$w.onReady(function () {
top of page


Most people think that picking a bracket is pure luck or simply madness. We disagree.

Last year we found several key patterns to help find those elusive Cinderellas and avoid the big upsets.

Adding several new insights, below are several key angles and statistics to listen to, plus several others to ignore!

Don’t chase too many upsets!

Round 1

i. #1 seeds almost never lose (99% win rate)

ii. #2 seeds lose extremely rarely (93% win rate)

iii. #3 seeds lose rather infrequently (about once every 2 years)

Making the sweet 16:

i. #1 seeds make the Sweet 16 85% of the time

Two key stats to find the Cinderellas

Slow paced teams with good defenses

i. This combination allows the underdogs to limit the total amount of possessions and keep the score low. In turn this raises the chances of a close game and provides an opportunity for the underdog to steal a victory*

ii. Recent examples:

Sister Jean’s Loyola of Chicago (2018): Slowest 15% in pace + a top 20 defense

Syracuse’s 11th seeded runs (2016 & 2018): Bottom 10% in tempo & 5th/18th in defense

St. Peters 15 seed run to the Elite Eight in 2022: Bottom 30% in tempo with a top 25 Defense

iii. The reverse of this is true too: Slow paced teams are more vulnerable to upsets

38% of #1 and #2 seeds that lose in the first weekend were in the bottom 70th percentile in tempo

Betting on futures – where is the value? Championship - Final Four?

Picking teams to win the Championship

a. Eleven of the past 15 champions started the tournament with single digit odds pre- tournament (less than 10-1)

b. Only once were the odds of a Champion greater than 25-1 pre-tournament (UCONN in 2014)

c. All of the past 7 winners have come from teams ranked within the fifth highest odds pre-tournament

d. With these low payout favorites winning most of the time, there may be limited value in picking outright championship futures

Picking teams to make the Final Four

8 of the past 9 years a team with double digit odds (pays 10-1 or more) has made the Final Four

4 of these (UNC 22, UCLA 21, Loy Chi 18, Syracuse 16) had odds over 20-1!

Unlike the champions who are consistently one of the best teams, many underdogs reach the final four with big payouts for their bettors

Pay attention to conference ratings!

i. It is extremely tricky to compare teams across conferences, since most top teams only play a few competitive games outside of their conference schedule

It is useful to look at the computer conference ratings, metrics that aggregate all the inter conference games to give insight how to compare teams across conferences

ii. Our analysis found that even among power 5 conferences, a better net rating of the conference a team is in, equaled a higher chance to win NCAA tournament games

iii. Three of the past 5 champions have come from the conference with the highest NET ranking

Six of the past 20 final four participants have come from one of the top two conferences in the net rankings in their respective years

Ten of 12 teams from mid major conferences who have made the Sweet 16 in the past five years have come from conferences in the top 12 in net ratings


Two statistics many pontificators will reference as predictive are team records in neutral games and margin of victory

However, neither of these provide predictive value while picking a bracket!

We ran statistical analyses to check the predictive value of these two statistics:

A team’s margin of victory had limited predictive value on a team’s success in the NCAA tournament. This may be because:

1. A single blowout win can skew this stat

2. Consistent blowouts may be indicative of a team’s weak strength of schedule

One hypothesis is that team’s that have performed well on neutral sites are more likely to perform well in the NCAA tournament, which is always played on a neutral site

However, we found there is a stronger correlation between wins at home and in the tournament than a neutral site

This can likely be attributed to the fact teams simply do not play enough neutral site games (generally ranging from 3-6) to make any concrete inferences from


bottom of page