Wednesday, March 26, 2014

# Bracket Patterns

The opening four days of the 2014 NCAA were about as good as it gets in the world of sports. Upsets, injuries, overtimes, and Cinderella stories highlight the headlines of this year's tournament. And for some, the excitement of the tournament was heightened by the offer of $1 billion to any fan who predicted a perfect bracket. As I'm sure you have heard, the chances of this are something like 1 in 9 quintillion (though I'd submit that so long as you're not an idiot, the chances are closer to 1 in 7 quadrillion) and nobody even produced a clean round of 64 (the chances of which are about 1 in 4 billion — maybe as high as 1 in 500 million ... if you are not an idiot).

All this got me thinking about perfect brackets, what it would take, and if there were any patterns over the past few years that might lead to advice on how to go about producing a bracket that resembles a typical, real bracket. And is there even such a thing?

A few notes before we get started though. I fully expect Warren Buffett to offer another $1 billion next year. I fully expect that nobody will win. I do not think anybody will ever produce a perfect bracket, but I honestly and sincerely hope that somebody does some day. This is not to help you produce a perfect bracket. This is to enlighten you about recent tournament history and help you produce a better bracket, one that at least fits the pattern of a real tournament bracket.

**Upsets**

There are always upsets. How many is an interesting thing. In the round of 64, here is the breakdown over the past five tournaments (I don't count 9 seeds over 8 seeds as upsets, more on that later):

2014: 7

2013: 8

2012: 9

2011: 6

2010: 8

So between 6 and 9 upsets on Thursday and Friday (maybe between 5 and 10 to be lenient). If you pick 12 upsets in the round of 64, that's too many. If you pick 4, that's too few. Another issue is choosing the biggest upset in the tournament.

**Highest Seed Upset in the Round of 64**

2014: 3 – Duke (by 14 – Mercer)

2013: 2 – Georgetown (by 15 – Florida Gulf Coast)

2012: 2 – Missouri (by 15 – Norfolk State); 2 – Duke (by 15 – Lehigh)

2011: 4 – Louisville (by 13 – Morehead State)

2010: 3 – Georgetown (by 14 – Ohio)

This is really where I think most people have trouble in their brackets. They are simply unwilling to pick one big upset. But truly, it's a necessity. Everybody loves 12s over 5s, but the pattern is that those are not enough. And if the last 5 tournaments are any indication, you should probably choose a successful program to be upset as Duke, Louisville, Georgetown are tough programs.

**8 vs. 9**

I hate choosing 8 vs. 9 more than any other match-up in the bracket. I constantly second guess myself. The 8 seed is 13-7 in the past five tournaments, but truly there is no discernible pattern. Good luck.

**Lowest Seed in the Sweet 16**

If there were no upsets, the lowest seed in the Sweet 16 would be a 4. Yeah, right. Here's the breakdown:

2014: 11 – Dayton and Tennessee

2013: 15 – Florida Gulf Coast

2012: 13 – Ohio

2011: 12 – Richmond

2010: 12 – Cornell

Another angle is:

**Number of Seeds Lower Than 4 in the Sweet 16**

2014: 6/16 (11 – Dayton; 11 – Tennessee; 10 – Stanford; 8 – Kentucky; 7 – UConn; 6 – Baylor)

2013: 5/16 (15 – Florida Gulf Coast; 13 – La Salle; 12 – Oregon; 9 – Wichita State; 6 – Arizona)

2012: 5/16 (13 – Ohio; 11 – North Carolina State; 10 – Xavier; 7 – Florida; 6 – Cincinnati)

2011: 6/16 (12 – Richmond; 11 – VCU; 11 – Marquette; 10 – Florida State; 8 – Butler; 5 – Arizona)

2010 8/16 (12 – Cornell; 11 – Washington; 10 – St. Mary's; 9 – Northern Iowa; 6 – Tennessee; 6 – Xavier; 5 – Butler; 5 – Michigan State)

So we are looking at 5 to 8 teams seeded 5 or below making it to the Sweet 16 (or perhaps more helpfully, 5 to 6 teams seeded 6 and below making it to the Sweet 16).

