Early Season College Basketball and Small Samples

The first weeks of a college season produce the most confident looking numbers and the least trustworthy ones. A guard averaging 22 after three games against overmatched opponents has told you something, just much less than the average suggests.

By SlatelinePublished
A wide, flat distribution slowly narrowing as more games are added to it

Opening day is the point of maximum ignorance. Rosters have turned over, new players have not played a minute together in a game that counts, and the first results will come mostly from games that were scheduled to be uneven. The men's season starts today, Sunday, Nov. 1, with Notre Dame against Villanova in Rome, per NCAA.com's season guide, and within a week there will be player averages everywhere that look precise and mean very little.

This is not a reason to ignore November. It is a reason to be clear about which parts of the early evidence carry information and which parts are mostly the schedule talking. The college basketball hub sets out why the college game is structurally different from the NBA. This piece covers the specific problem of the first few weeks.

In November, the schedule is the sample

Early season schedules are built around nonconference games, and many of them pair programs of very different strength on purpose. Wikipedia counts 364 schools across 31 conferences in Division I men's basketball for 2025 to 26, which leaves a lot of room between the top and the bottom. A strong program's first three opponents can be weaker than almost anyone it will face in conference play.

Against those opponents, the stars play fewer minutes because games are settled early, while their per minute numbers are inflated because the defense in front of them is overmatched. The two effects pull in opposite directions, so a per game average can look normal while both of its ingredients are distorted. Reversed, a player on the weaker side of those games sees a defense far better than his conference opponents and plays heavy minutes in a game his team is losing badly.

The schedule is not only mismatches. NCAA.com's guide lists a series of early season multi team events through November, which tend to produce the first meaningful games between comparable teams. Those games deserve far more weight than the scrimmage level results around them.

  • Sort each early game by opponent strength before reading any player line from it.
  • Read minutes from competitive games only. Minutes in a blowout tell you about the blowout.
  • Read per minute production with an opponent adjustment, or not at all.
  • Give the first games between comparable teams much more weight than their count suggests.

Last season is a prior, not a forecast

With a few games of new data, the best evidence about most returning players is still last season. The question is how much of last season still applies. Base rates are the right frame: start from what is known about players in similar situations, then move toward the individual only as far as the new evidence justifies.

For a returning player whose role is unchanged, last season's per minute rates are a strong prior and a few games in November should move them only a little. For a returning player whose role has changed, because a teammate left or a new coach arrived, last season's rates describe a different job. The prior should be weaker, and the direction of the change matters more than the size of last year's numbers.

Opponent adjustment belongs in the prior too. A player who put up big numbers in a weak conference and has moved up a level, or whose team's schedule has become harder, carries production that will not survive the move intact. The Sports Reference glossary defines a Simple Rating System that accounts for point differential and strength of schedule, and offensive and defensive ratings adjusted for opponent strength. Some version of that adjustment should sit under any number you carry forward.

Example: Carrying a prior forward, for illustration

Suppose a made up returning guard scored 0.45 points per minute last season in the same role, and has scored 0.70 per minute in three games against weak opponents this November. Weighting three soft games equally with a full season would be a mistake. A researcher who treats last season as worth many games of evidence, and discounts the November games for opponent strength, lands close to last year's rate and perhaps a little above it. The early number is a hint, not a new level. Every figure here is invented.

Transfers: known players in unknown roles

Player movement between programs is now a routine part of the college offseason. The NCAA's Division I Cabinet set the men's basketball transfer window as a 15 day period opening the day after the national championship game, which concentrates much of the movement into the spring. By November, many rosters include several players who have never played a game for their current coach.

A transfer is easier to research than a freshman because there is a college track record. The trap is carrying that track record over unchanged. A transfer's statistics were produced in a specific role, with specific teammates, against a specific level of opponent. Each of those three can change at once.

  1. Role first. Was he a primary option at his old school who is now one of several? Usage rarely transfers intact.
  2. Level second. Moving up in conference strength usually lowers efficiency, and moving down may raise it, though the size of the change varies widely.
  3. Fit third. A shooter joining a team that lacks spacing may get a bigger role than his numbers suggest. A big joining a crowded frontcourt may get less.

Early games are where you learn which of those guesses was right, and minutes and shot attempts in competitive games are the quickest signals. Shooting percentages over the same span are too noisy to tell you much.

The quieter beneficiaries: players who stayed

Every player who leaves a program leaves minutes and shots behind, and some of the most researchable early season changes belong to the players who stayed. A reserve whose minutes were capped by an upperclassman ahead of him may now start. A secondary scorer may become the first option because the primary one transferred out. His statistics from last season describe the smaller job, so both the market and a naive average tend to underrate the new one at first.

The method is the one the role changes article sets out: identify who is structurally positioned to absorb the vacated role, estimate the new share of minutes and usage, and widen the uncertainty around it rather than narrowing it. Vacated usage tends to flow to players who can create shots, not evenly across the roster, and a returning player who rarely created last season may not absorb much even with more minutes.

Freshmen: a debut with a lot of attention

Freshmen have no college sample at all, and the ones who get prop attention usually arrive with a reputation. Reputation is not a projection. Researching a debut covers the general method: role and opportunity first, base rates for similar players second, and any translated production from a lower level discounted heavily.

Two college specific points. First, freshmen minutes are often the least stable on the roster, because coaches are still learning what a first year player can handle and foul trouble hits inexperienced defenders hard. Second, early mismatches can hand a freshman big minutes in blowouts that will not recur in conference play. A freshman's November line against overmatched opponents is one of the least informative numbers in the sport.

How fast to let new games move you

Different statistics settle at different speeds, as the sample size article explains. Minutes and shot volume reflect coaching decisions and settle quickly. Rebounding and assist rates take longer. Shooting percentages, especially from three, take the longest by far and should barely move off the prior in the first month.

Tempo follows the same pattern. Early raw tempo figures are contaminated by opponents in the way described in the tempo piece, and adjusted figures lean on whatever prior the rating system started from. Both get more reliable through December as teams play more varied schedules.

The honest summary of November is that your estimates should be wide and should get narrower slowly. A projection that feels as confident in the second week of the season as in the second week of conference play is ignoring how little it knows.

Limits of early season research

Nothing in this method can see a coach's rotation plan before he uses it, and early season availability information in college is thin. A player can miss a November game for reasons that are never announced. Lines in the first weeks are also set with the same limited information you have, so an apparent gap may be nothing more than two noisy estimates disagreeing. That is the uncomfortable symmetry of early season research: the uncertainty that might make a line wrong applies to your own number just as much, and there is no way to know from the inside which of the two is further off.

References

Follow College Basketball on Slateline

The public College Basketball page lists each market with its graded record where one is published and, when a slate is open, the props the free plan opens. No account is needed to look.

Open the College Basketball page

Not ready to sign up? Get the weekly note.

The Slate

One weekly note: what published, what the graded record shows, what changed. No picks, and nothing is sent until you confirm your address.

Keep researching

Early Season College Basketball and Small Samples · Slateline