QuantEval — Real-Time Quant Analytics Platform

A full-stack analytics system for evaluating pair trading signals, live tick ingestion, alerts, replay mode, and mini-backtester.

🚀 Overview

This project implements a real-time analytical application designed for evaluating candidate quant developers. It demonstrates the ability to:

Ingest live Binance tick data

Sample & store data (1s, 1m, 5m)

Compute key statistical arbitrage analytics

Display real-time charts & summary stats

Trigger server-side alerts via rules

Replay uploaded OHLC files

Run a mini mean-reversion backtest

Provide an extensible and modular architecture

This codebase includes a FastAPI backend, Next.js frontend, candlestick charts using TradingView Lightweight Charts, and a clean dark UI.

🧱 Architecture Summary Backend (FastAPI + Socket.IO)

Binance WebSocket tick ingestion

Tick buffering & sampling (1s / 1m / 5m resampling)

OLS hedge ratio (intercept + beta)

Spread & z-score

ADF test (stationarity check)

Rolling correlation

Server-side rule-based alert engine

OHLC upload + parsing + replay mode

Mini mean-reversion backtester

CSV export

WebSocket push for alerts + ticks

Frontend (Next.js App Router + Tailwind + Lightweight Charts)

Live candlestick price chart

Spread & Z-score charts

Dark theme, collapsible panels

Alerts creation/deletion

Replay mode panel

Backtester panel

Summary statistics

CSV export

Toast notifications using react-hot-toast

▶ How To Run Locally Prerequisites

Python 3.10+

Node.js 18+

Windows PowerShell (or use .bat)

One-click startup

Use the included run_all.ps1:

.\run_all.ps1

This script automatically:

✔ Creates backend virtual environment ✔ Installs Python dependencies ✔ Starts FastAPI on localhost:8000 ✔ Installs Node dependencies ✔ Starts Next.js frontend on localhost:3000

📂 Folder Structure quant-eval/ │── backend/ │ ├── app.py │ ├── analytics/ │ ├── alerts/ │ ├── replay/ │ ├── backtester/ │ └── requirements.txt │ │── frontend/ │ ├── app/ │ ├── components/ │ ├── utils/ │ ├── styles/ │ └── package.json │ │── run_all.ps1 │── architecture.drawio │── architecture.png │── README.md │── ChatGPT_Usage.md

📊 Analytics Provided ✔ Hedge Ratio (OLS regression) ✔ Spread series ✔ Z-score (standardized spread) ✔ ADF test (stationarity check) ✔ Rolling correlation ✔ Summary stats ✔ Mean-reversion backtest (z>2 entry, z<0 exit) ✔ Replay mode from OHLC uploads ✔ Real-time candlestick chart 🔔 Alerts Engine

Server-side rules run continuously:

Examples:

latest_z > 2 rolling_corr_latest < 0.5 latest_spread > 10

Alerts are:

Evaluated on backend

Broadcast via Socket.IO

Shown as styled toast notifications

💾 Data Export

You can download:

Processed data

Spread & Z-score series

Full OLS/regression output

🧩 Extensibility

Designed so the following can be added without refactor:

Kalman filter hedge estimation

Robust regression (Theil-Sen, Huber)

Orderbook ingestion

Cross-correlation heatmaps

Multi-pair scanning

🎯 Final Notes

This system is designed as a professional demonstration of real-time quant engineering, analytics computation, data visualization, and modular system design.