Draft the minutes, not the names

A fantasy projection has two halves. One of them is well studied and the other is mostly guessed at.

The first half is a rate: what a player produces per minute on the floor. It is stable, it is cheap to estimate, and it changes slowly. A player who rebounded 0.20 times per minute last season will rebound at roughly that rate this season unless his position or his role changes.

The second half is minutes. Minutes are volatile, they are decided by a coach in October, and they move by 30% or more when the rotation ahead of a player is cleared out. Almost every difference between a good draft and a bad one lives in that second half.

Most draft boards bury it inside a single ranking number. This is a method for pulling it back out.

Your ranking is not a ranking

Settle what game you are playing before anything else, because the three common formats value the same player differently enough that a shared top-50 list is close to useless.

Format	What you are maximising	What it punishes
Head-to-head categories	Winning more of nine weekly categories than your opponent	Being average everywhere
Rotisserie	Season-long standing in each category against the whole league	Any single category left unattended
Points	One weighted total	Almost nothing, except low volume

Nine-cat leagues score field goal percentage, free throw percentage, three-pointers made, points, rebounds, assists, steals, blocks and turnovers. Eight-cat drops turnovers, which quietly raises the value of every high-usage guard in the pool.

The consequence is concrete. A centre who shoots 64% from the field and 52% from the line is a strong asset in head-to-head, where you can lose the free throw category in a given week and still win the matchup 6-3. In roto he is a season-long drag on a category you cannot escape, because the standing accumulates across 82 games rather than resetting every Monday.

Points leagues flatten all of it. A turnover costs a fixed amount there and a missed free throw costs nothing at all, so the same centre is simply a volume asset.

If you are reading a ranked list that does not say which of those three it was built for, you are reading someone else's league.

The minutes-first method

Build your own numbers in four steps. The arithmetic is deliberately simple, because the error in a fantasy projection is almost never in the multiplication.

1. Take a per-minute rate, not a per-game average. Per-game averages already contain last season's minutes, which is the thing you are about to change. Divide each category by minutes played and keep the rate.

2. Write down a minutes projection, in ink. One number per player. This is the step people skip, and skipping it is what turns a draft board into a list of vibes.

3. Multiply, then add a games-played assumption. A player at 32 minutes across 70 games and a player at 32 minutes across 58 games are not the same asset in roto, where counting categories accumulate. They are much closer in head-to-head, where a missed week is one bad matchup.

4. Say out loud which minutes you are unsure about. Then sort the board by how far the projection moves if you are wrong.

A worked illustration, using two hypothetical players so the arithmetic stays visible:

Case	Rebounds per minute	Projected minutes	Projected per game
Player A, base	0.21	34	7.1
Player A, low	0.21	26	5.5
Player B, base	0.27	22	5.9
Player B, high	0.27	30	8.1

Player A is the better rebounder on the box score and the worse one per minute. If the rotation breaks the way Player B's high case assumes, Player B is the better pick, and nothing about either player's ability changed. That gap is the whole argument for doing this by hand.

The habit worth building: when a ranking surprises you, ask what minutes number it assumed before you ask whether it rates the player correctly.

What the 2026 offseason actually did

Roster movement matters in fantasy for a narrower reason than it matters in real basketball. A trade becomes interesting to a manager mainly when it changes who is on the floor and who ends possessions, and those two effects often point in opposite directions.

Hoopspire's 2026-27 projection lays out the movement: Giannis Antetokounmpo to Miami, Jaylen Brown and LeBron James to Philadelphia, LaMelo Ball to Minnesota, Ja Morant to Portland, Kawhi Leonard to Toronto, Paul George to Boston, with Kevin Durant staying in Houston alongside Marcus Smart and Bogdan Bogdanovic.

Read that list twice, from two directions.

Arriving stars compress the players around them. Miami adding Antetokounmpo to a team already built around Bam Adebayo puts two high-usage interior players into the same possessions. The projection expects more rim pressure and more free throw attempts for Miami as a team, and treats the spacing cost as the open question. For fantasy the question narrows further: the shooters and secondary creators who were absorbing those possessions last season are now worth less than their 2025-26 lines suggest. Philadelphia has the same problem twice over, with Brown and James both arriving.

Departing stars are where the free minutes are. Milwaukee, Charlotte, Memphis and the Los Angeles Clippers each lost a primary option. Those vacancies are the most reliable source of value in a draft, and they are systematically underpriced, because a player's rank is computed from a season in which somebody else was taking his shots.

