If you look at the past 10 tournaments, Top 25 teams in the luck metric are 47-71 (39.8%) in the tournament. If you're high in the luck metric, it basically means that you're over seeded.
Here are all of the champions in KenPom history. '03 Syracuse and '14 UConn are the only ones in the top 25. And almost assuredly they weren't top 25 pre-tournament. 11/20 champs fall within the 76-150 range (15/20 are in the 51-200 range). Your luck rating post-tournament is going to be decent, considering you likely won 3-5 close-ish games. But, you likely don't have a top 25 luck rating, because it more or less means your success isn't likely to be duplicated.
2002 Maryland 104
2003 Syracuse 16
2004 UConn 140
2005 UNC 135
2006 Florida 194
2007 Florida 110
2008 Kansas 138
2009 North Carolina 195
2010 Duke 145
2011 UConn 58
2012 Kentucky 86
2013 Louisville 230
2014 UConn 23
2015 Duke 38
2016 Villanova 114
2017 North Carolina 85
2018 Villanova 225
2019 Virginia 62
2021 Baylor 80
2022 Kansas 88
You can't keep winning 50-50 games - if someone was a superior team, they would have simply won by more points and the game wouldn't have been decided in the final minute or two. Now, I acknowledge some teams are better coached and have more talent, so it's not like every team should be expected to win 50% of close games. The luck metric is looking at the standard deviation between a team's winning percentage and their expected record, using multiple analytics.
4 of the 6 losses for Kansas are by 14+ points. And KU won 9 of the 10 games by 5 points or less. It basically means, if you simulated all 31 games again using a widespread of analytics, more often than not, Kansas is much more likely to have 23 or 24 wins than they are to have 26 or 27.
