The TeamRankings.com 2013 Bracket Breakdown
Our Offical 2012 Brackets
We encourage you to read our full breakdown below, but for those of you that are anxious to see our official brackets for 2012, here you go:
- Small Pool (about 10 people), standard 1-2-4-8-16-32 scoring
- Medium Pool (about 100 people), standard 1-2-4-8-16-32 scoring
- Large Pool (about 1000 people), standard 1-2-4-8-16-32 scoring
How We Make Our Picks, What We Think Of This Year’s Tournament, And Why You Should Follow Our Advice
Before you even think of playing one of our expert brackets or using any of the NCAA prediction data we publish, you need to be confident that we're smart guys and that we know our stuff when it comes to intelligently picking a 2012 bracket. That's why we publish this writeup.
Our picks aren't going to win big every year, but the methodology, brainpower, and technology powering them is what makes them different and better than your cousin Frank's NCAA bracket advice -- or Seth Davis's bracket advice, or Barack Obama's bracket advice for that matter. (That was not a political statement, just talking about bracket picks here!) Just because someone watches a ton of college basketball, played college basketball, or wins national elections does NOT mean, by any stretch, that they understand how to identify the optimal strategy to win first place in a bracket pool.
Optimal bracket strategy is a mathematical challenge, not just a college basketball knowledge test. It's easy enough these days to find solid data on how good every team in the tournament is and who "should" win each game. TeamRankings.com, Ken Pomeroy, Jeff Sagarin, Vegas odds...all of our numbers are good predictors overall. The question is how you transform that data, complemented by other relevant sources of information, into the best decisions for your specific bracket pool scenario.
What does your points per round scoring system look like? Is there an upset bonus? How many people are you competing against? Different answers to those questions result in much different optimal strategies. Maximizing your odds to win a bracket pool is almost always much more complicated than just picking the expected winner of each game.
Some people look at our brackets and say, "What! I could have come up with that!" Well, of course you could have. Anyone out there can pick the same bracket that we're recommending in any given year. But by making a statement like that, you're completely (and unfairly) discrediting all of the underlying analysis that went into making our picks.
We don't decide on what we want the optimal strategy to be, and then go find a bunch of random numbers to back up our claims. (Lots of other people do, though!) Rather, we run the numbers to determine what the optimal strategy should be, then we do our own review and analysis of the results, double-checking the numbers and making final adjustments that we think make sense. If the numbers come back looking a certain way, we don't say, "Hey there, computer algorithms that we have spent ten years creating and refining and that we know work well over the long term -- we don't really like these picks! Too many favorites. How about a double digit seed in the Sweet Sixteen and maybe a sleeper pick in the Final Four?" That's not how it works.
If there is one message we want to stress heading into this document, it's that every year is different. That is the entire premise of "Moneyball" style analysis. It's not about about discovering some secret formula for predicting success that you blindly use year after year. It is about devising systems to identify whatever inefficiencies happen to exist in the market at the given time, and exploiting them. Those inefficiencies can change drastically from year to year depending on what teams make the tournament, what the matchups are, and what the general public happens to be thinking when NCAA tournament time rolls around.
That's what we do, and that's what you're paying us to figure out. We don't pre-determine the answers.
How We Come Up With Our Bracket Picks
We know you're here for the picks. But before we get to the those, it's worth reviewing our high level bracket prediction process so you have a clear sense of why we make the picks we make. First and foremost, our goal is to help you win a bracket pool -- to come in first place. Not to generally "do well"...you almost never win money or prizes in bracket pools for doing pretty well. You need to win this sucker.
You may not realize it, but that goal has major implications for bracket picking strategy. Think about it this way, with an extreme example. Let's say you compete in 10 bracket contests in the next 10 years. Would you rather win 2 of them, and come in last place in the other eight, or come in top 25% in all ten of them, but never win a prize?
