Glossary

The total volume of bets placed on a market or across multiple markets. Sportsbooks balance action across sides to manage risk, and sharp action refers to bets from professional bettors that often move lines.

A football statistic measuring the average distance downfield at which a receiver's targets occur or a quarterback's passes are thrown. Used in prop modeling to assess deep-ball tendencies and expected yards per reception.

In football, the distance a pass travels in the air from the line of scrimmage to the point of reception or incompletion, independent of yards after catch. Used in prop modeling to separate quarterback arm strength from receiver run-after-catch ability.

Alternative Over/Under or count-based prop lines offered at different thresholds (e.g., Over 23.5 points instead of Over 24.5), allowing bettors to achieve better prices or synthetic equivalents.

The standard odds format used in North American sportsbooks. Negative odds (e.g., -110) indicate how much you must risk to win $100. Positive odds (e.g., +150) indicate how much you profit on a $100 wager. All pricing analysis begins with converting odds to implied probability.

A derivative prop bet where you wager on whether a player scores a touchdown at any point during the game. Typically offered as a "Yes" option only in a one-way market with high vig, making it one of the highest-margin props for sportsbooks.

A betting term referring to picking winners based on the point spread, often used in the context of traditional sports betting performance tracking. ATS records measure how well a team covers the spread, not just wins outright.

The limitation sportsbooks face in specializing deeply in every player or prop due to the high volume of markets. This structural constraint creates inefficiencies that sharp bettors can exploit — books simply cannot dedicate the same analytical resources to every prop they post.

A game flow where one team dominates, leading to predictable prop outcomes like increased garbage time usage for backups. Blowout scripts systematically alter prop distributions by changing the opportunity structure mid-game.

Short for sportsbook — the entity that sets lines, accepts wagers, and pays out winnings. Used colloquially by bettors (e.g., "beat the books" or "the book moved the line"). Different books serve different roles: some are market makers, others copy lines from sharper sources.

The minimum win percentage required to produce an EV of exactly $0 at a specific price. At standard -110, break-even is 52.38%. Any true win rate above this threshold represents +EV.

The 5–10 second lag in TV broadcasts compared to real-time data feeds used by sportsbooks. This delay can disadvantage live bettors reacting to events they see on screen, as books have already repriced based on real-time information.

A sportsbook feature allowing a bettor to settle an active wager early for a guaranteed payout (often less than the full potential win), based on the current game state. Used to lock in profit or minimize losses, especially in live betting or hedging scenarios.

The final odds or line available just before an event begins, after all information and betting action have been incorporated. Consistently beating the closing line (positive CLV) is a strong indicator of betting skill and process quality.

The gold standard for measuring betting edge. CLV compares your entry price to the final market price before a game begins. Consistently beating the closing line is the single strongest predictor of long-term profitability — it proves you are capturing value the market has not yet priced in.

The process of removing the sportsbook's built-in margin (vig) from the implied probabilities to calculate the true or fair odds for each side. Enables bettors to compare their own probability estimates to market-implied fair probabilities — the foundation of edge detection.

An odds format common in Europe and on betting exchanges, expressed as a multiplier (e.g., 2.50) that represents total return per unit wagered, including the original stake. Easier for calculating payouts and implied probabilities than American odds: IP = 1 / decimal odds.

Bets derived from main stats, such as "Anytime TD" or "First Basket." Often carry high variance and are priced in one-way markets with elevated vig — making them simultaneously appealing to recreational bettors and profitable territory for sharps who can price them accurately.

A form of contest-based fantasy sports where participants build lineups for single-day or short-term events. Skills overlap significantly with prop betting — model building, projection, and understanding player usage all transfer directly.

The advantage a bettor has over the market, typically measured as the difference between your estimated true probability and the market's implied probability after vig. Without a quantifiable edge, every bet is a negative-sum proposition.

The structural inefficiency in live betting markets due to real-time volume overwhelming pricing precision. This gap creates exploitable windows for informed bettors who can process information faster than the book can reprice.

