Running a league people stay in

A single named award each week, announced in the same place every time, is the cheapest engagement tool there is. It gives people who are out of contention something to win and something to lose.

The trick is that at least one award has to be bad. An award for the highest score is a lap of honour for whoever is already winning; an award for the most points left on the bench is a running joke anybody can enter and nobody wants.

Name them after people in your league rather than after the stat. The Marcus Rule outlives Most Bench Points by about three seasons.

Standings reward luck. A 5-1 team that scored the fourth-most points in the league is not the best team, and everybody knows it without having a way to say so that sticks.

Rankings that account for points scored, points against and schedule give the league an argument every week, and arguments are engagement. Publish them with one line of reasoning per move.

Being ranked below your record is far more motivating than being ranked at it.

Pick the four matchups with genuine history and name them in the preseason. A week that has been on the calendar since August is one people plan around.

The pairings do not need to be close on paper. Two people who have argued for years make a better rivalry than two good teams who have never spoken.

Champion, last place, biggest bust, whose season falls apart first. Collected once, in writing, before anybody knows anything.

The value is entirely in the retrieval. A prediction nobody revisits is a form. A prediction read back in week twelve is the best thirty seconds of the year.

Ask for a reason alongside each answer. A name on its own cannot be argued with. A name plus because he traded his whole draft is a claim that either holds up or does not.

Everyone picks a winner for every matchup, including their own. Hidden until the deadline, then scored and ranked.

Hiding them until lock is the part that matters. Visible picks turn into everybody copying the consensus, and a leaderboard of the consensus is not worth reading.

Let people leave one line of reasoning on a pick. That line is what gets read back to them.

Straight after the draft, while people still have opinions, ask each manager what they make of everyone's roster. Two sentences is plenty.

Collect it in the window where it is cheap to give. A week later the same question gets nothing back.

Collusion, tanking, unresponsive managers, veto thresholds. Every one of these is a calm conversation in August and a fight in November.

Put a date on the document and do not change it mid-season. A rule that moves after a dispute looks like it moved because of the dispute.

The most common way a league dies is one person who never pays and one commissioner who does not want to chase it.

Whatever the amount, make it due before the draft rather than before the playoffs. Somebody who has not paid by week thirteen has already stopped caring.

Last-place punishments work when they are public, finite, and agreed before the season. They fail when they are invented in December by eleven people at the expense of one.

The good ones cost time or dignity rather than money. Whoever came last has already lost enough money.