Bracketology Because It's Bracketology Week
If you are filling in your brackets today there are a few good sites out there to help get you un-stuck. WSJ's blind bracket separates you from your biases and just gives you a 5-point scale for hotness, experience, size, offense, defense, and 3-point shooting, plus seed range, RPI and conference profile (HT Skiptoomylou22). Also from the board, user "entirely reasonable" linked Steve Czaban's all-everything pdf bracket. Considering most of these games are 60-40 anyway, choosing teams with pretty looking colors is also a tried and true method of winning your bracket. Just ask my friend's wife. #notbitter
My own device is an excel doc I have to rebuild every year that spits out a confidence % based on KenPom, next to supplementary information on injuries and site for that game. Here's that file if you want to use it. Put in the names of the teams to compare and which round (Round 1 is that which begins Thursday; we don't count play-ins) and it should spit out a confidence level and a site for that game. 100% is a 1-seed over a 16-seed, 50% is a pick-'em, and less than that means you're predicting an upset. You're responsible for adjusting your confidence based on injuries and site.
Here's that formula with the first round:
High Seed | Low Seed | Difference | Confidence | Site |
---|---|---|---|---|
Atlanta Regional | ||||
1 Kentucky | 16 W. Kentucky | 0.55 | 100.00% | Louisville, Ky. |
2 Duke | 15 Lehigh | 0.18 | 83.50% | Greensboro, N.C. |
3 Baylor | 14 SD State | 0.13 | 73.43% | Albuquerque, N.M. |
4 Indiana | 13 New Mexico St | 0.15 | 78.06% | Portland, Ore. |
5 Wichita State | 12 VCU | 0.12 | 73.23% | Portland, Ore. |
6 UNLV | 11 Colorado | 0.10 | 69.40% | Albuquerque, N.M. |
7 Notre Dame | 10 Xavier | 0.04 | 56.93% | Greensboro, N.C. |
8 Iowa State | 9 Connecticut | 0.03 | 54.80% | Louisville, Ky. |
Phoenix Regional | ||||
1 Michigan State | 16 Long Island | 0.47 | 100.00% | Columbus, Ohio |
2 Missouri | 15 Norfolk State | 0.56 | 100.00% | Omaha, Neb. |
3 Marquette | 14 Brigham Young | 0.10 | 67.76% | Louisville, Ky. |
4 Louisville | 13 Davidson | 0.13 | 73.34% | Portland, Ore. |
5 New Mexico | 12 Long Beach St | 0.08 | 64.92% | Portland, Ore. |
6 Murray State | 11 Colorado State | 0.07 | 62.43% | Louisville, Ky. |
7 Florida | 10 Virginia | 0.02 | 53.92% | Omaha, Neb. |
8 Memphis | 9 St. Louis | 0.03 | 54.74% | Columbus, Ohio |
Boston Regional | ||||
1 Syracuse | 16 NC Asheville | 0.32 | 100.00% | Pittsburgh, Pa. |
2 Ohio State | 15 Loyola MD | 0.37 | 100.00% | Pittsburgh, Pa. |
3 Florida State | 14 St. Bonaventure | 0.09 | 66.49% | Nashville, Tenn. |
4 Wisconsin | 13 Montana | 0.24 | 94.96% | Albuquerque, N.M. |
5 Vanderbilt | 12 Harvard | 0.08 | 64.45% | Albuquerque, N.M. |
6 Cincinnati | 11 Texas | -0.01 | 47.63% | Nashville, Tenn. |
7 Gonzaga | 10 West Virginia | 0.04 | 56.68% | Pittsburgh, Pa. |
8 Kansas State | 9 Southern Miss | 0.14 | 75.82% | Pittsburgh, Pa. |
St. Louis Regional | ||||
1 North Carolina | 16 Vermont | 0.32 | 100.00% | Greensboro, N.C. |
--or-- | 16 Lamar | 0.27 | 100.00% | Greensboro, N.C. |
2 Kansas | 15 Detroit | 0.32 | 100.00% | Omaha, Neb. |
3 Georgetown | 14 Belmont | 0.04 | 56.67% | Columbus, Ohio |
4 Michigan | 13 Ohio | 0.13 | 73.54% | Nashville, Tenn. |
5 Temple | 12 South Florida | 0.07 | 62.70% | Nashville, Tenn. |
--or-- | 12 California | -0.03 | 43.63% | Nashville, Tenn. |
6 San Diego St | 11 NC State | -0.02 | 45.54% | Columbus, Ohio |
7 St. Mary's | 10 Purdue | -0.06 | 38.80% | Omaha, Neb. |
8 Creighton | 9 Alabama | -0.02 | 45.74% | Greensboro, N.C. |
I am so happy Michigan missed a 3 seed and thus the most terrifying set of 14s since we put new tires on my grandpa's Cadillac: SD State, BYU, St Bon's, Belmont. Do not want. You've been warned previously of the weird KenPom-Wisconsin love affair; use with caution.
All it really does is convert KenPom differential into a prettier number and sticks that next to other useful info. I figure since a 16-seed has never beaten a 1-seed, I could create a constant from the difference between the worst 1 and the best 16 (so a hypothetical matchup of Syracuse and Lamar is 100%). Divide the KenPom difference in the game you're calculating by the constant, multiply that by .5, and add another .5.
The first time I used this thing I won a big pot of gold. Last year I finished behind two of my friends' wives. If you win something you can pledge to the Hail to the Victors Preview fund or something.