The mathematical advantage (or disadvantage) of a wager over an infinite sample size. A +EV bet has a positive expected return; a -EV bet is a losing proposition long-term. EV is the single most important concept in quantitative sports betting.

The total amount of risk a sportsbook or bettor has on a particular outcome, market, or across a slate. Books manage exposure by adjusting lines, while bettors manage exposure through position sizing and diversification across uncorrelated props.

The true or no-vig odds for an outcome after removing the sportsbook's margin, reflecting the actual probability without built-in profit. Calculated using de-vigging methods to enable accurate edge detection — the benchmark against which you measure your own probability estimates.

The flow of a game (e.g., blowout, close contest, comeback) that influences prop outcomes. Increased passing in comebacks, garbage time rushes in blowouts, and pace changes all systematically shift player stat distributions mid-game.

Late-game situations in blowouts where starters rest, leading to increased opportunities for backups and exploitable prop edges. Garbage time creates systematic distortions in box scores that must be filtered from projection models.

The total dollar volume of bets placed on a market. Main markets (spreads, totals) drive most handle and receive more sportsbook analytical attention. Prop markets, with lower handle, receive less scrutiny — creating the inefficiency gap that sharp prop bettors exploit.

Placing additional bets to reduce risk or lock in profit, often using negatively correlated props to offset potential losses. While hedging reduces variance, it also reduces expected value — making it a risk management tool, not an edge-generation strategy.

The sportsbook's built-in profit margin — the excess implied probability over 100% across all sides of a market. At standard -110/-110, the hold is ~4.76%. Prop markets often carry holds of 6–10% or more, especially in one-way markets.

The mistaken belief that a player experiencing a short-term streak is more likely to continue succeeding. Despite its intuitive appeal, evidence shows this is largely a product of small-sample noise and regression to the mean — a common trap in live betting decisions.

The probability embedded in a given price. Converting odds to implied probability strips away the format and reveals what the market is actually saying about the likelihood of an outcome. The formula differs for positive and negative American odds.

The sportsbook's built-in profit margin — the excess implied probability over 100%. At standard -110/-110 pricing, juice equals 4.76%. Synonym for vig, vigorish, and hold. Prop markets typically carry higher juice than main markets.

Last-minute information — injuries, weather changes, lineup decisions — that impacts props but creates repricing lags. Fast-acting bettors who process late-breaking news before books can adjust capture some of the most reliable edges in prop betting.

The potential amount a sportsbook stands to lose on a particular outcome or bet. Books manage liability by adjusting lines to balance action or by limiting sharp bettors who consistently win — a key reason why consistent CLV attracts restrictions.

The maximum bet size a sportsbook will accept on a given market. Props typically have much lower limits than main markets because they are harder to price accurately. Books use limits to protect themselves from sharp bettors — getting limited is often a sign you are winning.

The practice of comparing odds across multiple sportsbooks to secure the best available price. It is the single easiest way to improve profitability without developing better predictions — pure arbitrage of market friction.

The amount of betting volume available in a market. High-liquidity markets (main lines at major books) are tightly priced and efficient. Low-liquidity markets (niche props, exchanges) may offer wider spreads but also more opportunity for mispricing.

Placing wagers after a game has started, with odds updating in real-time based on game flow. Live prop markets are among the most inefficient in sports betting due to the speed at which books must reprice — creating the efficiency gap that informed bettors exploit.

Proposition bets offered and updated in real-time during a game, with odds adjusting based on current score, time remaining, and game flow. Generally less efficient than pre-game props due to speed and volume constraints on sportsbook pricing algorithms.

The primary betting markets for a game — typically the point spread, moneyline, and total (over/under). These receive the most betting volume, sharpest action, and tightest pricing from sportsbooks, making them significantly more efficient than prop markets.

Short-horizon bets on immediate outcomes, like "next play result" or "next basket scorer." Carry extremely high variance and situational dependence, with pricing that is often algorithmically generated and less refined than standard pre-game props.

