How Should You Bet March Madness, and What Do Seed Records Actually Tell You?

Betting March Madness and filling in a bracket are different games that use the same 63 results. A bracket pool rewards you for picking winners, usually more than your friends pick. A bet rewards you only when the price you took was better than the true chance. A No. 1 seed that wins by 30 helps every bracket in the pool. It helps a bettor only if the line on that team was wrong.

The same distinction governs how to use seed history. The NCAA's first-round records by seed go back to 1985. They are a useful base rate: how often, across four decades of different teams, did the lower seed win? They are not a price for this year's game, which the market sets from this year's teams. The rest of this guide covers what the records say, how futures differ from game lines, why the first weekend is its own market, and what changes in 2027.

What do the seed records actually say?

From NCAA.com's compilation of the official records, first round only, 1985 through 2025:

Matchup	Higher seed W-L	Higher seed win rate	Fair price at that rate
1 v 16	158-2	98.8%	about −7900
2 v 15	149-11	93.1%	about −1355
3 v 14	137-23	85.6%	about −596
4 v 13	127-33	79.4%	about −385
5 v 12	103-57	64.4%	about −181
6 v 11	98-62	61.3%	about −158
7 v 10	97-62	61.0%	about −156
8 v 9	77-83	48.1%	about +108

The 7 v 10 record excludes a 2021 game ruled a no-contest under COVID-19 protocols. First Four games are not included.

The fair-price column is conversion, not advice. For a favourite, American odds = −(p ÷ (1 − p)) × 100. For the 5 v 12 line: 0.644 ÷ 0.356 = 1.81, so about −181.

Two updates from the 2026 tournament. NCAA.com's running tallies now put No. 1 seeds at 162-2 against No. 16s. The two losses remain UMBC over Virginia, 74-54, in 2018, and FDU over Purdue, 63-58, in 2023. No. 12 seeds are 58-106 against No. 5s after High Point beat Wisconsin 83-82. That puts the 5 seeds at 106 ÷ 164 = 64.6%.

Why isn't the 5 v 12 record a betting system?

Because the market already knows it, and because it averages over very different games.

The popular story is that 12 seeds are underrated, so you should back them. The records show 12 seeds winning about 35% of the time. Whether that makes a 12 seed a good bet depends on its price this year. If the book has the 12 seed at +200, the implied probability is 100 ÷ 300 = 33.3%. That is close to the historical rate before accounting for margin. At +150 the implied probability is 100 ÷ 250 = 40%, well above it.

And "5 v 12" is not one matchup. Some years it's a strong at-large team from a big conference against a small-conference champion that went 30-3. Other years it's a bubble-quality 5 seed against a dangerous 12. The historical average mixes both kinds. This year's line prices the specific teams. When the two disagree, the line has more information about the game in front of you.

Seeds are also set by a committee using the NET and other measures (see our college basketball betting guide). They are not the market's own ratings. A mis-seeded team is interesting, but the betting line usually reflects the team's strength, not its seed.

How do futures differ from game lines?

A futures bet asks who will win the whole thing, or reach the Final Four, or win a region. A game line asks about one game. The pricing differs in two ways that matter.

Futures carry a bigger margin. With dozens of teams in the market, each price can be padded a little, and the implied probabilities across the field add up to far more than 100%. The same arithmetic makes first-basket bets expensive in the NBA; our first basket guide works an example.

A title is a chain of games. Take a team you rate highly, with illustrative win probabilities for its six games: 95%, 80%, 65%, 55%, 55%, 50%. Its title chance is the product:

0.95 × 0.80 × 0.65 × 0.55 × 0.55 × 0.50 = 7.47%

The fair price at 7.47% is (1 − 0.0747) ÷ 0.0747 × 100 = about +1238. If the book offers +800, the implied probability is 100 ÷ 900 = 11.1%. The book's price is worse than your own estimate, even though you rate the team highly. A book can shade a popular team's future because that is where casual money is likely to go, so check the price rather than assume the team is value because you like it.

Your money is tied up. A futures bet placed in November settles in April. The stake does nothing in between. Hedging a live future as it advances is a separate skill. The Academy's guide to hedging covers the arithmetic.

Settlement rules matter too, and they change. DraftKings' current Massachusetts house rules (implementation date 26 August 2025) settle "team to advance further" markets as a push if both teams go out in the same round, so the stake comes back. The version dated August 2024 treated college basketball differently: it applied dead-heat rules, under which a two-way tie returns half the projected payout. On a $100 bet at −110 that is $95.45, a small loss, where a push refunds $100. Read your book's current rules before betting a two-team head-to-head future.

Why is the first weekend its own market?

Because of the volume. The round of 64 is 32 games across a Thursday and a Friday, then 16 more on Saturday and Sunday. Many of the teams are small-conference champions that most bettors and some bookmakers watched rarely in the regular season.

That creates three effects.

Unfamiliar teams. A book's line on a small-conference champion rests on less information than its line on a high-major team. That makes it more likely to be wrong, in either direction.

Recreational money. The tournament draws people who rarely bet the rest of the year. If their money leans toward famous programs and popular upset picks, a book can shade those prices, and a big-name favourite can cost slightly more than its strength warrants. That is a reason to compare prices, not a rule to fade names.

Injury information. For the 2026 tournament the NCAA required public availability reports, the first time it had done so. Teams filed by 9 p.m. local venue time the night before each game and updated by two hours before tip. Each player was listed as available (more than a 75% chance to play), questionable (up to 75%) or out. Reports were posted on ncaa.com, with institution fines of up to $10,000 for a first violation, $25,000 for a second and $30,000 for a third or later, when a head coach could also be fined up to $10,000. For bettors, the two-hour update is the key moment in the first-weekend news cycle.

What changes in 2027?

The NCAA announced on 7 May 2026 that both tournaments grow from 68 to 76 teams from 2027. The men's tournament starts Tuesday 16 March 2027. The Tuesday and Wednesday games expand from two per day in Dayton to 12 games in all: three a day in Dayton and three a day in Wichita, Kansas, which the NCAA named as the second host in July 2026. The NCAA calls these games the Opening Round. The 12 lowest-seeded automatic qualifiers play in six of them. The 12 lowest-seeded at-large teams play in the other six. The round of 64 stays on Thursday and Friday, the round of 32 on the weekend.

For bettors that means more early games between closely matched teams, which tend to have tight lines. It also means more of the teams on the bottom seed lines will arrive in the round of 64 having already won a tournament game, often two days earlier. The historical seed records above were built on the 64-team and 68-team brackets. Whether they carry over to a 76-team field is something only the next few tournaments will show. Nobody should claim to know yet.

The short version

Use seed records as a sanity check on base rates, not as a price. Compare every line with your own estimate of this game. Treat futures as expensive, slow-settling parlays. And expect the first weekend to be the most crowded, information-poor stretch of the college season.