QB Consistency, Scoring Volatility, and Predicting Spread Outcomes

Analysis · by Rick's Picks Analytics

Not all 35-point scoring averages are created equal. A team that scores 35 every week is fundamentally different from one that scores 52 one week and 17 the next. We tested whether this consistency shows up in spread outcomes.

Measuring Consistency

We define offensive consistency as a team's mean points scored divided by their scoring standard deviation (plus 1 to avoid division by zero). A team scoring 35.0 points per game with a standard deviation of 5 has a consistency rating of 5.83. A team scoring 35.0 with a standard deviation of 15 has a rating of only 2.19.

Why this matters for the spread: Consistent teams are more predictable, making it easier for Vegas to set an accurate line. Volatile teams create wider prediction intervals, which should theoretically benefit bettors who can identify the high or low side.

We rank every team-season into quartiles by consistency rating and compare ATS performance.

Hypothesis 1: Consistent Teams Cover More

The question: Do top-quartile consistency teams (high mean scoring, low variance) cover the spread at a higher rate and have better ATS margins than bottom-quartile teams?

The test: Welch's t-test on ATS margins between top and bottom quartile, with Cohen's d measuring effect size. Validated on training (2015-2022) and test (2023-2024) sets.

What we find: Consistent teams do show a better ATS margin than volatile teams. The intuition is straightforward: if a team reliably scores in a narrow range, the spread is more likely to land on the right side. Volatile teams introduce randomness that sometimes helps and sometimes hurts, but on average doesn't create a positive-expected-value situation.

The output: This analysis produces the qb_elite_bonus coefficient -- the ATS margin difference between top and bottom quartile teams. This becomes a direct input to our prediction algorithm.

Hypothesis 2: Volatility as an Injury Proxy

The question: Do teams with high within-season scoring volatility (a proxy for key-player injuries or suspensions) have worse ATS outcomes?

The measure: We compute volatility as (max score - min score) / average score for each team-season, requiring at least 8 home games. A team whose scores range from 10 to 55 with an average of 32 has a volatility of 1.41 -- extremely erratic.

The injury connection: When a starting quarterback goes down, scoring typically drops sharply for a game or two before recovering (or not). This creates exactly the kind of max-min spread that our volatility metric captures, without needing explicit injury data.

What the data shows: High-volatility team-seasons do tend to underperform ATS. The effect produces the injury_impact_penalty coefficient in our model. When we detect high in-season scoring volatility, we adjust our prediction downward for that team.

Hypothesis 3: Scoring Standard Deviation Correlation

The question: Across all team-seasons, is there a Pearson correlation between scoring standard deviation and season-long cover rate?

The test: For team-seasons with at least 4 home games (to get a meaningful cover rate), we compute the correlation between scoring std dev and the team's home cover percentage. We require at least 30 team-seasons for statistical validity.

What we find: The correlation is negative -- higher scoring variance associates with lower cover rates -- but the magnitude is modest. This confirms the directional finding from H1 using a different statistical approach (correlation vs. quartile comparison).

Position Impact Ratings

Beyond the data-driven consistency metrics, our model incorporates domain-knowledge position impact ratings:

• Quarterback: 9.2 -- The most impactful position. A QB injury can swing a game by 10+ points.
• Offensive Line: 7.1 -- Underappreciated but critical for both run game and pass protection.
• Running Back: 6.8 -- Important but more replaceable than QB or OL in modern offenses.
• Defensive Line: 6.2 -- Sets the tone for the entire defense.
• Wide Receiver: 5.5 -- Important but depth is usually better than other positions.
• Defensive Back: 5.2 -- Critical against spread offenses.
• Linebacker: 4.8 -- Valuable but the least game-changing of the position groups we track.

These ratings scale injury-impact adjustments. Losing a starting QB (9.2) triggers a much larger prediction adjustment than losing a starting linebacker (4.8).

A Bug We Caught

Our original consistency formula was inverted -- it penalized teams with low variance instead of rewarding them. We also found that the qb_poor_penalty was computed using abs(x) * -1, which always produced a negative value regardless of direction. Both bugs were fixed in our February 2026 refactor.

How This Enters Our Algorithm

The player impact module produces three coefficients that feed directly into game predictions: 1. Elite consistency bonus: Applied to top-quartile teams 2. Poor consistency penalty: Applied to bottom-quartile teams 3. Injury volatility penalty: Applied when in-season scoring volatility exceeds thresholds

These coefficients are combined with weather, travel, stadium, and public-bias factors to produce our final prediction.

Analysis uses Welch's t-test, Pearson correlation, Cohen's d, and Bonferroni correction across three hypotheses. Results validated on 2023-2024 holdout data.

Rick's Picks publishes statistical analysis of college football for informational and entertainment purposes. Nothing here is betting advice. 21+. If gambling is affecting you or someone you know, call 1-800-GAMBLER.