Quantitative research,Analytics and AIfor portfolio decisions

Local-first. Private by design.

Two decades of institutional practice, built into tools that never leave your machine.

See the projects ↓

Quantitative researchPortfolio investmentMachine learningAnalyticsDecision science

IApproach

Local-first

Research data and models stay on your own machine. Nothing is uploaded unless you choose to upload it.

Evidence over narrative

Every method can be inspected, reproduced and tested. A claim without a test is just an opinion.

AI as an instrument

Models rank, extract and summarise. A person still makes the decision.

IIProjectsBuilt in-house

01 ML4T Academy A private, offline workbench for studying and researching machine learning for trading. It combines reading, notebooks, a learning mode and an optional AI tutor, all running on your own machine. Machine learningQuant researchOffline 01Read02Run notebooks03Ask the tutor Desktop workbench · Read the method → 02 QuantLab A self-contained Chrome extension that brings a Python quant and ML research library into the browser. There is no server and no remote code, so research data never leaves your machine. QuantML modelsIn-browser compute 01Load data02Compute in-browser03Export Chrome extension · Read the method → 03 JEV Decision Model A typed AI decision model that scores evidence from filings, news and research by its bearing on an investment question, and turns unstructured reading into a ranked, auditable basis for a decision. AIDecision scienceEvidence ranking 01Open a source02Score the evidence03Export the ranking Decision engine · Read the method → 04 Agentic Investment & Trading Coordinated AI agents for research, portfolio construction, risk and execution. Each agent works inside explicit limits, every action is logged, and a human signs off before capital moves. Agentic AIPortfolio constructionExecution 01Agents propose02Everything logged03Human approves Agent framework · Read the method →

IIIBriefings

An archive of market research from the desk, numbered by volume and issue. It records method and is not a recommendation.

IVAbout

A small group of researchers and engineers building systems that decide under uncertainty.

Most of what we learn stays with us.

Coverage: Singapore, Hong Kong, Southeast Asia
Disciplines: Quantitative research, machine learning, decision science
Founded
Access: By introduction

VContact

office@sigma82.comResearch collaborators and portfolio managers welcome.