AI-NATIVE QUANTITATIVE RESEARCH

We teach machines
to reason about markets.

Auditable agentic workflows for quantitative research & development.

Work with usExplore our services

OUR APPROACH

Machine reasoning sits at the centre of our investment process.

Most firms bolt AI onto human workflows. We built the reverse: machines do the research, signals and execution; humans own risk and allocation.

Every decision leaves a trace. Theses cite their sources; orders trace back to weights. Audit any of it.

We open-source the infrastructure, not the alpha. Our agents and tooling are public on GitHub; our quant research and strategies stay in-house.

THE SYSTEMS BEHIND OUR RESEARCH

Four services.
One reasoning engine.

From research to deployment, each layer is designed to be useful, inspectable and ready for production.

01

STRATEGY DEVELOPMENT

From research idea to live-ready strategy.

Build, test and evaluate beta and alpha strategies through one reproducible backtest-to-live workflow. Every strategy runs over the same investment universe, data layer and backtest engine, so results are comparable and reproducible. The infrastructure is open; the strategies stay in-house.

  • Reproducible backtest-to-live workflow
  • Beta spanning tests and alpha rank-prediction reports
  • Return attribution: market beta, leverage, selection alpha
Explore Strategy Pipeline↗
Fig 01Strategy lifecycle
RESEARCH ENVIRONMENTActive

$ initialise research workspace

Cross-sectional signal research

Normalising fundamentals across the shared investment universe.

Universe
1,284 equities
Dataset
Point-in-time
State
Exploring

02

RELIABILITY LAYER

Catch failures before they reach production.

Test research and quantitative workflows against clear, machine-checkable standards. The Evaluation Platform detects hallucinated figures, stale data, broken citations and strategy regressions before deployment — it is how the other three services stay reliable in production.

  • Verifiers grounded in primary-filing data
  • Baseline and regression testing against evals
  • Automated production gates
Explore Evaluation Platform↗
Fig 02Evaluation gates
Release confidence86.4%
92Evidence
84Freshness
88Citations
81Regression

03

FUNDAMENTAL RESEARCH

Investment research in minutes, not hours.

Ask for a company deep-dive, earnings analysis or sector comparison and receive an auditable thesis grounded in primary filings, not model memory. Research that previously required a full day completes in under ten minutes. Open source as zen-fundamentals.

  • Company and earnings analysis
  • Point-in-time universe, organised by supply-chain role
  • Traceable figures and citations
Explore Research Agent↗
Fig 03Traceable research output
ZR–142Earnings revision monitor12 sources↗ZR–138Semiconductor supply chain21 sources↗ZR–131Consumer margin signals9 sources↗

04

QUANTITATIVE DEVELOPMENT

A coding agent built for quantitative development.

Maintain repositories, connect proprietary quantitative data through MCP and run development tasks in a sandboxed environment. Every session tracks model usage, cost and latency for evaluation and control. Open source as zen-coding.

  • Repository-scale maintenance
  • Sandboxed execution and guardrails
  • Per-session cost and latency traces
Explore Coding Agent↗
Fig 04Protected execution dashboard
REPOSITORY CONTEXTReady to act
Zen Coding Agent guardrail blocking a write outside the permitted workspace
Boundary checkWrite stopped before execution

portfolio_system/ sits outside the permitted notebooks/ workspace. No file was created.

Safe stateWorkspace unchanged

Recorded workflow demonstration: the agent reuses PPO and TD3 mappings, attempts to create portfolio_system/__init__.py, and is blocked because the destination is outside the permitted notebooks workspace. No file is created.

Ready to replay
Source
notebooks/
Models
PPO · TD3
Next action
Review before write

WHO WE ARE

Built around the models.
Grounded in the real world.

We’re AI-native

Machines do the heavy lifting; people own the risk.

We’re builders

Everything we run, we built — pipelines, reasoning systems and execution.

We’re aligned

We trade on the systems we publish, with our own capital.

WORK WITH US

We research and trade our own strategies. Let’s talk.

Systematic

Strategies researched, tested and executed by our platform.

Risk first

Position limits, drawdown controls and monitoring are engineered in.

Auditable

Every decision is logged — thesis, weights and orders.

Get in touch

Interested in investing in our fund, or in strategy research collaboration? Contact yvette@zentradings.com.