A hedging strategy where bets are placed on both sides of a line (often pre-game and live) such that an outcome landing between the two lines results in winning both bets. Creates a low-risk profit window but requires careful timing and line movement.

A bet on which team will win outright, with no point spread. Odds are expressed in American format (e.g., -150 favorite, +130 underdog) and represent the price to risk or win relative to $100. One of the three main markets alongside spread and totals.

The cumulative cost of consistently betting at slightly worse odds — even 5 cents ("a nickel") of line difference. Over hundreds of bets, the nickel tax compounds into significant EV losses, making line shopping one of the highest-ROI habits in betting.

Fun, non-serious props like coin toss results, Gatorade color, or national anthem length. Often carry extremely high vig and are treated as entertainment rather than edge opportunities. No analytical framework can reliably price most novelty bets.

A prop structure where only one side is offered (e.g., "Yes" for touchdown scorer), with hidden and typically much higher vig — often 20–40%. The absence of a corresponding "No" side makes it impossible to directly observe the true margin.

The first odds or line posted by a sportsbook for an event, typically released days before the event. Sharp bettors target opening lines when they have an information edge before the market fully adjusts through subsequent action.

A two-way prop bet where you wager whether an outcome will be above (Over) or below (Under) a set line. The most common prop structure, allowing for direct probability modeling using statistical distributions like normal, Poisson, or negative binomial.

A combined bet on multiple outcomes, paying out only if all legs win. Same-game parlays (SGPs) are common in props and require correlation adjustments — multiplying individual probabilities together only works if legs are truly independent.

The Professional and Amateur Sports Protection Act — a U.S. federal law struck down by the Supreme Court in 2018, leading to widespread legalization of sports betting and the explosion of prop markets that created the current landscape of opportunity.

A sportsbook known for offering sharp, efficient lines with low vig, high limits, and a policy of not limiting winning bettors. Often used as a benchmark for line comparison and considered the sharpest book in the market — the closest proxy for "true odds."

A wager on a specific in-game occurrence not directly tied to the final outcome, such as player stats or game events. Prop markets are structurally less efficient than main markets due to the bandwidth problem — creating the exploitable edges this course is built around.

A tie outcome in a bet — when a player's performance exactly matches the line (e.g., exactly 24 points on an Over/Under 24). The stake is returned with no win or loss. Half-point lines (e.g., 24.5) eliminate the possibility of a push.

A casual bettor focused on entertainment rather than edge detection, often exhibiting predictable biases like favoring Overs, star players, and high-profile games. Recreational money pushes lines and creates systematic value on the opposite side for sharps.

Predictable tendencies of casual bettors — favoring Overs, stars, home teams, and parlays — which systematically push lines and create value on the opposite side. Understanding recreational bias is a structural edge that never fully disappears from prop markets.

A shift in a player's usage (e.g., backup becoming starter, new offensive scheme) that creates prop edges if not fully priced by books. Role changes are among the most exploitable situations in prop betting because books are slow to adjust projections.

A parlay combining multiple outcomes from the same game. Pricing SGPs correctly requires understanding the correlation structure between legs — most books exploit bettors by assuming independence when legs are positively correlated, systematically overcharging for SGPs.

Professional bettors who seek and exploit edges through models, data, and discipline. Sharp action moves lines toward efficiency. Sportsbooks often limit or restrict sharp bettors — being limited is paradoxically a badge of honor that confirms your edge.

Wagers placed by sharp bettors, which move lines toward efficiency. Sharp money is identifiable by its timing (often early), size (max or near-max limits), and the subsequent line movement it triggers across multiple books.

In football, the number of offensive plays a player was on the field for — used as a proxy for opportunity and playing time. Critical for projecting props like receptions, targets, and rushing attempts, especially after role changes or returning from injury.