The result of the former is usually some serious cash and prizes. The result of the latter is, maybe, a sort-of-impressive bragging right to discuss at the water cooler in early March. Maybe. The point here is that we'd rather win the two pools, lose eight, and laugh our way to the bank. And if we're not going to win, then pretty much any other result means the same to us. Nothing. If our bracket picks win way more often than most people, but totally stink the other years, would you be unhappy? We hope not, because that's achieving the primary goal!
With that goal in mind, here is how our picking process works at a high level. In short, we:
- Start by looking at our round by round team survival odds numbers to identify the teams we think are most likely to advance. These projections are based on objective computer simulations driven by our sophisticated team power ratings.
- Create an index of public picking trends from the biggest national bracket contests in order to understand how many people in a typical bracket pool will pick each team to advance to any specific round
- Look at Las Vegas odds for every team (futures odds) to understand the implied survival odds that the betting markets, which are good predictors and tend to adjust for "intangibles" better than our models do, are giving each team
We combine all these numbers to figure out where the biggest value opportunities appear to be in 2012. This includes identifying specific teams that are highly "underpicked" by the public despite having great survival odds, as well as more general observations like, "Wow, there is a lot of hype about the Big East teams this year but they don't look nearly as likely to make a deep run in the tournament as the public thinks."
We then put all these data inputs into a sophisticated tournament and bracket contest simulation engine we've created -- all computerized -- which lets us simultaneously model two things:
(1) What teams opponents in your pool will pick in their bracket sheets
(2) What teams will actually win in the tournament
Simulating both of these things allows us to estimate the likelihood that any particular bracket you fill out is actually going to win your pool. We run these simulations many thousands of times, for a huge number of different possible brackets you could fill out. For each bracket pool scenario (for instance, a 50-person pool with a 1-2-3-4-5-6 scoring system with upset bonuses, or a 1000-person pool with a 1-2-4-8-16-32 system without upset bonuses, or a 50-person pool with a 1-2-4-8-16-32 scoring system that is full of Kansas fans), we find the best possible bracket for you to pick. Finally, we give all the resulting brackets the eye test, and make any final adjustments we think are warranted. (Those adjustments are almost always very minor.)
Make sense? Hopefully that gives you a clear understanding of what we're doing here, and why you paid us money to do this stuff instead of doing it -- or more likely, not doing it -- yourself. You're not flying blind here. A ton of objective data and analysis is behind these picks.
As we mentioned above, though, you absolutely must be comfortable with the fact that our picks will not help you win money and prizes every year. That's simply an impossible goal, especially if you're competing against 20 or 50 or 100 or more people. In fact, starting this year, for every bracket we post for each pool size, we tell you exactly what we think are that bracket's odds to win a pool of that size. Notice how those odds are always far, far from 100%.
However, those are also far better than the odds for your average bracket picker. We've been posting "official" brackets on the site since 2005, and using these methods on our own for a long time before that. Year after year, this level of analysis typically results in a 2x to 10x edge in bracket win odds. And generally, the bigger your pool, the bigger the relative edge, even though overall you have a much lower chance to win a huge pool than a small one.
Let's put that into context. Think about your last paycheck. Now triple it. Wouldn't you rather be making that kind of money? You're still not as wealthy as Bill Gates or Mark Cuban or even the cast of Jersey Shore for that matter (we're guessing), and you may never be. But you're a lot better off than you used to be. That's why you use BracketBrains.
Now that you know about how our complex computer simulation engine works to find the best brackets for many different types of pools, it also helps to explain how different characteristics of your pool shape what our recommended bracket ends up being. The computer crunches the numbers in a much faster time than we can on our own, but there are methods to its madness. And we never blindly follow the algorithmic recommendations until we personally review and vet the strategy being proposed.
Strategy Differences For Large And Small Pools
Pool size is one of the most important factors influencing the selection of an optimal bracket. The ideal strategies for large and small pools are fairly different.
