Macfax Glossary

AdjO estimates how many points a team would score per 100 possessions against an average D1 defense. Raw offensive efficiency is iteratively adjusted based on the strength of defenses faced, giving a more accurate picture of offensive quality than unadjusted points-per-possession.

AdjD estimates how many points a team would allow per 100 possessions against an average D1 offense. Raw defensive efficiency is iteratively adjusted for the strength of offenses faced. Lower is better — elite defenses allow fewer points per possession.

AdjEM is the single most predictive measure of overall team quality on Macfax. It captures both offensive and defensive performance adjusted for opponent strength. A positive number means the team scores more efficiently than it allows; a larger number means a wider quality gap.

Tempo measures how fast a team plays, expressed as possessions per 40-minute game after adjusting for opponent pace tendencies. Possessions are estimated from box-score statistics using the standard approximation. Tempo is an independent dimension from efficiency — fast teams are not inherently better or worse.

eFG% is the most important of the Four Factors. It adjusts raw field goal percentage to account for the extra value of 3-point shots. A made 3-pointer counts as 1.5 field goals in the formula, making eFG% a fairer measure of shooting efficiency across teams with different shot distributions.

TOV% measures how often a team gives the ball away. Lower is better for offense; higher is better for defense. On the margin, what matters is forcing more turnovers than you commit — a positive TOV edge means the team benefits from this factor.

ORB% measures a team's ability to extend possessions by recovering their own misses. It's calculated relative to available rebounds — only rebounds the offense could have gotten. Defensive rebound rate (DRB%) is the same concept from the other end, measuring how well a team limits opponent offensive rebounds.

FTR captures a team's ability to draw fouls and convert from the stripe. On offense, a high FTR means earning extra scoring opportunities at the foul line. On defense, a low FTR allowed means not fouling often. The FTR margin compares how often a team gets to the line versus how often it sends opponents there.

FFI combines eFG% margin, TOV edge, rebounding edge, and FTR margin into a single 0–100 score. Each factor is converted to a z-score against the national distribution for that season, weighted by its importance, and scaled to a 50-point baseline. A score of 50 is exactly average; higher is better. Scores are season-relative and not directly comparable across seasons.

BPR is Macfax's primary player evaluation metric. It combines box-score contributions (scoring, rebounding, assists, steals, blocks, turnovers) with on-court impact signals to estimate a player's net effect on team performance per 100 possessions. The Bayesian component stabilizes estimates for players with limited minutes. BPR = OBPR + DBPR.

OBPR isolates a player's offensive impact: scoring efficiency, assist creation, offensive rebounding, and avoiding turnovers. It is normalized per 100 possessions and adjusted for teammate and opponent quality. Higher OBPR means the player meaningfully improves the offense when on the court.

DBPR captures defensive impact through steal rates, block rates, defensive rebounding, and defensive on-court influence. It is harder to measure than offensive impact because defense shows up less clearly in box scores. The Bayesian component is especially important for stabilizing defensive estimates with limited data.

WAB measures resume quality by comparing a team's actual results to what would be expected from a hypothetical bubble-caliber team playing that exact schedule. A positive WAB means the team outperformed bubble-level expectations; negative means they underperformed. WAB accounts for location (home/away/neutral) and the strength of each opponent.

SOS expresses schedule difficulty as the expected winning percentage of an average D1 team if they played that exact slate of games, accounting for home/away/neutral location for each contest. A lower SOS percentage means harder opponents — an average team would struggle to win many of those games. Higher SOS means easier competition.

SOR evaluates resume quality: given this team's exact win-loss record and schedule, what is the probability that a strong reference team would match or exceed it? A high SOR rank means the team's record is impressive relative to schedule difficulty. SOR rewards beating tough opponents and penalizes losses to weak ones.

Win probability is generated from Macfax's efficiency-based matchup model before each game. It incorporates both teams' adjusted offensive and defensive ratings, home-court advantage, and tempo context. The model outputs a probability for each outcome — not a guarantee — and reflects uncertainty in the prediction.

The projected spread is the expected margin of victory based on the efficiency matchup. Positive values favor the home team; negative values favor the away team. The spread is derived from efficiency differentials and adjusted for home-court advantage. It is not intended as a betting line — it's a model estimate with inherent uncertainty.

The projected total is the expected combined final score of both teams. It is calculated from each team's offensive efficiency and the opponent's defensive efficiency, scaled by the expected number of possessions (determined by both teams' tempo ratings). A higher tempo game generates more possessions and a higher projected total.

Winner accuracy tracks how often the team with the higher predicted win probability actually won the game. All predictions are locked as snapshots before games are played — no postgame adjustments. A coin flip would yield ~50%. Good college basketball models typically land in the 68–74% range over a full season.

Spread MAE measures how far off the projected margin was on average, in absolute value. An MAE of 9.5 means the model was off by 9.5 points on average. Vegas lines on college basketball typically run 8.5–9.5 MAE. Signed margin error (positive or negative) reveals directional bias — whether the model systematically over- or under-predicts margins.

Brier score measures the accuracy of probability predictions, not just win/loss picks. It is calculated as the squared difference between the predicted probability and the actual outcome (1 = win, 0 = loss). Lower Brier scores mean better-calibrated probabilities. A perfect predictor scores 0.0; a coin flip scores 0.25.

The Trapezoid of Excellence plots every team simultaneously using AdjO on the X-axis and AdjD on the Y-axis (inverted, so better defense is higher). Teams are color-coded by their bracket seeding or tier. The "trapezoid" shape outlines the realistic envelope of team quality — very few teams are both elite offensively and defensively, creating the characteristic shape.

The Efficiency Landscape provides a comprehensive visual snapshot of a team's offensive and defensive profile across the Four Factors. Each factor is displayed as a bar relative to the national average, making it easy to see where a team excels or struggles. It's designed to answer: "How does this team win games?" at a glance.

The Crystal Ball simulates the NCAA tournament thousands of times, using Macfax's matchup model to determine game-by-game probabilities. Each simulation produces a bracket winner; aggregating across simulations yields each team's probability of reaching the Round of 32, Sweet 16, Elite Eight, Final Four, and winning the championship.

The Cinderella Index combines statistical factors (efficiency ratings, seeding gap, recent performance) with contextual signals to estimate how likely a team is to make a surprising run in the NCAA tournament. Higher scores mean a team has the profile of a historical Cinderella: respectable efficiency, underseeding, and a favorable path.