A handicap applied to the favorite in a game, creating a near-even betting proposition. The favorite must win by more than the spread, and the underdog must lose by less (or win outright) for the bet to cash. One of the three main markets.

A live betting line that has not yet fully adjusted to new information — an injury, game script change, or momentum shift. Stale lines create temporary edge windows that informed bettors can exploit before the book reprices.

A sudden, significant line movement across multiple sportsbooks caused by heavy sharp action on one side. Steam indicates that respected money has entered the market, often due to new information or a sharp betting syndicate. Steam chasers attempt to follow these moves.

Equivalent bets constructed using alt lines, derivatives, or combinations to replicate a desired wager at better odds or prices. Building synthetic positions is an advanced line shopping technique that expands the universe of available edges.

The absolute best available odds or line across all sportsbooks at a given time. Consistently finding top-of-market prices through line shopping is one of the simplest ways to improve long-term profitability without changing your handicapping.

A bet on whether the combined score of both teams will be over or under a set number. One of the three main markets alongside spread and moneyline, with props being individual or team-specific variations of the same concept.

The actual likelihood of an outcome occurring, independent of market prices or sportsbook margins. Bettors estimate true probability through models and data to identify edges when the market-implied probability differs from their estimate.

A prop structure offering both Over and Under sides, typically with balanced vig (e.g., -110/-110). Two-way markets are more transparent than one-way markets because the vig is visible and can be directly measured and removed via de-vigging.

The transaction cost or "hold" built into the market by the sportsbook. Also called juice or margin. Standard vig on sides is ~4.5%, but prop markets often carry 6–8% or more — making them simultaneously harder to beat and richer in inefficiency.

The percentage of bets won over a sample, calculated as Wins / Total Bets. While important, win rate alone does not determine profitability — bettors must also account for odds and vig, making expected value and ROI more accurate measures of long-term success.

The total capital allocated exclusively to betting operations. In Wall Street terms, this is your trading capital. Professional bettors treat their bankroll as a portfolio — never risking more than a calculated fraction on any single position.

The process of sizing bets appropriately to avoid ruin while maximizing growth. Uses frameworks like Kelly Criterion, fractional Kelly, and flat betting to maintain long-term sustainability — emphasizing that survival is the prerequisite for profitability.

A period of losses due to negative variance, even when betting with a positive edge. Downswings are part of normal statistical fluctuation and are mathematically guaranteed over any sufficiently long betting career — the question is not if but when and how deep.

A peak-to-trough decline in bankroll value. Even with a verified +EV strategy, drawdowns of 20–40% are mathematically inevitable over hundreds of bets. Understanding drawdown is essential for bankroll survival and psychological resilience.

A bankroll management strategy where you wager the same fixed amount (typically 1–2% of bankroll) on every bet regardless of perceived edge size. Simpler than Kelly-based approaches, flat betting sacrifices some growth optimization for reduced complexity and lower variance.

A bankroll management approach where bettors wager a fraction (commonly 1/4 to 1/2) of the amount recommended by the full Kelly Criterion. Significantly reduces volatility and risk of ruin while retaining most of the growth rate — preferred by professionals due to edge estimation uncertainty.

The original Kelly Criterion formula that maximizes long-term bankroll growth by betting a percentage of bankroll equal to edge divided by odds. Highly aggressive and requires perfect edge estimation — leading most bettors to use fractional Kelly instead to account for model uncertainty.

A specific fractional Kelly approach where bettors wager 50% of the amount recommended by the full Kelly Criterion. Captures approximately 75% of full Kelly's growth rate while reducing volatility by 50%, offering a balanced risk-reward profile for experienced bettors.

Situations where a bettor has access to or processes information (e.g., late injury news, advanced metrics) faster or better than the sportsbook. Information asymmetry is the fundamental mechanism behind most betting edges — you know something the market has not yet priced in.

The optimal position-sizing formula derived from information theory: f* = (bp - q) / b. Kelly calculates the exact fraction of your bankroll to wager to maximize the geometric growth rate of capital. Most professionals use fractional Kelly (1/4 to 1/2) to reduce drawdown risk.