Pro Tips: If you're going against only a few people, play it safe; if you're in a large pool, I recommend filling out several brackets each with a major upset and a big run for a middling seed you like. This is because it's easier to win a big pool by getting big points from one team nobody else in the winners circle has than hoping a lot of good early picks can carry you through an end game with 20 other Kentucky-OSU people. Picking a lot of upsets is a bad gamble.
i don't recommend picking OSU (for obvious reasons, traitor), but this article:
http://www.slate.com/articles/sports/sports_nut/2012/03/ncaa_basketball…
makes a good point. in a large pool, your best chance of winning is picking an undervalued team that plausibly could win, but that most participants aren't selecting.
Can someone explain the mgoblog obsession with KenPom? In all honesty, I'm quite the gambler and NCAA BB man but I never found their odds or projections to be superior to other sites. Someone shed some light I'm not seeing, thanks.
I don't suggest using him to exclusivity. For my part however, like Rivals.com, he makes his bank of knowledge available and freely accessible. Sagarin for example has a wall called USA Today between us.
Traversing intellectual property is a scary process because none of it makes any sense and the limits of what you can do are not based on your rights but the whimsy of someone with a lot more money than you choosing to sue one time. So that which is most freely available becomes the standard, elevating the author by the scope of authors who build upon that work rather than trying to milk an idea like it's a product.
The best pure analysis I have seen so far is from Nate Silvers.
He uses the stats from all of the big number crunchers (Kenpom, Sagarin, etc.) then adjusts it based on latest injuries and how far they are playing from home.
I am using this to guide my picks this year.
March 15th, 2012 at 10:22 AM ^
I did this, but also added in conference tournament performance and past performance in the NCAA tournament. I got wildly different results.
Quick question - why is Duke so low compared to the other 2 seeds. Lehigh doesn't seem to be a particularly good team, and those types of upsets tend to happen when you have a middling 2-seed, not a team like Duke that nearly won the ACC. Just wondering.
does not play defense. Top 2 seeds tends to be balanced and good with offense and defense.
Duke relies a lot on three point shots - and they tend to go cold, a lot. The other 2s - Missouri, Ohio, and Kansas - tend to rely more on points in the paint. Points in the paint are a bit easier to count on, that is, unless the opponent clogs the middle.
Duke was 14-4 in games decided by 10 points or less; their record is better than you'd expect it to be from their points scored and given up. A system that uses point differentials rates Duke a lot lower than a system like RPI that uses wins and losses only. Kenpom has Duke #17, MasseyPower has them #12. RPI has them #5.
Which you should believe depends on whether you think Duke won more than their share of close games because they were lucky or because they have some advantage in those contests. And also how much weight you want to give to their blowout losses (18 to UNC, 22 to OSU). Nobody else on the 2 line lost by 18 or more to anyone. And, finally, whether you think their failure to blow out really bad teams is because they aren't as good as the other top teams or because they showed some mercy to the Southeast Missouris of the world.
They're the best of the 15 seeds--KenPom thinks so by some distance. They never lost by ten or more, including competitive games at Michigan St. and Iowa St. Detroit lost by double digits 4 times, Loyola 5 times, Norfolk 5 times (including two losses to RPI #250 Delaware St.).
March 15th, 2012 at 10:27 AM ^
If a team tends to consistently win close games, I view that as a good thing and something to bet on in the tournament. You won't see a lot of blowouts once you start getting past the first few rounds of the tournament. I'd rather have a team that has lots of experience in close games than a team that either won by a large margin or lost because that shows me the team doesn't respond well to adversity.
I just want Jamiemac to fill my bracket out.
/count cash
My Final 4 ended up being Kentucky, MSU, Kansas, and Vandy, with Kansas over Kentucky in the final.
March 14th, 2012 at 10:43 PM ^
My Final Four was:
Baylor (beating Kentucky to get there)
Michigan State
Ohio State
Kansas
With the final game being Kansas v. Baylor, and Kansas taking it all. Hmm. Interesting.
March 14th, 2012 at 10:25 PM ^
Well, if this formula is close to accurate, St. Louis should be the regional for upsets, with four possible upsets.
March 14th, 2012 at 10:39 PM ^
it's not infallible.
March 14th, 2012 at 10:47 PM ^
Your confidence percentage for some games is listed as 100%. Does that mean it's inherently broken? There is no such thing as a really true sure thing.
March 14th, 2012 at 11:22 PM ^
Taking a 1 over a 16 might not be 100%...but it's damn close.
the entire system of confidence calcuations is based on the assumption that a hypothetical matchup between the worst #1 seed and best #16 seed is 100% in favor of the #1 seed (since no #16 seed has ever beaten a #1 seed).
I capped it at 100%. This isn't meant to be accurate, it's meant to be a guide for filling out brackets. It was wrong twice last night...
March 15th, 2012 at 10:39 AM ^
That Wall Street journal thing is a hoot. I went in deciding I didn't care about size and I was going to put a huge weight on 3-point shooting. I got a fun bracket to submit, really different than I would usually submit. I took it as it turned out, only over-ridiing the Michigan loss because if I am going to suck at predictions, by god I am going to enjoy my rampant homerism while I do it.
I have South Dakota State & Belmont making the Sweet 16.
March 15th, 2012 at 11:52 AM ^
I started by looking at experience and defense, then offense and 3 point shooting, then size, then everything else. I got SDSU too and figured that their judgement of a team's ranking in those categories must be off somehow.
Comments