The operating system for AI-run investing
Every AI decision in your fund, provable.
LocusQuant verifies what your AI reads, governs what it does and records every decision, so you can show an allocator or an examiner exactly why. We are proving it first on our own fund.
The problem
Funds are adding AI to research and trading. Allocators now ask how it is validated, and regulators ask how it is supervised. Most firms answer with a policy document.
We answer with a record.
Three instruments a fund can use this month.
Each one names a problem, hands back a result you can forward, and starts with a short conversation. No prices on the site; every engagement starts with Request Access.
Find out how much of your Sharpe ratio is luck, before an allocator does.
Illustrative figures.
Weigh what each data subscription is really worth to your decisions.
Every claim your AI pulls from a filing, checked against the sentence it came from.
One operating system beneath every instrument.
Eighty seconds, ten instruments, each shown doing its one job. Every one of them runs on LoQOS.
Start with one. Each runs on LoQOS, so each one you add reuses what the last one built. Open the film on its own page →
06 · Evidence
Most AI asks for trust. We would rather show the arithmetic.
From an Assay report
Momentum, top decile, monthly rebalance
Backtest supplied with 400 recorded variations
| Sharpe, as reported | 1.84 |
| Sharpe, after the trials behind it | 0.52 |
| Share of the result explained by selection | 71% |
| Holds out of sample | partly |
Most of the headline result comes from choosing the best of many tries. A smaller, real effect remains.
Layout specimen. Figures are illustrative, not a client result.
Effective panel size
Five analysts who read the same news, with errors correlated at 0.6, are worth 1.47 independent opinions, not five. A room of agreeing agents does not get to bet bigger.
The model proposes. Deterministic code disposes.
Read the maths →On the record, on ourselves first.
We run our own fund on the same system. Today it is in paper trading, and every decision is recorded.
- Live market feedroadmap
- Thematic segregationroadmap
- Research and selectionlive
- Risk and direction gatesroadmap
- Execution and allocationroadmap
- Continuous stewardshiproadmap
Outcomes flow back into research. Fills, not code, set the date for live capital. See the roadmap
For firms that have to explain their AI to someone else.
Emerging managers
Due-diligence ready from day one.
Quant teams
Validation before promotion, not after the drawdown.
Data vendors
Independent proof that your data predicts.
Platforms and wealth apps
Governance for the AI features you ship to clients.
Data science across credit-bureau and fintech systems. Builds the quantitative and modelling core of LocusQuant.
Enterprise integration background. Wires the operating system into production-grade infrastructure.
See your own numbers, on the record.
Does our data leave our environment?
No. Raw data stays with you. We agree what is needed in the first conversation, and where possible the work runs inside your environment.
How do we start?
Request access and tell us what you are working on. A founder replies, and the first call settles scope. Nothing is priced on the site.
What is built today, and what is roadmap?
Assay and Tare are available to design partners, Scribe is in pilot, and seven more instruments are in build, previewed on the instruments page. Our own fund runs its research stage live, on paper, with every decision audited. The rest is on the roadmap, stated plainly.
Does the AI trade on its own?
It proposes. Rules set by people decide what is allowed, and the rules are enforced below the agent, not in its prompt.
Can we see the maths?
The ideas, with toy numbers, are on the maths page. The detail comes in the conversation.