In a small pool, you generally just want to maximize your expected points. That is, get as many points as you can, which in most scoring systems mean picking as many winners right as you can. This typically involves playing a conservative strategy where you pick mostly favorites to advance. If, in last year's case, a Butler or a VCU -- or even a UConn -- makes a deep run as an unheralded seed, you won't get those points. But in a small pool, chances are good that no one else will either. Your opponents for the most part won't be gaining ground on you with the big upsets, but they will be falling back when their own upset picks fall through. The best strategy here is to play it safe, grab the easiest points available, and hope that your opponents do something dumb, because they usually do. Nevertheless, you still need to be smart about making sure your key picks, like your NCAA champion, check out from risk/value perspective.
In a big pool, on the other hand, it's likely that at least one person will pick all or most of the key late-round games correctly, so a conservative strategy may not be optimal. Your best bet here is to exploit market inefficiency. Basically, if there's a team that's got a reasonable chance to advance to a certain round -- not a total long shot, mind you -- but no one is picking them to do so, that's better than going with the team that has a great chance to advance, but is picked by everyone. If possible, you want to be one of only a few other people who got most of the end games right, rather than one of 100 other people that did. To do that, you should strive to make bets on a few valuable late-stage games that distinguish you from others. If you get some or most of them right, you leapfrog a large percentage of your opponents. No matter how many first round upsets you pick, in most scoring systems, it's not going to make much difference.
Strategy Differences Based On Scoring System
Scoring systems have an enormous impact on your strategy.
In a pool where the point values double each round (1-2-4-8-16-32), most of the what sets you apart from your competitors is the later rounds. A "great" first round might be 26 points out of a possible 32; a lousy one 18 points. That entire difference is wiped out by the value of choosing a single Final Four team correctly. A great score for the first two rounds might be 50 points out of a possible 64; a lousy one, 30 points. Even that difference is easily overcome by picking a national champion someone else doesn't pick.
We've done thousands of simulations, and the conclusion we've come to is that for the 1-2-4-8-16-32 points per round system, which is by far the most popular, your best bet is to take a calculated risk or two in the later rounds, but for early round games to stick with favorites. Remember that "favorites" and "better seeds" are not always the same thing, though, because sometimes the NCAA Selection Committee gets it wrong. And some early games are so close to being coin flips (Alabama-Creighton, Cincy-Texas, Temple-Cal, ND-Xavier, St Mary's-Purdue, UConn-Iowa State, Gonzaga-WV) that there's not really a "wrong" option.
Still, while picking huge early round upsets is certainly fun, and making one or two won't hurt you much, it's definitely not your best strategy in a 1-2-4-8-16-32 pool. The risk of losing a team with potential to make a deep run for one measly point just isn't worth it.
It's a very different story in a pool where each round is worth only slightly more than the one before it. The most common example of this is round point values of 1-2-3-4-5-6. Because you only get 6 points for picking a champion, it's more important to distinguish yourself in the early rounds. If you pick Belmont in the Sweet 16 and are right (you've got a 1 in 5 chance), based on public picking trends, you'll get two points that 95% of your pool isn't getting and one point that 88% isn't getting. With decreased later-round point values for correct picks, those early round differentiation points can make a difference. You'll note that our strategy for a 1000-person 1-2-3-4-5-6 pool indeed has Belmont in the Sweet 16.
Still, in this system, you don't want to go overboard. Pick a few strategic underrated teams with solid odds, but don't put all of the 15 seeds in the Sweet 16.