The degree to which available information is already reflected in the current odds. Main markets (spreads, totals) are highly efficient. Prop markets, especially player props, are significantly less efficient — creating exploitable edges for bettors with superior information or models.

The discipline of determining how much capital to allocate to each individual bet. Proper position sizing — informed by Kelly Criterion and confidence level — is the difference between long-term compounding and ruin.

A betting philosophy emphasizing the quality and consistency of decision-making rather than short-term outcomes. Winning bettors focus on making +EV bets with sound reasoning, trusting that results will follow over time despite variance. Judge the process, not the scoreboard.

A conservative fractional Kelly approach where bettors wager 25% of the amount recommended by the full Kelly Criterion. Minimizes volatility and risk of ruin while still achieving steady growth — suitable for beginners or those with lower risk tolerance and less confidence in their edge estimates.

The probability of losing your entire bankroll given your edge, bet size, and variance profile. Even a +EV bettor using oversized stakes can face near-certain ruin — making proper position sizing non-negotiable for survival.

Net profit divided by total capital risked, expressed as a percentage. A 5% ROI means $500 profit on $10,000 wagered. Elite prop bettors typically sustain 3–8% ROI over large sample sizes. ROI is a more complete measure than win rate because it accounts for odds.

Losing emotional control after losses, leading to poor decisions like chasing losses, increasing bet sizes, or abandoning your model. Countered by discipline, bankroll management rules, and the understanding that downswings are statistically inevitable.

A standardized bet size, typically defined as a fixed percentage (e.g., 1–2%) of a bettor's total bankroll. Using units rather than dollar amounts allows for consistent bet sizing relative to bankroll changes and simplifies performance tracking across different bankroll sizes.

A starting projection or reference point before adjustments, typically derived from historical averages or simple models. Used as the foundation for building more refined estimates that account for matchup, role, pace, and other contextual factors.

A discrete probability distribution modeling the number of successes in a fixed number of independent yes/no trials, each with the same probability of success. Used in betting to calculate win rate probabilities over a sample of bets and assess whether observed results are due to skill or variance.

Describes players whose performance is highly variable, leading to overdispersion in stats like touchdowns or three-pointers. Outcomes cluster in extremes (zero or multiple) rather than averaging out consistently — requiring negative binomial rather than Poisson modeling.

A function that gives the probability that a random variable is less than or equal to a specific value. In prop betting, the CDF is what you use for Under bets — P(X ≤ line). In Excel: NORM.DIST(..., TRUE) and POISSON.DIST(..., TRUE) return cumulative probabilities.

The upper end of a player's realistic range of outcomes in a given game, representing a high-probability best-case scenario. Used alongside floor to establish the bounds for prop modeling and risk assessment — understanding ceilings helps identify value in alt lines.

The probability of one event occurring given that another event has occurred — e.g., P(WR touchdown | QB 2+ passing TDs). Used to adjust for correlation in same-game parlays where legs are not independent.

A range of values within which a parameter (e.g., a player's true mean or a correlation coefficient) is likely to fall with a specified probability (commonly 95%). Reflects estimation uncertainty and sample size limitations — wider intervals demand more conservative position sizing.

An adjustment of ±0.5 applied when using a continuous distribution (like normal) to approximate a discrete outcome. For an Over 24.5 prop, you model P(X > 24.5) directly; for Over 24 (whole number), use P(X > 24.5) to account for the discrete boundary.

Advanced statistical models for combining marginal distributions of correlated variables while preserving their dependence structure. Used for joint probability calculations in multi-leg prop parlays where standard correlation coefficients are insufficient to capture the full dependency.

A standardized measure (−1 to +1) of how two variables move together. In SGP pricing, positive correlation between legs means the parlay is worth more than the independent product of individual probabilities implies. r > 0.4 is considered strong in betting contexts.

