Editor's Note: This is a guest post from TeamRankings.com, a site that has provided data-driven bracket tools and analysis since 2004. They also offer premium bracket picks.
It's officially March Madness, which means that bragging rights (and more) are on the line in your annual NCAA bracket pool.
Here at TeamRankings.com, we may not be fantasy experts, but we've been researching how to win March Madness pools since 2004, and have built sophisticated tools to maximize your edge in bracket contests.
Backed by nearly 15 years of research, here are three tips to pick a smarter bracket in 2018:
1. Trust data, not opinions
There are over 4,000 games in a single college basketball season. To develop an intimate knowledge of every NCAA tournament team, a human brain would need to assimilate and process data from most (if not all) of those games.
It’s simply not possible. As a result, everyone from casual fans to the so-called "expert" college basketball commentators on TV form biased opinions on teams based on imperfect data.
Recency bias also comes into play, since a lot of people (including some of the folks providing analysis for the big networks) don't really start tuning into college basketball until the conference tournaments start. Then they get impressed by a team like Michigan, a surprise conference tourney winner, even though there's not much evidence that teams who get hot at the end of the regular season outperform expectations in the NCAA tournament.
In a season-long picking contest, only a (mostly lucky) handful of humans would pick college basketball winners more often than sophisticated computer models or projections implied by the betting markets. So trust the markets and the models more than the humans.
Our approach is to use an "ensemble of models" approach to make bracket picks. We evaluate team ratings and projections from multiple algorithmic models and top ratings systems, along with the latest Vegas odds.
2. Your Scoring System Matters
Your pool's scoring system can make a huge difference in determining the optimal strategy to win. You shouldn't even start to think about making any specific picks until you've fully analyzed the implications.
Let's look at an example. The structure of the most popular 1-2-4-8-16-32 points by round bracket pool scoring system places a very high importance on getting your late-round picks correct. For example, getting just one of your two NCAA finalists right is worth the same as getting a whopping 16 first round games right.
If your pool uses this scoring system, focus the vast majority of your analysis time on your Final Four and beyond, because pulling your hair out over first round picks is almost certainly going to be a waste of time.
However, if your pool's scoring system is flatter (say 1-2-3-4-5-6 points by round), or if it has upset bonuses, it's a completely different story.
In those types of bracket pools, early round games are much more likely to have a big impact in determining the pool winner, so your bracket should look much different. In fact, our research shows that most players in upset bonus pools aren't nearly as aggressive as they should be when it comes to making risky picks.
The math behind optimizing your bracket for your particular scoring system is incredibly complex, though. Our solution was to develop bracket pick optimization algorithms that take into account both your pool's scoring system and upset bonus structure.
3. Consider opponent picks
In bracket pools, there is no prize for getting a certain number of picks right. You win your pool if, and only if, you score more points than everyone else. And in order for that to happen, you need to pick at least one (and most likely several) games right that your opponents get wrong.
This is such a foundational element of bracket pool strategy that it's amazing how many bracket pickers just don't get it. We constantly field questions like "How many Final Four teams did you get right over the last five years?" In a vacuum, success rate picking Final Four teams is a relatively meaningless statistic.
Getting two Final Four picks correct could be an incredible performance, or not particularly great depending on the year. In 2011, the Final Four consisted of teams seeded 3, 4, 8, and 11. If you got just one of those right, you probably had a great shot to win your pool.
What really matters is how often you win pools, not how many picks you usually get correct, and those two measures are often less related than most bracket pickers think. And when your goal is to outscore your opponents, the picks your opponents make impact your odds to win.
Let's say you're considering two teams as possible NCAA champion picks: Team A and Team B. Team A has a slightly better chance to win it all, but you also expect them to be a much more popular pick in your pool. Usually, you're better off picking Team B.
Why? Because in the long run, your expected bracket pool prize winnings should be higher. You’ll get your champion pick right slightly less often than your opponents do, but in years when your less popular champion pick does come through, you’ll leapfrog a lot more people in the standings and be in a much better position to win.
To figure out which bracket picks are the best to make based on expected opponent picking trends, we collect public pick data from the popular nationwide bracket contests multiple times a day, and that data plays a critical role in our bracket picks and Data Grid value analysis tool, which helps identify the smartest upset picks.
Wrapping it up
Picking the bracket that gives you the best chance to win your pool is a daunting task, especially with 67 individual pick decisions to make. Other factors in addition to those mentioned in this article (e.g. the number of entries in your pool) also impact your optimal strategy.
Hopefully the information in this post was useful to help frame your thinking about the 2018 NCAA bracket. If you'd rather outsource the heavy lifting to the bracket pros, check out our premium NCAA Bracket Picks (or read more about them in this month's WIRED).