In pools that provide bonus points for picking upsets, however, it's a different story yet again. For these pools, it's often the case that a big early round upset can deliver as many points (or more) than a late round game. For example, picking a #15 seed over a #2 seed is likely worth 13 or 14 points if you use the common "difference in seed" upset point bonus -- nearly as much as picking a Final Four game correctly! If you have a 20% chance at nabbing those 13 points, it's probably worth the risk. You'll note that our picks for a 1-2-3-4-5-6 bonus pool with 100 people is conservative in late rounds but quite aggressive in trying to gather early round bonuses. This makes intuitive sense: a 20% (or even 10%) chance for Lehigh to beat Duke in the first round can justify picking that upset in this scoring system. But the odds of Lehigh being in the Final Four are tiny, so it's still not worth picking huge underdogs to go that deep.
Why Early Round Upsets May Not Make Sense
Early round upset picks are always "fan favorites"...everybody loves 'em! But while they're entertaining to watch and hypothesize about, their role in bracket strategy tends to be a lot less important than most people think, especially in certain situations.
Take a big pool (1,000 people). The odds of early round games making a difference in how you finish are tiny in a 1000-person pool UNLESS you pick the same late round teams as everyone else. If you pick the most popular Final Four -- right now, Kentucky over UNC in the final, with Syracuse and Michigan State also in the Final Four -- you'll likely be one of about 20 people total in your pool with that exact same combination. And if you do that, your early round games will matter and you should pick a couple of strategic upsets (we like Wisconsin, Vandy, Baylor, and Memphis as sleepers in the Elite Eight).
However, the "most popular Final Four" strategy doesn't make much sense for that pool. Overall, the odds of the final three games playing out as the "most popular" option says they will is about the same as the odds of a high-value underdog like Wichita State or Wisconsin winning the NCAA tournament. But if you pick the Badgers or the Shockers to win it all, you'll be one of five or ten people with that pick. 1 in 6 is better than 1 in 20.
Finally, the odds of the "favorite" Final Four playing out are much, much smaller than the odds of 2 seed Ohio State winning the tournament. If you pick Ohio State to win a big pool and they do, you'll be one of a small group of competitors with a viable chance to win the pool. In that case, the best way to distinguish your picks is to throw in a semi-upset finalist like Missouri. If you get that right too, it will yield 16 points and basically seal the victory for you. Missouri in the final is by no means a stretch of the imagination, and It would take getting a lot of first round upset picks right at one point apiece to have the same impact.
To close the loop here: Don't convince yourself you need to pick a bunch of first round upsets just because lots of people and the talking heads on TV talk about them. Even in really large pools, picking lots of upsets is likely less valuable than you may think from an optimal strategy perspective. If your goal is to win a million person pool, by all means, pick a few big upsets (Belmont, Texas, Cincinnati, Memphis, and Cal are all reasonable Sweet 16 picks). But to win a 1000-person pool, you're much better off minimizing the number of bets that are less than 50-50, but making those bets big.
Breaking Down This Year’s Field
The media like to talk about seeds as if they are a perfect measure of quality. If the number next to your team name is lower than the number next to the other guys' name you're the favorite. End of story. However, the NCAA regularly botches seeding, and the relative strength of teams varies year to year.
So what's unique about this year? There are a few things, and they drive the way our final bracket picks look. So please read the following carefully:
1) Kentucky is very, very good. The Wildcats are much stronger than any other team in the bracket, and they are the biggest favorite the tournament has had in years.
2) After Kentucky, the next six teams -- all of the 1 and 2 seeds except Duke -- form a strong group. Each of these teams is well above the rest of the field. The 2 seeds besides Duke (Kansas, Missouri, Ohio State) are all about as strong as the 1 seeds in their region, and each has a legitimate chance to win it all. This is very different from last year, where the 2 seeds were relatively weak, and didn't appear any stronger than the 3 and 4 seeds.
3) There are a few cases where the NCAA has completely mangled the seedings, but they've managed to mostly match overrated teams against other overrated teams, and underrated teams against other underrated teams. For instance:
- Vanderbilt is underseeded as a 5 (they should be a 4) but if seeding holds they'll play the strongest #4 seed (Wisconsin) in the second round. We project Wisconsin as a slight favorite to advance to the Sweet 16, but if Vandy ended up playing Michigan, we'd peg the Commodores as a clear favorite.
