Private research · In development
Markets are uncertain. Responsibility shouldn’t be. We’re building an AI forex research system with distinct teams, independent checks and human authority.
Explore our approachEvidence over conviction
Separation of authority
Human-defined boundaries
THE APPROACH
A convincing explanation isn’t evidence. Our intended workflow gives every strategy a path to scrutiny before it earns any authority to trade.
The intended broker workflow.
External execution is still in development.
Explore bounded strategy ideas, record experiments and preserve the assumptions behind each result. A proposal begins a review; it cannot approve itself.
Research may propose a strategy.
It may not activate its own proposal.
Review results independently, include trading costs and examine evidence beyond the data used to develop an idea. An inconclusive result is a reason to keep researching.
Validation may reject the evidence.
It may not rewrite results to make them pass.
Allocate only within approved limits. Check each proposed order against explicit risk rules, preserve an audit trail and restrict new activity when required checks fail.
AI may recommend a change.
It may not raise its own risk ceiling.
THE COUNCIL
A system organised around specialised AI roles and deterministic controls, with the owner setting its objectives and limits. These are software roles, not a claim of an employed trading team.
Develop and document strategy proposals.
Cannot activate its own proposals.Challenge the method and assess the evidence.
Cannot alter results to secure a pass.Allocate among eligible strategies.
Cannot increase the overall risk ceiling.Enforce limits and block invalid activity.
AI recommendations cannot override rules.Carry out authorised trading instructions.
Cannot rewrite strategies or award capital.Reconcile records and investigate discrepancies.
Restrictions follow predefined rules.Maintain the system and prepare tested fixes.
Trading-behaviour changes require approval.Role separation is part of the design. The local implementation is being tested; complete broker integration and trading qualification remain release gates.
WHERE WE ARE
MarketHelm is a private project in active development. The local research environment is implemented. Connecting the intended workflow to a demo broker and establishing genuine strategy evidence are the next stages.
More about the projectResearch workflows, mock execution, audit records and operational monitoring are implemented.
Account verification, execution integration and broker acceptance remain in development.
Prospective evidence and independent evaluation are required before automated demo activation.
Implementation is not proof of trading performance.
No verified return figures are published.
A FEW DETAILS
MarketHelm is a private AI forex research project exploring how specialised AI roles, independent validation, explicit risk rules and human oversight can work together. Its current environment is local and experimental.
No. MarketHelm does not currently offer public accounts, investment products, managed funds or trading signals. This website introduces the research project; it does not accept deposits or provide access to trading controls.
No. The intended system gives AI bounded responsibilities. The owner approves objectives, risk limits and executable trading-code changes. Strategy activation is intended to operate only within an approved policy after independent qualification.
The system records experiments and reconciled outcomes to support evaluation. Proposed strategy or model changes need separate evidence and review before deployment. Recording a result does not establish that a model has improved.
No. Forex trading involves risk, and losses can exceed planned thresholds. Simulation and demo results do not guarantee future performance. This project has not established a verified trading edge.