**Sweet 16 Sums**

Stick with me on this next point. If you add up the total number of seeds for each team that made the Sweet 16, the numbers you get from year to year are interesting. If there were no upsets, you would have 40 (1+1+1+1+2+2+2+2+3+3+3+3+4+4+4+4).

2014: 79

2013: 81

2012: 73

2011: 80

2010: 80

Personally, I find that fascinating. So next year, add up the seed numbers of the teams you are thinking of putting in the Sweet 16. If it's not between 70 and 85, try something different.

I won't break down the Elite Eight quite as far as the Sweet 16.

**Lowest Seed in the Elite Eight**

2013: 9 – Wichita State

2012: 7 – Florida

2011: 11 – VCU

2010: 6 – Tennessee

In 2014, this will be either a 10 or 11 (10 Stanford takes on 11 Dayton on Thursday and 11 Tennessee faces 2 Michigan on Friday). This is personally where I struggle the most. I have a hard time putting a team in the Elite Eight that isn't ranked in the AP top 25. But the tournament history shows one very unexpected team will make a run. Choose wisely. If your Elite Eight is all 1s, 2s, and 3s, think again.

**The Final Four is a Bit More Variable**

2013: 1, 4, 4, 9 = 18

2012: 1, 2, 2, 4 = 9

2011: 3, 4, 8, 11 = 26

2010: 1, 2, 5, 5 = 13

There's not a terribly discernible pattern there. 2011 was really an oddity seeing the first Final Four without a 1 or a 2, and on top of that having an 11, tying the lowest seed to make the Final Four.

**What About Conferences?**

Over the past five NCAA tournaments, no conference has had more than four teams in the Sweet 16. Very generally speaking, about half (or below) of a conference's teams in the tournament will make the Sweet 16. But even in 2011 when the Big East had 11 teams in the tournament, only two of their teams made the Sweet 16.

What does this mean? Don't expect any conference to do too well. This year's SEC having three teams in the tournament (Florida, Kentucky, Tennessee) all make the Sweet 16 is a rarity.

So let me summarize what every tournament will look like for the foreseeable future:

* Between 5 and 10 round of 64 upsets

* At least one big (seeds 2, 3, or 4) upset (possible two, probably not more than three)

* An 11 to 15 seed in the Sweet 16 (15 is a stretch; Florida Gulf Coast is the only 15 to make it that far)

* 5 or 6 teams seeded 6 or worse in the Sweet 16

* A 6 to 11 seed in the Elite Eight

* Not too much conference loyalty and bias

There you go. Don't expect this to be the key to a perfect bracket, but hopefully it helps you produce a bracket that at least stands a chance of being a perfect bracket.

Finally though, the current bracket scoring systems are stupid and I'd like to offer a counter system.

How much sense does this make? The chances of you picking the overall champion is 1 in 64 (1.56%); your chances of picking a perfect round of 64 is 1 in 4 billion (.000000025%). And yet in every round, the amount of points available remains the same: typically 32 or 320. We desperately need a new system.

What would be fair? Your odds of predicting 6/6 games correctly in the Round of 64 is the same as predicting the National Champion. So in reality, if Round of 64 games are 1 point a piece, the national championship should be worth 6 points, not 32. Breaking that down further we would find that the scoring should be more like this:

From 64 to 32 – 1 point per correct – 32 points available

From 32 to 16 – 2 points per correct – 32 points available

From 16 to 8 – 3 points per correct – 24 points available

From 8 to 4 – 4 points per correct – 16 points available

From 4 to 2 – 5 points per correct – 10 points available

From 2 to 1 – 6 points per correct – 6 points available

This probably sounds terrible to you. I admit, I'm not in love with this either. Perhaps a hybrid is in order, just splitting the difference.

From 64 to 32 – 1 point per correct – 32 points available

From 32 to 16 – 2 points per correct – 32 points available

From 16 to 8 – 3.5 points per correct – 28 points available

From 8 to 4 – 6 points per correct – 24 points available

From 4 to 2 – 11.5 points per correct – 21 points available

From 2 to 1 – 19 points per correct – 19 points available

That might work for me. What do you think?

In the end, I can say this has been one of the best NCAA tournaments so far and though I doubt anybody will be a billion dollars richer by filling out a bracket, we can always dream ... right?