- Florida State is overseeded at a 3. If they were to play Florida (a 7 seed) or Memphis (an 8) in the second round, they'd be a significant underdog against either of those teams. But their potential Sweet 16 opponent, Cincinnati, is also relatively overrated at a 6, so Florida State would be a slight favorite should they get through to the second round. And if the Seminoles had been matched up against under-seeded 14 seed Belmont rather than St. Bonaventure, we'd have given Belmont close to 50/50 odds to beat them.
- The other side of that is that underrated teams seeded 5 through 9 -- Kansas State, Florida, Vanderbilt, Memphis, St Louis -- all have pretty tough draws. Kansas State's may be the most fortunate; Southern Miss is not deserving of a 9 seed, and Syracuse may be vulnerable due to the loss of Fab Melo. But KSU still has to play a #1 seed to get to the Sweet 16. Memphis and St Louis have each other, then Michigan State to beat. That's a brutal path. And Florida would have to beat a very tough Missouri team.
What all of that means for a bracket picker is that an optimal early round bracket is going to include a lot of "chalk" (favorites) this year. On balance, chalk looks good and underdogs have a really tough road ahead of them this year.
Three of the 4-5-12-13 pods could easily go to either the #4 or #5 (only Michigan is a fairly strong favorite), so picking some minor 5-over-4 upsets probably won't affect your odds negatively. But the 1, 2, and 3 seeds are all solid favorites this year, with the exception of Florida State. Your best bet in most pools is to go with all favorites in those games.
For those that insist on a big upset pick for the Sweet 16, we'd favor #14 Belmont in a potential second round game against NC State or San Diego State, but the Bruins are a slight underdog against Georgetown. If you want to pick something that could really impress people but still has a reasonable chance of happening, put Belmont in the Sweet 16.
Those are our high level opinions on the 2012 field -- now let's get to some actual brackets.
A Quick Rundown Of Our Brackets
We’ve significantly expanded our number of expert brackets this year, thanks to our new simulation engine. We know it’s a lot, but optimizing based on specific dynamics of your pool is critical. To make it easier to get to the bracket you need, we’ve broken our lists of expert brackets up across pages:
- Best Brackets - The top brackets we recommend playing (no upset bonus)
- Runner Up Brackets - More bracket options for your pool size
- Upset Brackets - If you're in a pool with bonuses given for seed upsets
- Champion Brackets - If you're dead set on picking a specific team to win it all
- Conference Brackets - If you think we are underrating/overrating a conference
- Team Bias Brackets - If your pool is filled with people biased towards one team
- All Brackets - If you want the entire list....it’s a lot!
Simply filter on your pool size and scoring system, for completely personalized results, and every column header on the table is sortable too.
Writeups For The Three Brackets We’re Tracking
This is very close to a “highest probability” bracket, the one you would pick if you went purely by survival odds. The one game exception is UNC losing in the regional final to Kansas. Our models see the Tar Heels as a slight favorite versus the Jayhawks, but because it’s very close and because Kansas is a strong and undervalued team, we advance them in this bracket based on the value perspective.
Another dynamic in play with this bracket is that picking Ohio State to reach the championship game increases the likelihood of winning a small pool if an unlikely champion emerges this year. If, say, #3 Baylor ends up winning the national title, there probably wouldn’t be anyone in a 10-person pool who picked them to do so. In that case, if Ohio State does make the final game, you’ll probably be the only one in a small pool who gets one of the two finalists right, since no one would have picked Baylor. You’d be in fantastic position to win the pool. If you had picked UNC in the final, probably at least three other people in your 10-person pool would have also, so you’d have lower odds to win.
