-end by Justin — an enterprise AI systems architect with 20+ years of experience and an active options/equities trader — using LangGraph multi-agent orchestration, Agentic RAG, deterministic guardrails, and real-time 13F/SEC regulatory filing surveillance.

## Core Quantitative Research Methodology

- [Market Regime Classification](https://k2j.net/#research): Multi-state macro risk classification using GBDT ensembles to detect risk-on, risk-off, and transitional phases before factor calculation.
- [Market Breadth Analysis](https://k2j.net/#research): Full-market cross-sectional participation metric measuring internal equity breadth beyond headline indices.
- [5-D Orthogonal Factor Attribution](https://k2j.net/#research): Five-dimensional cross-sectional statistical attribution using robust WLS and PCA to isolate true alpha drivers from noise.
- [Latent Theme Clustering](https://k2j.net/#research): High-dimensional mover clustering using HDBSCAN and dense semantic embeddings to identify emergent market narrative clusters.
- [13F & SEC Filing Surveillance / Smart Money Tracking](https://k2j.net/#research): Real-time ingestion and parsing of SEC EDGAR submissions (Form 13F quarterly holdings, 13D/13G beneficial ownership, Form 4 insider transactions, 8-K material events) to track institutional positioning and identify price-volume divergences.
- [Real-Time Financial News Intelligence](https://k2j.net/#research): 24/7 NLP surveillance across 70+ financial wire sources performing causal event-ticker relevance scoring, polarity decomposition, and tail-risk extraction.
- [Portfolio & Tail-Risk Optimization](https://k2j.net/#research): Dynamic portfolio stance modeling constrained by CVaR (Conditional Value-at-Risk) and Bayesian Risk Parity.

## AI Systems Architecture

- **Agentic Orchestration**: Autonomous multi-agent pipelines orchestrated via LangGraph.
- **Retrieval-Augmented Generation**: Agentic RAG with Pinecone vector database and hybrid dense/sparse indexing.
- **Reliability & Guardrails**: Deterministic guardrails preventing hallucination and enforcing financial data consistency.
- **Triple-Engine Resonance**: Synchronized alpha synthesis uniting price/volume factors, news intelligence, and smart money regulatory filings.
- **Production Cadence**: Rebuilt and published autonomously every trading morning before market open.

## Author & Background

- **Author**: Justin (AI Systems Architect & Quant Trader)
- **Experience**: 20+ years in enterprise technology architecture (Dell, Microsoft, IBM, private funds)
- **Profile & Contact**: Open to advisory, consulting, and full-time/part-time collaboration for teams seeking dual expertise in AI engineering and quantitative markets.
- **Email**: k2jysy@gmail.com
- **LinkedIn**: https://www.linkedin.com/in/justin-k2j

## Key Links & Resources

- [Official Website](https://k2j.net/): Homepage with methodology, interactive FAQ, and report archive.
- [Methodology & Reading Guide](https://k2j.net/guide.html): Comprehensive guide on the 8 data pillars (100+ metrics), top-down macro-to-stock analytical logic, 5-D attribution formula, 20 focus equities screening funnel, and sector divergence forensics.
  - [Chinese Reading Guide](https://k2j.net/guide.zh.html)
  - [English Reading Guide](https://k2j.net/guide.en.html)
- [Privacy Policy](https://k2j.net/privacy.html): Official data privacy and compliance policy.
  - [Chinese Privacy Policy](https://k2j.net/privacy.zh.html)
  - [English Privacy Policy](https://k2j.net/privacy.en.html)
- [Full LLM Knowledge Base](https://k2j.net/llms-full.txt): Complete detailed text representation for LLM context windows and deep retrieval.
- [Daily Reports Archive](https://k2j.net/#archive): Full repository of daily market quantitative analysis reports (Bilingual: English & Simplified Chinese) under `/reports/`.