Pointing NCAA Brackets Differently

Pointing models NCAA tournament brackets are almost exclusively based on how many winners of the 63 games you correctly pick. Even when a round multiplier is applied (e.g., Round 1 winners are 1 point each; Round 2 winners are 2 points each, etc.), more winners result in more points: however, just picking which teams make the Final Four doesn’t offset massively screwing up the rest of your bracket.

No pointing models I know of or found online reward bravery in selecting lower-ranked teams when completing your bracket: the risk could damage your bracket’s viability if the favored team won. Anyone picking UMBC over Virginia in 2018 would be considered certifiably nuts until, for the first time in the NCAA Men’s Tournament, a #16 defeated a #1. And who actually picked UMBC? Players, families, friends and alums who dreamed of the unprecedented. Their reward for such bravery (insanity): a single point. Meh.

Introducing ….

The points received for correctly picking a winner is determined by round and degree of upset. For those mathematically-inclined, the per-game formula is:

points_awarded = (1 + MAX(winner_seed - loser_seed, 0)) * ROUND

where ROUND is 1-4 for regionals, 5 is the national semifinals, and 6 is the championship game. MAX(winner_seed - loser_seed, 0) takes the difference of the seeds and, when positive, adds to the points awarded.

For those whose math skills required cheatsheets to pass the tests, here’s how the points for regional matchups are determined: the points awarded are found at the intersection of the winner’s seed (first column) and the loser’s seed (second row). The green-shaded columns indicate the favored team won; the peach-shaded columns are for upsets.

Note that the only upset worth less than any non-upset is when #9 defeats #8 in Round 1 (2 points). A #5 over #4 upset and #13 over #12 upset awards the same four points as the favored team winning the regional. Otherwise, points awarded for upsets are larger than the points awarded for any other non-upset, sometimes substantially.

Going Chalk

You often hear pundits talk about going chalk, choosing the higher-seeded team for whatever game under discussion, which might seem the safer approach when filling out your bracket. So how many points are available if you strictly pick favorites?

Ignoring play-in and First Four games for simplicity and consistency, each of the four regions has sixteen teams seeded 1 to 16. When only favored teams are selected, the points are based solely on round without an upset bias:

Regional Round 1: 8 games worth 1 point each = 8 points
Regional Round 2: 4 games worth 2 points each = 8 points
Regional Semifinal: 2 games worth 3 points each = 6 points
Regional Final: 1 game worth 4 points = 4 points

The minimum points available for each regional is 26 points, or 104 points for all four regionals.

No regional upsets means no Final Four upsets as only #1 seeds are present, therefore the points available are also solely based on round without an upset bias:

National Semifinals (Round 5): 2 games worth 5 points each = 10 points
National Championship (Round 6): 1 game worth 6 points = 6 points

The minimum points available for the Final Four is 16 points.

So picking only favorites AND the favorites won every game, the maximum available points is 120 points. As we’ll see, that’s not great.

Real World, Please.

OK, so let’s use the 2002 South Regional, won by #5 Indiana, which had a fairly-typical five upsets:

#13 UNC Wilmington defeating Southern California in Round 1;
#10 Kent State defeating Oklahoma State in Round 1;
#10 Kent State defeating #2 Alabama in Round 2;
#5 Indiana defeating #1 Duke in the Regional Semifinal;
#10 Kent State defeating #3 Pittsburgh in the Regional Semifinal.

The points available are shown in RED:

Someone’s bracket that correctly picked all fifteen games in the regional would receive 83 points, more than three times the points than a regional without any upsets. A bracket picking just the three Kent State upsets and striking out on every other game still receives 46 points. It’s the upsets, baby!

Across all regionals and the Final Four, 291 total points were available in 2002. The range of available points when applying an upset bias is between 163 points (2007) and 336 points (1985).

Blow My Mind

Upsets occur regularly – something I’ll address in subsequent posts – but without an upset bias, there’s less impetus to pick upsets: sure, you miss one or two, but in the end if you’ll be fine if you pick more winners than anyone else.

An upset-biased pointing model requires more skill and judgement to get at least a couple of upsets correct, as upset points add up quickly: a bracket with Kent State reaching the Regional Final has at least 16% of the total points available in 2002 (46 of 291).

What’s Next?

In subsequent posts, I’ll dig deeper into the data for both the Men’s and Women’s tournaments and see what we can learn. Did you know that 1985 had the only perfect regional, the largest-pointed game, and the largest-pointed tournament? Perhaps I can show you something that will be useful when you complete your 2025 bracket!