Wizard
Joined: Oct 14, 2009
• Posts: 21509
December 5th, 2013 at 8:54:21 PM permalink
I just hand entered the outcome of every March Madness game since the 64-team system started in 1985. When the props come out, I'll be rip roaring and ready to analyze them. I know a frequent one is over/under total wins in the x conference. For example, over/under 9.5 wins in the PAC 10. We don't know who is playing or the seeds yet, but this table shows the expected wins according to seed. With this, we should get a decent estimate by summing the seeds for any given conference. The average wins column shows the exact average number of wins for that seed. However, there is a problem. Note, for example, the average is higher for a six seed than five. I would attribute that to a small sample size problem. The right column smooths out the ups and down.

Seed Avg. Wins Estimate
1 3.35 3.19
2 2.41 2.39
3 1.86 1.93
4 1.53 1.60
5 1.13 1.34
6 1.19 1.13
7 0.83 0.95
8 0.71 0.80
9 0.57 0.66
10 0.63 0.54
11 0.51 0.43
12 0.53 0.33
13 0.27 0.24
14 0.16 0.15
15 0.08 0.07
16 0.00 0.00

Any other props or questions you'd like to examine while we're waiting for the props to come out?
It's not whether you win or lose; it's whether or not you had a good bet.
Mission146
Joined: May 15, 2012
• Posts: 12609
December 5th, 2013 at 9:21:34 PM permalink
Sure, at what point would you consider a Moneyline fair on a 16th seed to win a game, since it has never happened?
Vultures can't be choosers.
Wizard
Joined: Oct 14, 2009
• Posts: 21509
December 5th, 2013 at 9:24:43 PM permalink
Quote: Mission146

Sure, at what point would you consider a Moneyline fair on a 16th seed to win a game, since it has never happened?

Around 200 to 1.
It's not whether you win or lose; it's whether or not you had a good bet.
ThatDonGuy
Joined: Jun 22, 2011
• Posts: 4035
December 5th, 2013 at 10:40:07 PM permalink
You also have to keep in mind that the seeding committee is allowed to move a team one seed up or down in order to, for example, prevent two teams in the same conference from playing each other before the Elite Eight (unless the conference gets nine teams).
michael99000
Joined: Jul 10, 2010
• Posts: 2091
December 5th, 2013 at 11:22:17 PM permalink
I don't think the lower seeds having more wins than a higher seed is that unusual at all. Here's why..

They don't reseed after each round. So for example, let's say for a given region the 10 and the 12 seed both pull off upsets, while the other six higher seeds all win. In round 2 the 10 seed actually has to face a more difficult team (the 2 seed), than the 12 seed (who now faces the 4 seed).

So I'd say that if you did that table for just Round 1 wins, you get a uniform distribution of wins vs seed. But since after round 1 it's not reseeded, you can't really say that a higher seed is more likely to win in round 2 or thereafter.

Also, once you get to the 6th and 7th seeds, your talking the 21st best team, 22nd best, etc.. And I honestly don't think the tournament committe has only clue whose better than whom
Wizard
Joined: Oct 14, 2009
• Posts: 21509
December 6th, 2013 at 6:07:12 AM permalink
Quote: michael99000

I don't think the lower seeds having more wins than a higher seed is that unusual at all. Here's why..

Good point. Here is a table showing the probability of winning the first round.

Seed Wins Prob. Win Estimate
1 116 1.00 1.00
2 108 0.93 0.93
3 99 0.85 0.87
4 91 0.78 0.80
5 75 0.65 0.73
6 78 0.67 0.67
7 70 0.60 0.60
8 58 0.50 0.53
9 58 0.50 0.47
10 46 0.40 0.40
11 38 0.33 0.33
12 41 0.35 0.27
13 25 0.22 0.20
14 17 0.15 0.13
15 8 0.07 0.07
16 0 0.00 0.00

Note those 6-seeds have more wins than 5-seeds. The column on the right, again, smooths the ups and downs, for better or worse.

By the way, what is this kind of tournament called, where the best plays the worst, the second best plays the second worst, and so on? When I was in the Baltimore chess club they did the first round in the tournaments this way.

Also, here is Stanford Wong's estimated wins from Sharp Sports Betting (page 202) alongside the actual average.