The discipline is to be specific about which vacancy you are buying. Usage left behind by a departing star does not distribute evenly. It goes disproportionately to the players who already had the ball in their hands, which means the third option on a bad team is usually a worse bet than the second option looks.

Watch the other direction too. A player whose team improved sharply can lose value in roto even while his team wins more, because a rotation that is winning comfortably plays its starters fewer fourth quarters.

Rookies

Rookies are the one group where the rate half of the projection is missing entirely. There is no NBA per-minute production to work from, so everything rests on the minutes projection, which is the volatile half. That is a bad combination, and it is why rookies tend to go about a round earlier than they should.

Three filters, in order of how much they are worth:

Where is he drafted, and what is ahead of him? A lottery pick on a rebuilding roster with no established starter at his position is a different asset from an equally talented player behind two veterans. Depth chart beats draft slot.

Does his game produce fantasy categories, or does it produce winning? These are not the same thing. Defensive forwards who guard the other team's best player, set screens and take open threes are valuable in basketball and thin in nine-cat. High-usage guards who dominate the ball produce assists and turnovers, which cuts both ways depending on whether your league counts turnovers.

What does he do to your free throw line? A rookie big who shoots 58% from the line moves your team's free throw percentage across a whole season. In roto that is a category you then have to repair somewhere else.

Cooper Flagg, now in his second season, is the useful counterexample. Hoopspire's projection expects Dallas to improve without a full transformation from a 26-56 team, which describes a player whose minutes are already secure. A second-year player with established minutes has both halves of the projection available, and that is worth more than the upside premium a genuine rookie carries.

The general rule: draft one rookie you believe in, late, and let the waiver wire hand you the others in November once the rotations have declared themselves.

Players who did not play last season

This group is the largest source of genuine mispricing in a draft, and the most dangerous.

The mispricing is real because ranking systems built on last season's totals either exclude a player who missed the year or rank him near the bottom. Either way the number in front of you is wrong, and a manager willing to do the work can buy a starter in the eighth round.

The danger is that the reason he missed the season has not gone away. Split the group before you bid.

Missed for a reason with a known recovery curve. A single structural injury with a documented timeline is the most predictable case. The rate half of the projection usually returns close to where it was, and the minutes half returns on a ramp: restricted minutes early, no back-to-backs, a real role by December. In head-to-head that ramp costs very little, because early matchups are worth the same as late ones and you can stream around him. In roto it costs real counting stats, because November is roughly a sixth of your season and it never comes back.

Missed for a reason with no clear curve. Recurring soft-tissue problems, a load-management history, or a situation that was never fully explained. The honest projection here is not a lower number, it is a wider one. Price him by what he does to your roster if he plays 45 games rather than by his ceiling.

Missed for a non-medical reason. A contract situation, a trade standoff, a suspension. The body is fine. The role is the question, and the role is usually answered by the transaction that resolved it.

In all three cases, write the games-played assumption next to the name before the draft starts. A manager holding "58 games" on paper is much harder to talk into a fourth-round pick than a manager holding "he is a first-round talent" in his head.

Punting is a plan, not an accident

Punting means deliberately conceding a category to concentrate value in the others. It is a head-to-head strategy, and it is close to indefensible in roto, where every conceded category is a fixed loss in the standings that no amount of excellence elsewhere recovers.

If you punt free throw percentage in a nine-cat head-to-head league, you are targeting a 5-4 or 6-3 weekly win and accepting a guaranteed loss in one column. That makes low free throw percentage bigs cheap for you and expensive for everyone else, which is the actual source of the edge.

Two conditions make it work. Commit before the draft rather than discovering the punt in round seven, and leave one category you can still win in a bad week. A roster built on two conceded categories has no slack.

The draft-day list

What to have written down before the first pick:

• A minutes number and a games number for every player you would take.
• Your format, in one sentence, and the three or four categories you intend to win.
• The five vacancies you most believe in, drawn from departing usage rather than from arriving talent.
• One rookie, and the round you will not reach past.
• For every returning player, the games-played figure you will hold yourself to.

None of that requires a projection system. It requires writing the assumption down where you can see it while somebody else is on the clock.

The names in a draft room are mostly the same for everyone. What separates two boards is whether the minutes behind those names were reasoned about or inherited, and inherited minutes are last season's answer to a question the offseason already changed.