HAQUE TRADING

CONNECTING

HAQUE TRADING

Autonomous AI-Powered Trading Platform

Portfolio NAV
$—
Combined stock + crypto
XGBoost Sharpe
—
Validation risk-adjusted return
Max Drawdown
—
Peak-to-trough decline
Live Performance

Real-Time Portfolio
Monitoring

Live P&L, market regime classification, and position-level performance — updated every 2 seconds via WebSocket.

Today's P&L
—
Weekly P&L
—
Cumulative Return
—
Market Regime
—
Current Positions —
TickerSharesEntryCurrentP&L %Unrealized
Connecting...
Strategy

How Haque Trading Works

A fully autonomous 6-stage pipeline — from universe screening to daily model retraining, running 24/7 with monthly ticker rotation.

📊
Scan & Select
S&P 500 + NASDAQ-100 → 25
→
🧠
Train Models
XGBoost (Stocks) + SAC (Crypto)
→
✅
Validate
Walk-forward + rank IC
→
⚡
Execute
IBKR hourly rebalance
→
🔄
Self-Learn
Daily retrain + adaptation

Haque Trading uses an asset-specialized architecture: XGBoost Alpha is the sole stock model — a cross-sectional ranking engine that scores 25 dynamically-selected equities using 41 features including price action, technical indicators, market sentiment, and fundamental quality — retrained daily with 750 days of data. Soft Actor-Critic (SAC) reinforcement learning handles crypto markets (BTC, ETH, SOL), retrained weekly with 1M timesteps. A SMA-50 momentum overlay filters out stocks in downtrends, while earnings avoidance protects against binary event risk. The stock universe auto-rotates monthly from the S&P 500 and NASDAQ-100.

41
XGBoost Features
25
Universe Stocks
24/7
Crypto Trading
Daily
Model Retrain
Risk Management

19 Layers of Protection

Institutional-grade capital preservation — every trade passes through 19 independent risk gates.

1
Pre-trade risk gates (position & sector limits)
2
Market regime detection (5-tier classification)
3
Drawdown recovery mode (auto-reduce exposure)
4
Trailing stop-loss (8% from high-water mark)
5
Hard stop-loss (12% from entry price)
6
Out-of-distribution detection
7
Graduated circuit breaker
8
Incremental rebalancing (2% drift threshold)
9
TWAP execution (large order splitting)
10
Data quality gates (4-stage validation)
11
Walk-forward embargo (7-day buffer)
12
Daily loss limit (-3% partial, -5% full)
13
PerformanceGuard (5 severity levels + kill switch)
14
Autonomous capital protection (3-layer self-healing)
15
Portfolio correlation guard
16
Position correlation sizing
17
Observation normalization warmup (cold-start elimination)
18
SMA-50 momentum overlay (trend filter on allocations)
19
Earnings avoidance (graduated close before binary events)
Stress Test Grade
A — 100/100
Guard Status
NORMAL
Circuit Breaker
100%
Historical Crisis Performance Regime-aware stress testing
CrisisPeriodMax DrawdownAlpha vs S&P 500
2008 Financial Crisis2007–2009-10.6%+45.8%
2011 European Debt2011-4.2%+35.3%
2015 China/Oil Crash2015–2016-7.7%+37.7%
2018 Volmageddon2018-8.2%+19.9%
2020 COVID-192020-8.3%+43.3%
2022 Rate Hike Bear2022-10.0%+26.3%
Technology

Institutional-Grade
Infrastructure

Unified cloud architecture with XGBoost daily retraining, automated model deployment, 24/7 uptime monitoring, and real-time WebSocket data feeds.

Model Training

Google Cloud Platform
  • Daily XGBoost Alpha retrain (Mon–Fri)
  • Walk-forward validation (4 folds)
  • Model validation & deployment gate
  • Automated preflight checks
⇄

Production Trading

Google Cloud Platform (24/7)
  • Stock runner (hourly, 25 equities)
  • Crypto runner (24/7, BTC/ETH/SOL)
  • Monthly universe auto-rotation
  • Real-time WebSocket dashboard
  • Production email alerts
Python 3.10 XGBoost Alpha SAC (Stable Baselines 3) Interactive Brokers Google Cloud FastAPI WebSocket Walk-Forward Validation
2.33
XGBoost Val Sharpe
A
Stress Test Grade
19
Risk Layers
v5.53c
Current Version