The NCAA Is Expanding March Madness to 76. The Arithmetic Says Go to 365.

The question: the NCAA is expanding March Madness from 68 teams to 76. What would it actually cost to invite everybody?

The answer. Invite all 365 Division I teams, seed them off ten games, and run double elimination. The tournament gets smaller and fairer at the same time.

Why that is surprising. Everyone assumes a bigger field means more chaos. It is the reverse; a second life makes the tournament more reliable, not less.

The NCAA has confirmed the change. From 2027 there are eight more at-large bids and a twelve-game “Opening Round” replacing the First Four. After that the familiar bracket of 64 carries on as before.

It is a strange size to land on; big enough to annoy everyone who liked 64, small enough to change nothing.

So here is a better question. What would it cost to invite everybody?

The Shape of the Alternative

Picture it. All 365 Division I teams play a ten-game opening slate, every game televised; those ten games do one job, which is to seed the bracket.

Then everyone enters a double-elimination tournament. Two losses and you are out.

Not a group stage; those turn the last games into sums about point difference. Just: lose twice, go home.

Run it as two brackets if you like. An elite division for the revenue programs, a second for everyone else; the numbers barely move, and we will come back to that.

The Counting

A double-elimination tournament of N teams takes 2N − 2 games. Every team but the champion has to lose twice, so the count rises in a straight line with the size of the field.

For 365 teams that is 728 games.

Against the 76-team bracket’s 75.

Now watch the counting, because this is the part that surprises people.

Count what you already watch. A Division I season runs 32 regular-season games per team, plus 31 conference tournaments, plus the bracket; about 6,245 games. Of those, 75 end somebody’s season. That is 1.2%.

The alternative plays 1,825 seeding games and 728 elimination games. 2,553 in all.

That is 59% fewer games than the sport plays now, with almost ten times as many that end somebody’s season.

So we would not be adding basketball. We would delete about 3,700 low-stakes games and turn the survivors into games that matter.

The Objection, and Why It Is Backwards

Your first reaction is probably the same as everyone’s: invite 365 teams and the tournament becomes a coin-flipping contest.

It is the opposite, and the reason is simple.

Single elimination is the most random format ever built. One bad shooting night ends a great season. Double elimination makes you beat a team twice, and two bad nights is a much higher bar than one.

Simulate three thousand seasons. Under the 76-team single-elimination bracket, the best team in the country wins the title 45% of the time.

Open it to all 365 with two lives each and that rises to 55%. The champion’s median true rank improves from second-best in the country to best.

Notice what the extra 289 teams actually do: they lose and go home, mostly in the first week. What adds order is the second life.

Two Things That Fall Out of the Maths

Splitting into two divisions is free. Because the game count is linear, an 80-team elite bracket plus a 285-team second bracket costs 726 games against 728 for one combined field.

A difference of two. Whatever you think of splitting the revenue programs off, the sums do not object.

Ten games is not enough to seed anything, and that is a feature. A ten-game record tracks true team strength at 0.87. That is genuinely useful.

But it names the single best team in the country as the top seed only 4% of the time. Picking one maximum out of 365 from ten noisy games is close to hopeless, and you should not want it to be otherwise.

Which settles a design question. If seeding is that unreliable, chasing it is a poor use of the opening slate.

So a coach could spend those ten games sorting a rotation instead of stacking wins, and give up almost nothing. Under the current format, where one bad night ends everything, no coach can afford that.

Under the current format, where one bad night ends everything, no coach can afford that. The new format would create a real choice where today there is only one sane answer.

Which quietly settles a design question. If seeding is that unreliable, chasing it is a poor use of the opening slate — so a coach can rationally spend those ten games sorting a rotation instead of stacking wins, and give up almost nothing. Under the current format, where one bad night ends everything, no coach can afford that. The format would create a real strategic choice where today there is only one sane answer.

What It Costs

Now the price, honestly.

A champion would play about 14 tournament games, up to 17 through the losers’ bracket, against seven today. Call it six weeks rather than three; you would be watching a different sport in March.

Total games for sale fall by more than half; that is real money. That is a real problem for any program that leans on ticket money from twenty home dates. And the ten-game slate would gut the money that keeps a lot of small athletic departments afloat.

Those are real objections, and we should take them seriously. None of them is “it would be too random” — the one everybody reaches for first, and the only one the numbers actually refute.

What to take home: every tournament is a measuring tool

The NCAA is expanding because more teams means more games to sell. Fine.

But if the argument for 76 is that more access is good, the argument does not stop at 76. It stops wherever someone decided to stop it. And 76 is a number chosen to disturb as little as possible.

So the lesson we take home is about formats, not basketball. Every tournament is a measuring tool, and its design decides what it measures.

Single elimination measures your best day; nothing more. Double elimination measures how steady you are. Round robin measures quality.

None is more correct than the others. But a sport ought to know which question it asks, and nobody has ever made college basketball choose.