Win Probability Engine — Bayesian pWin with 8-Factor Model

The Win Probability Engine is the bid/no-bid decision-support component of the Aliff Solutions platform. It calculates a 0-100 Win Probability (pWin) score for any federal opportunity by weighing 8 factors that historically predict federal source selection outcomes. Output includes a GO/NO-GO recommendation, ranked strengths and gaps with mitigation strategies, and a confidence indicator.

The 8 factors are: technical capability match, past performance relevance, incumbent vulnerability, pricing position, set-aside eligibility, customer relationship, teaming strength, and competitive landscape density. Weights are agency-adjusted — the same factor predicts differently at different agencies based on historical source selection patterns.

The engine calculates a pWin score by combining the 8 factor scores using agency-adjusted weights, then compares the score to a client-configurable threshold. Opportunities above the threshold receive a GO recommendation; below receive NO-GO. The threshold is typically set at the pWin level where expected proposal investment ROI turns positive given the client's proposal cost structure and historical win rate.

Mitigation strategies are tied to specific gaps identified per factor. Low past performance relevance might trigger teaming recommendations with partners holding the missing past performance. Weak customer relationship might trigger customer engagement plan recommendations. High competitive density might trigger price strategy recommendations from the Pricing Intelligence Engine. If multiple factors are weak simultaneously, the engine surfaces NO-GO and the recommendation is to redeploy capture investment to a higher-pWin opportunity.

Each pWin score carries a confidence indicator reflecting the completeness and reliability of the underlying signal data. High-confidence scores come from data-rich situations — established NAICS at familiar agencies with multiple data sources confirming each factor. Lower-confidence scores indicate data gaps — newer NAICS, unfamiliar agencies, or sparse historical pattern. Low-confidence scores trigger additional Phase 03 qualification work before final GO/NO-GO commitment.

pWin is opportunity-specific — it predicts the probability of winning a specific opportunity given its specific characteristics. Win-rate forecasting is portfolio-level — predicting the percentage of pursuits that will win across a pipeline. Both are useful: pWin drives individual bid/no-bid decisions; portfolio win-rate forecasting drives capture investment budgeting and pipeline sizing decisions.

Federal recompete data shows that approximately 63% of services recompetes retain the incumbent. Pursuing well-performing entrenched incumbents is a low-yield use of capture investment. Pursuing vulnerable incumbents — where the Incumbent Vulnerability Scoring model surfaces 4+ visible weakness signals — substantially improves challenger win probability. The factor links the two engines so vulnerability evidence directly raises pWin.

Set-aside eligibility is binary at the threshold (you either qualify or you don't) but binary changes can produce large pWin swings. A client with valid SDVOSB certification competing in an SDVOSB set-aside has a substantially smaller eligible competitor pool than the same client in a full-and-open competition. The engine credits the set-aside advantage explicitly and surfaces the implication for capture strategy.