Trust and transparency
Methodology, assumptions and limits
Read the rules behind TakaCompass's educational planner and the differences between the browser demo and reference engine.
TakaCompass uses deterministic code to illustrate how goals, liquidity, time horizon and risk constraints can affect a plan. The browser app is a simplified demo: its capacity score and allocation bands are rules-based teaching proxies, and its scenario chart is illustrative arithmetic. A separate reference engine contains more detailed policy and calculation logic; neither is validated personalized advice or a return forecast.
Suitability logic in the reference model
The reference methodology models risk tolerance, financial capacity and required risk separately. It caps the recommendation at the lower of tolerance and capacity. Emergency reserves and known near-term withdrawals are ring-fenced before an allocation is considered. If the goal requires more risk than capacity permits, goal-rescue options include contributing more, extending the horizon, reducing the target, building reserves or addressing expensive debt.
Browser demo boundaries
The shipped browser calculation in `apps/web/lib/demo.ts` is not the portfolio engine or FastAPI service. It estimates capacity from a horizon proxy, income-stability cap, loss-tolerance proxy and reserve penalty; an illustrative allocation then respects that cap and uses whole-taka rounding. Its fan chart derives a centre and spread from contributions, horizon and equity weight. Research rows are fictional fixture records. It does not build executable security lots or run a stochastic simulation.
Reference policy and uncertainty
The separate reference engine uses a versioned horizon/strategy ceiling, eligibility-first filtering, concentration and denomination constraints, whole-share rounding and a visible residual-cash line. Its scenario module uses a seeded random stream, monthly return draws and explicit assumptions; the stated percentiles are estimates, not forecasts. Coefficients, normal-return assumptions and historical-quantile tail measures are educational model choices, not empirically validated suitability thresholds or complete representations of market risk.
Versioning and source record
The source repository records methodology, data lineage, compliance and model limitations. A result's data and transformation dates must not be confused: effective time describes when an observation applies; retrieval time records when it was obtained. The current browser dataset is fictional and remains labelled `DEMO DATA — NOT LIVE`.