Alpha Variance

AlphaVariance is a private finance intelligence platform built for internal market research, model validation, and AI-assisted prediction workflows across multiple financial instruments. The initial focus is Indonesian equities, with the architecture designed to expand toward crypto, FX, indexes, ETFs, commodities, futures, options, bonds, and macro data.

The project is intentionally private, so the public explanation stays high-level. Its core focus is to turn historical market data, technical context, machine learning predictions, and risk filters into disciplined decision-support signals without exposing proprietary methodology.

Frontend — Built with Next.js, React, TypeScript, Tailwind CSS, and shadcn/ui to create a refined private dashboard for scanning instruments, reviewing model status, and reading validation results.

Backend — Powered by Go with Echo, PostgreSQL/TimescaleDB, and Redis for market data ingestion, indicator processing, scheduled workflows, signal storage, and fast internal access to model outputs.

ML Service — Uses a separate Python FastAPI service for research workflows, including feature engineering, model training, challenger evaluation, batch prediction, backtesting, and paper-trading validation.

Infrastructure — Managed as a monorepo with Turborepo, containerized with Docker, and structured around a generic instrument model so the system is not locked to a single asset class.

I owned the project end-to-end:

• Product Strategy — Defined the private research workflow for market scanning, model review, and signal validation
• System Architecture — Designed the multi-instrument data model, scheduler, model registry, and model-gating flow
• Frontend Development — Built the landing page and private dashboard with a calm, finance-focused interface
• Backend Development — Implemented APIs, ingestion workflows, technical analysis processing, and internal data orchestration
• ML Workflow — Integrated research, backtesting, prediction, and paper-trading flows behind controlled validation stages
• Deployment — Shipped the production deployment and operational workflow for ongoing research usage

• Multi-Instrument Foundation — Generic instrument, provider-symbol, and candle model designed for stocks and future asset classes.
• Market Data Pipeline — Ingests historical OHLCV data, validates data quality, and prepares market context for research.
• Technical & ML Research Flow — Combines technical analysis, feature engineering, model training, and probabilistic prediction review.
• Model Gate Validation — Keeps signals behind readiness, backtest, paper-trading, and forward-evidence checks before they become useful.
• Private Signal Dashboard — Provides an elegant internal view for freshness, model status, scores, active instruments, and research activity.