Our approach
Regime first, then conditions, confirmation and exposure.
The platform is organised around one research sequence, and its models are built so that each output can be traced, tested and argued with.
The research process
Four layers, each feeding the next.
Regime
What environment are we in?
Classify growth, inflation, liquidity, risk appetite and policy, and the direction each is moving. The classification is the shared starting point for every later step.
A regime read with its conviction and the conditions that would overturn it.
Conditions
Is the financial system adding impulse or removing it?
Read liquidity, credit creation, real rates, the curve and the priced policy path together, as one transmission system.
The direction and strength of the financial impulse.
Confirmation
Is the market agreeing with the thesis?
Test the macro view against credit, currencies, commodities, breadth, volatility and positioning. A thesis that markets do not confirm stays provisional.
A thesis marked confirmed, challenged or unresolved.
Exposure
Where does the thesis meet the portfolio?
Move to sectors, factors and securities, size the idea against the book, and keep the original conditions and invalidation points attached.
A position, its risk contribution and the conditions to watch.
How the models are built
Six rules that apply across the platform.
These are engineering rules built into the code, applied in every module.
- Release-dated history
- Where vintages exist, models use the value as first published, gated on its release date. A backtest cannot see data that did not exist yet.
- Coverage on every score
- Each score prints how many of its inputs are live, so conviction always reflects the evidence behind it.
- Evidence next to the result
- Distributions, correlations and seasonal statistics print their depth of history, so every signal is weighed correctly.
- Fit statistics on fitted output
- A volatility surface reports its R² and residual error, and a currency fair value is shown where the model's out-of-sample fit earns it.
- Explainable output
- Every regime, fair value and model book shows the inputs that produced it.
- AI with provenance
- Where AI extracts or summarises, it runs on your firm's chosen provider and the underlying document is kept with the output.
Request a demo
See it on your own questions.
A demo is a working session in the platform on a question you bring: a release, a company, a position or a portfolio. We reply within two business days.