How StrategyNet turns signals into portfolios
The path from a market observation to a portfolio is not one model call. It is a chain of distinct research objects, each with its own timing rules and its own test. A feature becomes a signal; a signal becomes a factor-mimicking portfolio; validated FMPs become sleeves in a final allocation. Keeping those objects separate is what makes the process inspectable.
This series follows that chain in four short animations. The written notes fill in the definitions and mathematics that move too quickly on screen.
From signals to a pure FMPPoint-in-time features are normalized, composed, and purified against the active risk model.
Realtime Factor ProjectionsThe prior model state and live-session evidence become coverage-aware rankings across stocks.
Walk-forward FMP validationFMP weights are frozen before the next return, then tested on ranking quality, net performance, and stability.
From selected FMPs to a final portfolioValidated sleeves are combined with covariance and mapped into a constrained asset portfolio.The objects that pass between stages
A feature is a measured value for one security at one time. A signal is a cross-sectional forecast: it says how securities rank relative to one another. An FMP is different again. It is a vector of security weights built to preserve exposure to one signal while removing unwanted market, sector, or style exposures.
Once that FMP is held through time, it produces a return history. Only then can it be treated as a sleeve, tested out of sample, compared with other sleeves, and considered for allocation. The final portfolio is therefore not a larger factor score. It is a constrained combination of already constructed and validated portfolios.
The same risk model appears twice for a reason. It defines the geometry used to purify individual FMPs, and later maps the selected FMP combination back to a consistent asset-level optimization problem. The four pages show where that continuity enters the process.
This walkthrough is for research and educational purposes. It illustrates how strategynet.ai organizes signal evidence into factors and scenarios. It provides no recommendation, investment advice, or instruction to trade any security.
Back to Insights