README.md

metadata

title: Kronos + Laya Portfolio & Trade Analysis Engine
emoji: 🚀
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.20.0
app_file: app.py
pinned: false

🚀 Kronos Foundation Models + Laya Portfolio + Dual Trade-Analysis Engine

A 24/7/365 production quantitative intelligence and self-improving trading suite running natively on Hugging Face CPU Basic:

1. 📈 Kronos Foundation Models (Kronos-mini, Kronos-small, Kronos-base): Autoregressive K-line price path simulation with 5%, 50%, 95% quantile cones.
2. ⚡ Laya Portfolio & Risk Engine (System 1): Sub-50ms calibrated portfolio risk gating, dynamic stop loss/take profit levels, and capital preservation.
3. 🔬 Trade Execution Post-Mortem Analyst (Qwen/Qwen2.5-1.5B-Instruct): Grades closed trades (A+ to F), diagnoses statistical edge vs execution error vs macro shock, and produces calibrated rule adjustments for self-improving bots.
4. 📰 Financial Catalyst Correlator (ProsusAI/finbert): Sub-15ms macroeconomic news and headline sentiment analyzer scoring market bias (-1.0 to +1.0).

---

🌐 API Endpoints

1. POST /v1/predict-trade — Quantitative Forecast & Laya Risk Signal

curl -X POST https://devinblaze-laya-engine.hf.space/v1/predict-trade \
  -H "Content-Type: application/json" \
  -d '{
    "symbol": "BTC-USD",
    "model_size": "mini",
    "interval": "1h",
    "pred_len": 12,
    "paths_count": 10,
    "headline": "Fed signals potential rate pause in upcoming FOMC meeting"
  }'

2. POST /v1/analyze-trade — Dual-Model Trade Diagnosis & Self-Improvement

Sent automatically by your trading bot after each closed trade to grade execution and recalibrate strategy parameters:

curl -X POST https://devinblaze-laya-engine.hf.space/v1/analyze-trade \
  -H "Content-Type: application/json" \
  -d '{
    "symbol": "BTC-USD",
    "trade_type": "LONG",
    "entry_price": 85000.0,
    "exit_price": 83400.0,
    "pnl_pct": -1.88,
    "duration_hours": 4,
    "kronos_q5": 84100.0,
    "kronos_q50": 86200.0,
    "kronos_q95": 87500.0,
    "macro_news": "Hot CPI inflation release caused sudden flash dump"
  }'

3. POST /v1/chat/completions — OpenAI Compatible Endpoint

High-speed CPU-optimized quantitative inference for OmniRoute, Hermes Agent, and custom tools:

curl -X POST https://devinblaze-laya-engine.hf.space/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-1.5b-analyst",
    "messages": [{"role": "user", "content": "Explain why volatility cones widen during high CPI prints."}]
  }'

4. POST /v1/systemone — Laya 33ms Reflex Router

Fast zero-token portfolio triage and risk filtering.

5. GET /health — Service Status & Active Model Registry

Check runtime health and verify active model weights.