A measure of the directional relationship between two variables. Unlike correlation, covariance is not standardized — its magnitude depends on the scale of the variables. Used in the raw computation of SGP adjustments and as the basis for calculating correlation.

In the negative binomial distribution, a parameter that reflects a player's inherent boom-or-bust nature. Estimated as r = μ² / (Variance - μ), where lower r indicates more overdispersion. Critical for correctly modeling props where variance significantly exceeds the mean.

An advanced football metric created by Football Outsiders that measures a team or player's efficiency relative to league average, adjusted for opponent strength. Used in prop modeling to assess matchup difficulty and adjust baseline projections.

An advanced football metric that quantifies the value of a play by measuring the change in expected points before and after the play. Used to evaluate player and team efficiency beyond traditional box score stats — a more predictive signal than raw counting statistics.

The higher probability of extreme outcomes in distributions like negative binomial compared to normal or Poisson. Fat tails mean that "boom" games (e.g., 3+ touchdowns) and "bust" games (zero) are more likely than standard models predict — ignoring them leads to systematic mispricing.

The lower end of a player's realistic range of outcomes in a given game, representing a high-probability worst-case scenario. Used alongside ceiling to establish risk bounds and assess downside in prop modeling — understanding floors helps identify when Unders offer value.

A statistical model for count data that separately estimates the probability of zero versus the distribution of positive outcomes. Useful for props with structural zeros (e.g., a player scripted out) where the zero-generating process differs from the count-generating process — more flexible than ZIP.

When two events or props have no relationship (correlation ≈ 0), allowing probabilities to be multiplied directly for joint outcomes. True independence is rare in same-game contexts — most prop legs within the same game share some degree of correlation.

The probability of two or more events occurring together, adjusted for correlation. For dependent events: P(A and B) = P(A) × P(B|A). For independent events: P(A and B) = P(A) × P(B). The foundation of correct SGP pricing.

The expected count parameter in a Poisson distribution. For a player averaging 0.8 touchdowns per game, λ = 0.8. Lambda simultaneously represents both the mean and variance of the distribution — a property unique to Poisson.

The average value of a statistic (e.g., average assists per game). Used as a core parameter in distributions like Poisson and normal. The mean is the starting point for all prop projections — but must be supplemented with variance information for complete modeling.

The middle value in a dataset when ordered from lowest to highest, such that half of observations fall above and half below. Less sensitive to outliers than the mean and useful for skewed distributions common in prop betting — especially boom-or-bust stats.

The most frequently occurring value in a dataset. In prop betting, useful for identifying the most common outcome in boom-or-bust stats like touchdowns, where the mode (often 0) may differ significantly from the mean (e.g., 0.8).

An extension of Poisson that adds a dispersion parameter to handle overdispersed count data (where variance exceeds the mean). Critical for props like receptions and assists where player usage varies significantly — produces fatter tails than Poisson.

When two variables move in opposite directions (r < 0) — e.g., RB1 carries and RB2 carries on the same team. Useful for reducing portfolio variance through diversification and for identifying natural hedges within same-game parlays.

Random variation or irrelevant information in data that obscures true patterns (signal). In betting, separating noise from signal is critical for building accurate models and avoiding overreaction to small samples — most short-term results are noise.

The bell curve. A continuous probability distribution defined by mean (μ) and standard deviation (σ). Used for modeling continuous stats like passing yards, fantasy points, and total points. The foundation of Z-score analysis and the most commonly used distribution in prop betting.

When the variance of count data exceeds its mean (VMR > 1), indicating more variability than a Poisson model can capture. Ignoring overdispersion leads to systematically overconfident probability estimates — a common and costly mistake in prop modeling.

The speed of play in a game, influencing the number of opportunities (e.g., possessions in basketball, plays in football). Pace directly affects prop projections — a fast-paced game environment inflates counting stats, while slow pace suppresses them.

