Consistency In College Basketball Statistics: Conference-Level Variability Profiles

The second installment of this three part series

Sound of Confusion

Now that we've explored how box score statistics vary across Division I, it's time to take a closer look.

Part one used the coefficient of variation (CV) to measure consistency and found that certain metrics, like game pace and effective possession ratio, remain relatively stable across the season. In contrast, composite stats such as points per shot, along with steal and block rates, show much higher variability.

But national trends only tell part of the story. College basketball is a collection of conferences, each with its own rhythm, defensive identity, and structural quirks. What holds true across 11,000 games can shift dramatically when viewed through the lens of individual leagues, non-conference games, or roster composition.

Clear patterns emerge when comparing each conference’s overall average consistency to its overall winning percentage.

The most consistent conference in the country this season was one highlighted earlier in the year: Conference USA. The league showed minimal separation between its top and bottom teams, creating an environment that naturally supports consistency. The Atlantic 10, by contrast, lacks C-USA’s parity, but with nearly twice as many teams, sheer volume appears to dampen extreme statistical swings across the board.

The top-tier conferences tell a different story. The sport’s two best-performing leagues this season showed only slightly less variability overall, while many smaller or lower-ranked conferences landed among the most statistically erratic.

One high-major outlier was the Big East, which recorded the third-highest average coefficient of variation among all conferences. Several notable programs, including Villanova, Providence, and Butler, ranked among the 20 most inconsistent teams. Georgetown and DePaul also finished in the bottom 50. Nearly half of the league’s 11 teams showed extreme inconsistency, and none of them came close to making the NCAA Tournament.

Southern Birds
No discussion of conference trends in 2025 is complete without examining the SEC, a 16-team powerhouse that may be the strongest the sport has seen.

The diverging box plot above shows that the SEC is on average more consistent on a game to game basis. This doesn't mean the SEC outperformed other conferences in free throw rate or block percentage. Instead, those stats showed less fluctuation across games.

On a composite level, the conference stands out in three-point metrics, with greater consistency in percentage, rate, and attempts. The SEC didn't shoot especially well from deep, finishing 20th out of 31 conferences at 33.7%. Still, the stability of their shooting may have contributed to their success.

Three of the four least consistent variables relate to defense: defensive rating plus, opponent points, and opponent points per shot all varied significantly. This contrasts with the steadiness in opponent three-point shooting. Inside scoring, however, was more volatile. The SEC ranked near the middle nationally in two-point percentage at 51.6 percent.

Two-point shooting carries more context than threes. Teams can't fully prevent open threes, but they can design schemes to limit interior scoring. Twos fluctuate more because their quality and volume depend heavily on matchups, tempo, and shot selection.

Salem Sisters
Rather than rely on heatmap color patterns, we can now directly compare average values across specific stats. Converting each CV to a z-score, which normalizes it within its statistical category, lets us identify smaller-scale discrepancies with greater precision.

Along with the SEC, the Patriot League (5th in three-point percentage) and the Southern Conference (14th) show strong consistency in several three-point shooting metrics. The Sun Belt, SWAC, and WAC form another group with similar shooting profiles. However, these conferences were not only inconsistent from the perimeter but also ranked in the bottom five in three-point accuracy.

As shown in the average CV graph earlier, the SWAC stands out as the least consistent conference in the country and also holds the lowest winning percentage. The heavy presence of red in their row reinforces this point.

By evaluating each statistic individually, we can create variability profiles for every conference. Instead of relying on visual cues from a heatmap, we can measure the difference between each conference's average across multiple stats. Taking the average of these differences allows us to group conferences into “sister conferences” based on statistical variability.

This doesn't suggest that two conferences would perform at a similar level, but rather that they tend to show comparable levels of consistency, even with different overall averages. For example, among power conferences, the ACC most closely aligns with the Patriot League in terms of statistical consistency.

Unsurprisingly, the most polarizing conferences in terms of overall CV, as shown in the second figure, have statistical profiles that are hard to replicate. The SWAC and C-USA, the most and least variable conferences respectively, were the furthest from the consistency profile of the power conferences.

The SWAC was so polarizing, relative to the ACC, that the gap between its profile and the next most similar one matched the difference between the first and 17th most consistent profiles.

Contextual factors help explain why the SWAC stands out as a statistical outlier. Its 11-122 non-conference record is a clear contributor to that inconsistency. With a .0827 winning percentage over the first two months of the season and a .500 mark in conference play, SWAC teams faced a wider range of opponents than any other conference.

Unsurprisingly, the NEC, which finished with a 28-81 record in non-conference play, showed a similar level of statistical inconsistency to the SWAC. Using the same CV value, the next most comparable sister conference would rank in the ACC’s bottom five for season-long consistency. That the Big East and Mountain West appear in the SWAC’s top five sister conferences says more about the SWAC’s uniqueness than any true similarity.

More is Less
In this case, it's easy to link inconsistency with lower win percentages. However, the correlation shown in the first figure suggests that, on a broader scale, there is no clear relationship between statistical inconsistency and overall success.

By taking a closer look, we can explore how specific variables correlate with performance on a conference-by-conference basis. With an understanding of which conferences are most consistent, which perform best, and which stats are most or least stable, we can begin to connect more pieces of the puzzle.

Since a higher CV reflects greater inconsistency, a negative z-score means that a higher conference win percentage is associated with more consistent game-to-game performance for a given statistic. The five most statistically significant variables all had p-values below 0.005, indicating a strong relationship with win percentage.

The strongest correlation emerged between conference win percentage and consistency in points per shot, highlighted by the most successful conference also being the most consistent in this metric, and the least successful being the least consistent. Notably, the SEC and SWAC ranked as the most and least consistent in points per shot, and also the most and least effective.

Other highly correlated statistics also reflect scoring efficiency, with some redundancy across metrics that tell a similar story. Points per shot, as a composite measure, best captures how consistent offensive performance relates to winning.

On the other hand, raw counting stats show weak consistency, likely because they aren't adjusted for volume, pace, or other factors that can obscure meaningful differences. Two teams might share the same raw numbers, while their pace of play tells a completely different story.

In the third and final part, we'll explore how individual teams within each conference contribute to statistical consistency. By looking beyond conference lines and filtering out the noise from inconsistent teams, we can better understand how consistency connects to performance. The puzzle is only getting more complex.