One Live Book Across Stocks, Crypto and Mutual Funds: Quant Research Infrastructure for Licensed Firms

For firms licensed to research and trade client money, the first bottleneck is rarely the model — it is the data. Positions live inside separate trading applications: one for listed stocks, one for crypto, one for mutual funds. Before any research can happen, someone reconciles them by hand into a spreadsheet that is stale before it is finished. Chronexa built the layer underneath: every account connected through its API, every position normalised into one continuously updated book, and on top of that, the infrastructure for building and testing trading algorithms with classical machine learning and modern AI.

One boundary was fixed before the first line of code: this system is built for licensed firms only, and it does not trade on its own. Models research, rank and forecast; the licensed desk reviews and approves every action; and every step is logged for audit.

The Challenge

A Portfolio Spread Across Apps Is Not a Portfolio

With stocks, crypto and mutual funds held in different applications, there was no moment at which the firm could see its whole book at once. Exposure questions — how concentrated, how correlated, how hedged — were answered from a manually assembled spreadsheet that was hours or days old by the time anyone read it.

Research Without a Standard

Algorithm ideas were tested ad hoc — different data pulls, different assumptions, different time periods. Two back-tests could not be compared with each other, so promising ideas died in argument and weak ones survived on enthusiasm.

A Hard Regulatory Boundary

Everything in scope touches regulated activity. The infrastructure had to keep the human decision — and the licence that authorises it — unmistakably in charge, with an audit trail capable of proving it after the fact.

The Solution: One Book, One Research Loop, Human Control

Every Account, One Live Book

API connections to each trading application feed positions, fills and cash into a single normalised view across all three asset classes. The book updates continuously. The research desk stops reconciling and starts researching.

Algorithm Research Infrastructure

Classical machine learning and AI models are built and evaluated inside one standardised loop: the same historical data, the same back-testing method, the same reporting for every idea. A price-behaviour model and a signal-ranking model are judged the same way, so results are comparable — and reproducible months later.

Models Advise — the Licensed Desk Decides

Model output arrives as ranked signals and forecasts with the reasoning attached. Nothing executes autonomously. The licensed team reviews, approves or rejects, and the system records who decided what, when, and on which evidence.

The Daily Research Digest

Each morning the desk receives the book and the signals in one digest: current exposure across asset classes, model output since yesterday, and the back-tests that finished overnight.

Results

Three Asset Classes on One Screen

The firm's whole book — stocks, crypto and mutual funds — became a single live view. The reconciliation spreadsheet is gone, and exposure questions are answered from data that is current rather than hours old.

Research Became Repeatable

Every algorithm idea now runs through the same loop — same data discipline, same back-test, same review. Ideas are compared on evidence instead of enthusiasm, and a result from last quarter can be reproduced, not just remembered.

Compliance Held by Design

The licence boundary is structural, not procedural: models cannot act, humans must, and the audit trail writes itself. The firm accelerated its research without moving a single decision outside its regulatory authority.

Why This Project Matters

In regulated finance, the edge from AI is not a black box that promises returns — regulators, clients and good sense all push back on that. The edge is infrastructure: clean unified data, disciplined testing, and models that make licensed humans faster and better informed. That is what Chronexa builds, and it is why this pattern holds for any firm whose research has outrun its data.