A function that gives the probability of each specific discrete outcome. In prop betting, the PMF answers "what is the exact probability of exactly k events?" In Excel: POISSON.DIST(..., FALSE) returns the PMF for a specific count.

A discrete probability distribution for rare, independent count events (e.g., touchdowns, goals, strikeouts) where variance equals the mean (VMR ≈ 1). Defined by a single parameter λ (lambda) — the expected count. Ideal when events are rare and independent.

When two variables move in the same direction (r > 0) — e.g., QB passing TDs and WR receiving TDs. Positive correlation increases joint probabilities in parlays beyond what independent multiplication would suggest, making correlated SGP legs more valuable.

An estimated or forecasted value for a player's performance in an upcoming game, typically based on historical data, matchup factors, and contextual adjustments. The foundation for comparing against market lines to identify edges — the core skill of prop betting.

The spread of possible outcomes for a player's performance, bounded by a realistic floor and ceiling. Understanding a prop's expected range helps assess risk, avoid overconfidence in point estimates, and identify value in alt lines.

The statistical tendency for extreme performances (high or low) to be followed by outcomes closer to a player's historical average, due to variance rather than skill changes. Counters biases like the hot hand fallacy and is essential for separating signal from noise in small samples.

The number of observations or trials in a dataset. Larger samples reduce noise and increase confidence in estimates, while small samples lead to high variance and unreliable conclusions. A common pitfall in sports betting is drawing conclusions from insufficient data.

A measure of dispersion in a dataset, calculated as the average squared deviation from the mean (divided by n-1 for sample). The numerator of the VMR calculation and a key input for determining whether Poisson or negative binomial modeling is appropriate.

Meaningful information or true patterns in data that can be used to make better predictions. In betting, extracting signal from noise is essential for building edges and avoiding false confidence from randomness — the higher the sample size, the clearer the signal.

A computational method for modeling outcomes by generating many random trials based on probability distributions. Used in prop betting to estimate probabilities for complex scenarios, test strategies, and validate models when closed-form solutions are impractical.

The square root of variance — a more intuitive measure of spread expressed in the same units as the data. In a normal distribution, ~68% of outcomes fall within ±1 SD of the mean, and ~95% within ±2 SD. The key to converting projections into probabilities.

In the negative binomial distribution, a derived parameter calculated as p = r / (r + μ). Used alongside the dispersion parameter r in probability calculations for overdispersed count data — a higher p value indicates less overdispersion.

A test statistic used to determine if a correlation is statistically significant (not due to random noise). Calculated as t = r × sqrt(n - 2) / sqrt(1 - r²). Helps distinguish real relationships from spurious correlations in small-sample prop data.

In football and basketball, the percentage of a team's total targets or shot attempts directed at a particular player. Used to project volume-based props like receptions, receiving yards, or points — a more stable and predictive metric than raw counting stats.

In basketball, the percentage of team possessions a player uses while on the floor. In football, the percentage of team plays involving a specific player. Used to project opportunity-driven props — higher usage correlates with higher counting stat output and lower variance.

The measure of how far a set of results spreads out from their average value. In betting, variance is the "noise" that obscures your true edge. High-variance strategies require larger bankrolls and longer time horizons to realize theoretical returns.

The diagnostic test for overdispersion. VMR = Variance / Mean. VMR ≈ 1 suggests Poisson is appropriate; VMR > 1 indicates overdispersion and the need for Negative Binomial modeling. The single most important diagnostic before choosing a count distribution.

In football, the yards gained by a receiver after catching the pass, distinct from air yards. Used in prop modeling to separate receiver ability to break tackles and gain additional yards from quarterback accuracy and pass location.

The number of standard deviations a value lies from the mean. Z = (X − μ) / σ. Used to convert any normally distributed variable into a standardized probability. A Z-score of +1.65 corresponds to the 95th percentile.

A two-component model combining a Poisson distribution with a point mass at zero. Used for props where a player has a meaningful probability of zero participation (injury, game script), producing excess zeros beyond what standard Poisson predicts.