In this bracket designed for slightly larger pools, just playing the conservative champion card with Kentucky doesn’t cut it. Instead, we extend our Ohio State pick one round further. The Buckeyes have a better than 10% chance to win it all, but less than 4% of people are picking them to do so. This means in a pool of 100, you’d expect about 4 other people to pick OSU. If they win, you’re effectively competing in a 5-person mini-pool, where it pays to be very conservative with the rest of your picks.
When we first saw the optimal bracket for large pools, we’ll admit it, we had trouble believing it was optimal. Ohio State seems like way too much of a mainstream pick as a national champion to win huge pool, right? And in past years, haven’t we always advocating picking a significantly undervalued but less popular tournament winner? In 2011 we had #4 Texas as our 500-person pool champ; in 2010 we had #3 Baylor.
But the more we investigated, the better this bracket looked. Our algorithms were right. As mentioned above, Ohio State has a better than 10% chance to win it all, but less than 4% of America is picking them to win. So that’s a great start.
If you pick Ohio State to win your 1000-person pool, there’s a 10% chance you’ll be right. If you are, at that point you’re effectively competing in a mini-pool against around 40 other people who also picked OSU to win. At that point, winning becomes a secondary optimization challenge. If you pick OSU to win, and they do, how should you pick the rest of the games to maximize your odds to beat the 40 others who also picked OSU to win? This is getting intense!
The answer in this case, which we alluded in our above notes about upset picks in general, is to make one moderate bet -- #2 Missouri making the final game -- and otherwise play it pretty conservative. Mizzou is underrated by the public so only about five people are expected to make that pick.
Now we’re down to five people, including you, who picked an Ohio State vs. Missouri final with OSU winning. Once we get here, your smart (i.e. conservative) early round picks should give you around a 40% chance to win the face-off; add in a few other less likely scenarios, and it yields a 0.9% chance of winning a 1000-person pool this year.
That may not sound like a huge number, but it’s nine times the average chance to win of your competitors.
We also took close look at a picking any of a handful of underrated teams as champion: Memphis, Kansas State, Wisconsin, Vanderbilt, Wichita State, New Mexico, Florida. All of these would be decent picks in a very large pool, since you may be the only one to pick them. But none are quite as strong a pick as OSU, who is undervalued but has much higher odds to win. The basic reasoning behind this decision is that if you pick a team like Memphis to win it all, your fate lies almost solely in the hands of that a big long-shot winning the tournament. But if you pick Ohio State, we think our crafty game theory for optimizing picks for the resulting 40-person “mini-pool” will give you a higher chance to win overall.
A Deeper Look At Multi-Bracket Strategy For Large Pools
Even though we like Ohio State in large pools if we are entering one bracket, large pools are also great for multi-bracket strategy. Play our Best Bracket here with a few in the RunnerUp-1000 category, and you can get to 3-4% odds to win by entering four pools. (Just go to our Runner Up Brackets page and filter on pool size of 1000 to see those brackets.)
Here’s an example of a Runner Up bracket with a big upset winner, Wichita State: http://www.teamrankings.com/ncaa-tournament/runner-up-brackets/bracket/?bracket_id=35521
The math is similar to the Ohio State bracket we describe above. There’s a 1% chance WSU wins it all, but only 0.1% of the population picking them to win. So you could be the only person in your 1000-person bracket pool picking the Shockers to win it all, and if not, you’d likely be one of two or three. All told, you have a 0.8% chance to win that pool with this bracket -- not bad, and pretty similar to the Ohio State bracket above.
More importantly, though, those two brackets along with these other Runner Ups:
would give you a tremendous, diverse portfolio if you wanted to enter several brackets in a pool. One of these additional three brackets has Kentucky winning over Kansas State in the final, which is the “big bet” of that bracket; one has Missouri over Kansas, and a third has Wisconsin over Kentucky. You could enter all five of these brackets in a 1000-person pool, with virtually no risk of stepping on your own toes. And for $100 in entry fees if it’s a $20 buy-in, you’d have an expected return of around seven times that amount.