AI Decision Infrastructure
Weethley applies machine learning to large volumes of market data, so independent earners can act on risk-adjusted recommendations without hiring a research team.
Model Output — Illustrative
Representative visualisation of predictive signal strength. Not indicative of actual account returns.
The Challenge
Gig economy earners and independent analysts rarely work fixed hours, and the time available for market research changes from week to week. Spreadsheet tracking and ad hoc news-reading were not built for continuous, high-volume data. By the time a pattern is spotted manually, the window to act on it has often narrowed considerably.
Setup Protocol
The onboarding sequence is deliberately short. Each step completes once the previous one is confirmed, with no manual data entry beyond the initial connection.
Link an existing brokerage, exchange or portfolio API. Weethley reads read-only account and market data; no trading permissions are required at this stage.
Choose a risk band and a capital allocation ceiling. These settings govern every recommendation the model produces and can be adjusted at any time.
The model returns a ranked set of recommendations with confidence bands and a stated rationale, ready for review before any action is taken.
Core Capabilities
The forecasting layer combines time-series models with gradient-boosted decision trees, trained on historical pricing, volume and volatility data. Outputs are expressed as a probability-weighted range rather than a single projected figure, reflecting genuine market uncertainty.
Every recommendation passes through exposure checks before it reaches the interface: position concentration, correlation with existing holdings, and distance from the selected risk band. Items that breach a threshold are flagged rather than hidden.
Market data is ingested continuously rather than on a fixed schedule, so the model's view of conditions updates as new information arrives. Processing load scales automatically with data volume, without manual intervention.
About Weethley
Weethley was designed around a simple constraint: most independent earners do not have a fixed block of hours to dedicate to research. The platform compresses the groundwork — data gathering, pattern recognition and risk checking — into a process that runs in the background and surfaces only what requires a decision.
The underlying infrastructure is maintained by a small analytics team based in the UK, with models reviewed and recalibrated on a regular cycle as market conditions shift.
Methodology
Predictions are produced by an ensemble of time-series forecasting models and gradient-boosted trees, trained on a shared feature set covering price history, traded volume, volatility and liquidity. Outputs are weighted by recent accuracy before being combined into a single recommendation.
Market data is drawn from regulated exchange and brokerage feeds and cross-checked against a secondary source before use in a live calculation. Any feed that falls out of tolerance is excluded from that cycle rather than estimated.
Each model version is tested against historical data spanning multiple market cycles before deployment, with performance reviewed on a rolling basis. Backtested results are not a guarantee of future performance and should be read alongside independent judgement.
Connect a read-only data source, set a risk band and receive a first set of recommendations before the kettle boils. No onboarding call is required.
Initialise PortfolioSetup typically completes in under 60 seconds, from connection to first output.
