# AlphaForge x K2 Think V2 — Institutional Quant Platform

A **10-module institutional-grade quantitative trading system** built for the **Build with K2 Think V2 Challenge** by MBZUAI.

**Live Space:** https://huggingface.co/spaces/Premchan369/alphaforge-k2think

## 10 Modules

| Module | Feature | Institution Equivalent |
|--------|---------|------------------------|
| 📈 Technical | 18+ indicators, candlestick, Bollinger, Ichimoku, VWAP, ADX, MFI, OBV | Bloomberg Terminal |
| ⚡ Backtest | 5 strategies, ATR sizing, equity curves, drawdown, trade logs | Jane Street backtest engine |
| 💼 Portfolio | Markowitz MPT, 10K-portfolio MC, efficient frontier, Sharpe max | AQR, D.E. Shaw |
| 📐 Options | Black-Scholes + full Greeks (ΔΓΘVΡ), scenario P/L, 6-panel charts | Goldman Sachs derivatives desk |
| 🔗 Pairs | Cointegration, OLS hedge ratio, OU half-life, Z-score signals | Two Sigma stat arb |
| 🪙 Crypto Arb | Cross-exchange spread heatmap, funding rate concepts | Jump Trading |
| 🛡️ Risk Engine | VaR 95/99, stress testing, correlation breakdown | Bridgewater risk |
| 📰 Sentiment | Composite gauge (-100 to +100), keyword extraction | Citadel NLP |
| 🌍 Macro | Cross-asset regime dashboard (S&P, 10Y, Gold, Oil, DXY, BTC) | Bridgewater All Weather |
| 🤖 K2 AI | Chain-of-thought: entry/stop/target, catalyst calendar, contrarian | MBZUAI K2 Think V2 |

## Setup

1. Add `K2_API_KEY` in Space Settings → Repository Secrets
2. No key required — all quant modules work standalone

## Stack

- yfinance / synthetic fallback (market data)
- Plotly (Bloomberg Terminal aesthetic)
- NumPy/Pandas (vectorized quant math)
- K2 Think V2 (MBZUAI reasoning)

## Data Notice

When Yahoo Finance rate-limits shared IPs (common on HF Spaces), the app automatically falls back to **deterministic synthetic data** seeded by ticker + date. Same ticker = same realistic data on the same day. All indicators, backtests, and optimizations work normally.

---
*Built by Premchan | Build with K2 Think V2*