Investment management,
run by models.
InferenceIM builds elaborate quantitative models that research the market, manage the portfolio and execute — continuously, as conditions change, at machine speed.
Data-inferred investment.
Why autonomous
A discretionary manager reads the market a few times a day. A model reads it continuously.
InferenceIM's models ingest market data as it forms, re-estimate their view, and adjust positions without waiting for a human in the loop. The result is discipline that doesn't fatigue, hesitate or drift from its mandate — and reaction times measured in fractions of a second rather than meetings.
Autonomy applies to execution, not to accountability. Mandates, risk limits and controls are set — and enforced — above the models, by people. The model manages the portfolio; the firm governs the model.
Method
Three stages. Each one has to earn the next.
-
01
Research
Every strategy begins as a hypothesis tested against data. Ideas are examined for statistical robustness, capacity and decay; those that fail out-of-sample never reach the portfolio.
-
02
Modelling
Validated signals are engineered into elaborate quantitative models — systems that estimate, decide and size positions within strictly defined risk parameters, and keep learning as new data arrives.
-
03
Execution
Models execute their own decisions at machine speed, as markets move. No committee sits between signal and trade — only the risk framework the models must operate inside.
Risk & governance
Autonomy without control is just risk.
Every model operates inside hard limits — exposure, concentration and drawdown constraints enforced by systems independent of the models themselves. Breaching a limit halts the model, not the safeguard.
Human oversight is continuous and deliberately separate from model development. Positions, behaviour and limit utilisation are monitored in real time, and every model can be halted manually at any moment.
- Independent limits
- Risk constraints enforced outside the models' own code.
- Real-time monitoring
- Continuous human supervision of positions and behaviour.
- Manual override
- Any model can be halted by the team, at any time.
Team
A small partnership of researchers and engineers.
InferenceIM is built and run by the people who write its models. Research, engineering and risk sit at one table — there is no distance between the idea, the code and the accountability for both.
-
Founding Partner
Research
Biography to follow.
-
Founding Partner
Technology
Biography to follow.
-
Founding Partner
Risk & operations
Biography to follow.
Contact
InferenceIM works with professional clients and eligible counterparties only.