Seed Avg. Wins Wong
1 3.35 3.7 to 4.2
2 2.41 2.70
3 1.86 1.90
4 1.53 1.30
5 1.13 1.00
6 1.19 0.80
7 0.83 0.70
8 0.71 0.60
9 0.57 0.50
10 0.63 0.50
11 0.51 0.50
12 0.53 0.50
13 0.27 0.40
14 0.16 0.20
15 0.08 0.10
16 0.00 0.10
It's not whether you win or lose; it's whether or not you had a good bet.
beachbumbabs
Joined: May 21, 2013
• Posts: 14217
December 6th, 2013 at 6:12:27 AM permalink
Quote: michael99000

I don't think the lower seeds having more wins than a higher seed is that unusual at all. Here's why..

They don't reseed after each round. So for example, let's say for a given region the 10 and the 12 seed both pull off upsets, while the other six higher seeds all win. In round 2 the 10 seed actually has to face a more difficult team (the 2 seed), than the 12 seed (who now faces the 4 seed).

So I'd say that if you did that table for just Round 1 wins, you get a uniform distribution of wins vs seed. But since after round 1 it's not reseeded, you can't really say that a higher seed is more likely to win in round 2 or thereafter.

Also, once you get to the 6th and 7th seeds, your talking the 21st best team, 22nd best, etc.. And I honestly don't think the tournament committe has only clue whose better than whom

Without having statistics in front of me, it's been my strong perception in recent years that the 12 and 13 seeds perform well above expectations as a group. I'd be interested in knowing whether that's a true trend over time. OTOH, it may just be that those teams get more attention because they're upsets by definition.

There also tends to be a bias towards certain conferences, specifically the ACC, in filling berths going to non-conference champions. I think an interesting prop bet might be to determine how many berths go to non-champion teams, and then pick how many additional teams from each of the 8 or 10 strongest conferences make the dance (before the brackets are announced). The bet could also include a field bet for those that get there without being from one of the named conferences.
If the House lost every hand, they wouldn't deal the game.
Joined: Jun 20, 2012
• Posts: 297
December 6th, 2013 at 7:50:30 AM permalink
Be sure to take into account that sometimes teams are "mis-seeded" due to bracket rules and committee errors.
Two years ago an 11 seed was a 2 point favorite over a 6 seed. I have much more confidence in the Vegas line than the committee.

Since the four worst teams in the field play each other for the chance to make the 64 team field, the quality of 16 seed that get to actually play the 1 seed, has slightly improved since the two worst teams (in theory) have already been eliminated. This also pushes other teams down (as 6 teams have a 16 seed instead of 4 or 5), thus 15 seeds are better relative to the field than they were before the field expanded.

I have had generally positive results betting on conference totals. I bet the ACC under 7.5 +145 in 2012. Duke going out after the first round really helped and then my Jayhawks finished it off when they beat UNC. In 2013, I didn't think any of the totals had enough of an edge to place a bet. I did like the Mountain West not to win the championship at -5000 and the WCC not to win the championship at -1800 or so.

I have not had success on the will a 14,15 or 16 seed win a game prop. My numbers have said NO is a strong wager both years as historical data indicated that a 1,2, or 3 seed loses in the round of 64 about every other year. I've failed to cash with odds of +300 and +210.

Edited for 1st/2nd Round Confusion.
wroberson
Joined: May 11, 2011
• Posts: 426
December 6th, 2013 at 8:38:05 AM permalink
The 1st round has 4 or 8 teams playing for the last two or four seeds.
2nd round is the round of 64.
Buffering...
Joined: Jun 20, 2012
• Posts: 297
December 6th, 2013 at 8:57:56 AM permalink
Quote: wroberson

The 1st round has 4 or 8 teams playing for the last two or four seeds.

The NCAA changed the round numbering system in 2011 when it added 3 more games to what was previously the Play-In game. The 1st round, round of 68 or "first four" (referring to number of games not teams) is played in Dayton, OH on the Tuesday and Wednesday (2 games per night) between the announcement of pairings on Sunday and the beginning of the round of 64 (the new 2nd round, old 1st round) on Thursday.

The lowest four automatic qualifiers play on the s-curve and the last four at-large selections play in the round of 68. The automatic qualifiers are assigned to play another automatic qualifier, while the at-large selections are assigned to play another at-large selection.

I don't think calling it the round of 64 the 2nd round will catch on until the field is to 96 or 128. If half the teams played I think it could be legitimately called a "round", but when only 8 of 68 teams are playing it hardly seems worthy of a "round" designation.