Picking your Final Four Bracket

For the last two years I've written a post around March Madness time displaying some interesting trends about the percentage of seeds advancing in the tournament. Here are the trends from the past 26 seasons as well as the results that followed last year.

In the first round:

· #12 seeds upset the #5 seeds 34% of the time (last year it was 25% with #12 Richmond beating #5 Vanderbilt)

· 21% of #4 seeds go down (#4 Louisville was upset last year)

· Meanwhile only 15% of #3 seeds lose their first game (last year all four #3 seeds advanced)

In the second round:

· More #5 seeds make the sweet 16 than #4 seeds, (15 to 10) (25 year range is closer with #4 seeds making it 43-36 times, but last year backed it up with #5 Butler and #5 MSU advancing while only one #4 seed Purdue made it)

· Only 12% of #1 seeds were upset (Has happened two years in a row though)

· Meanwhile 37% of #2 seeds have been upset by the #7/#10 winner (#10 Florida State upset #2 Notre Dame last year)

· Only or 13% has a region gone #1, #2, #3, #4 (Which did not happen again last year)

Sweet 16 round:

· 71% of #1 seeds make the elite eight (Was only 25% last year when only #1 Kansas survived)

· On the other side of the region, 71% are either #2 or #3 seeds (75% last year)

Elite Eight round:

· Only one time in 31 years of the 64 team bracket has all four #1 seeds made the Final Four (Which was no exception last year)

· 42% of #1 seeds make the Final Four (No #1 seeds made it last year)

· #2 seeds make it 21% of the time (No #2 seeds made it last year as well)

· If you add up all the seeds that make the final four, the average total is 10.5. Meaning if you pick all #1 seeds, that’s less than half the average, or if you throw in a #11 George Mason with a #3, #2, and #1, that’s double the average. (Last year’s total of 26 (#3, #4, #8, #11) was more than double the average)

While last year followed these patterns pretty closely, it does not mean I picked the correct upsets nor won my pool. Boo. ESPN must have read this post earlier, since they have a version as well.