Macfax Roster Outlook
Next-season team projections built from the roster up
The Roster Outlook builds a forward-looking team projection from the ground up — starting with each player's individual talent estimate, then layering in expected playing time, how well the roster's styles fit together, and how much continuity or roster turnover introduces uncertainty. The result is a projected efficiency rating with a national ranking estimate, uncertainty bands, roster fit grades, and continuity diagnostics for the upcoming season.
What It Measures
The Roster Outlook estimates what a team's current roster is likely to produce next season in terms of adjusted offensive efficiency, adjusted defensive efficiency, and net efficiency margin. It separates talent from fit — a roster with individually strong players can still have structural weaknesses if the pieces do not complement each other. The feature also measures roster continuity (how much production is returning versus transferred in or newly added), and flags where transfer dependence or roster composition creates meaningful projection uncertainty.
Why It Matters
In-season ratings tell you what a team has done. The Roster Outlook tells you what a team is likely to become. Transfer portal activity, recruiting classes, and roster turnover make next-season projections fundamentally different from extrapolating current ratings. By building the projection player by player — incorporating each player's talent estimate, minutes, and fit contribution — the Roster Outlook captures the actual mechanism by which rosters translate into team performance, rather than assuming the current team quality persists unchanged.
How to Interpret
The projected AdjEM, AdjO, and AdjD are the model's central estimates for next season. The national ranking estimate shows where that efficiency profile would rank among Division I teams in a typical season. Uncertainty bands (shown as a projected rank range) reflect how wide the realistic distribution of outcomes is — a team with a narrow range has a more predictable outlook, while a team with a wide range has more variance in either direction. Fit grades run from A+ (elite roster composition) to F (significant structural weaknesses), separated into offensive fit and defensive fit. A team can have strong individual talent but a B- fit grade if the roster lacks spacing, playmaking, or defensive structure. Player tiers — All-American, All-Conference, Starter, Reserve, Bench — reflect the projected quality level for each player relative to Division I averages. Recruitment type labels (Returner, Transfer, Newcomer) show where projected production is coming from and where it carries more uncertainty.
Technical Notes
- Player talent estimates use Bayesian Performance Rating (BPR) as the core input. For returning players, BPR is based on prior college performance with appropriate regression toward the mean. For transfers, prior college BPR is adjusted for the change in competition level and role. For freshmen and other newcomers, recruiting rank and star rating serve as priors where college performance data is unavailable.
- Minutes allocation is projected using a minutes model that accounts for role, quality tier, roster depth, and historical minutes patterns. Minutes projections are a key uncertainty source — a player whose role expands or contracts significantly will produce differently than the model expects.
- Roster fit is evaluated separately for offensive and defensive dimensions. Offensive fit assesses playmaking coverage, spacing distribution, ball security, finishing quality, and free-throw pressure. Defensive fit assesses rim protection, defensive rebounding, perimeter coverage, and structural composition. Each dimension receives a letter grade. The exact subcomponent weights and structural penalty logic are internal to Macfax.
- Contextual adjustments incorporate the team's stylistic identity — pace tendency and offensive and defensive scheme — to assess whether the projected roster aligns with how the coaching staff has historically deployed talent.
- Continuity is measured as both a minutes fraction (how much of projected playing time comes from returners) and a talent-weighted score (how much of projected BPR production is retained). High transfer dependence without strong fit scores increases projection uncertainty.
- Team-level projected ratings are derived by translating the aggregate talent and fit signals into an efficiency margin estimate, calibrated against the historical relationship between roster quality and on-court performance. The translation is not linear — fit matters, and the same aggregate talent level produces better ratings when the pieces complement each other.
- Uncertainty bands represent roughly a ±2 standard deviation range around the central projection, reflecting player projection variance, minutes uncertainty, and continuity risk. The central estimate is more likely than any outcome at the edges of the range, but outcomes outside the range are possible.
- The scenario editor on the Roster Outlook page allows users to modify the projected roster — adding, removing, or swapping players — and see how those changes would shift the projection. Scenario results use the same model as the baseline projection.
- •The Roster Outlook is a projection, not a guarantee. Basketball roster construction involves significant uncertainty — injuries, role changes, player development, and coaching adjustments can all cause actual outcomes to diverge from the projection.
- •Transfer portal movement is ongoing throughout the offseason. The projection reflects roster information available at the time of the most recent pipeline run; subsequent transfers, decommitments, or additions are not automatically reflected.
- •Minutes projections are a major source of uncertainty, especially for freshmen and transfer players whose college roles are inherently difficult to predict.
- •The model does not account for coaching changes unless they are already reflected in the team's current coaching profile.
- •Defensive projection carries more uncertainty than offensive projection at the player level, which propagates to the team-level defensive fit and projected AdjD estimates.
- •Early offseason projections (before most transfer portal decisions are finalized) carry wider uncertainty than late-offseason projections with a more complete roster picture.
- •The scenario editor produces estimates, not predictions. Manually constructed rosters may include combinations that the model has limited data to evaluate confidently.
Illustrative: a team projects AdjEM +18.5 with a national rank estimate of #22, but with a rank range of #14–#35. Their offensive fit is B+ (strong spacing and playmaking, slight ball-security concern from a high-usage transfer) and defensive fit is B (solid rim protection, but limited perimeter stopper in the backcourt rotation). Continuity score is 62 — meaning roughly 62% of projected production comes from returners, with the rest dependent on two transfers. If the transfers perform to projection, the upper end of the range is realistic. If they underperform their college history, the team slides toward the lower end.