How to Use BracketBrains Betting Mode

Using BracketBrains

Using BracketBrains to analyze betting lines is a similar process to picking bracket winners; you are trying to find potential inefficiencies where it appears the market (in this case, the lines) are either overvaluing or undervaluing the true odds of an event occurring, whether it be a team winning a game by a certain amount, scoring a certain number of points, etc.

Our Guarantee and Principles

Our guarantee is this: sophisticated math and objective, honest analysis. That's it. We are not touts, and we don't guarantee that any of our predictions will be winners. When you buy BracketBrains, you are buying the opinions of a team of people who paid way too much for their college engineering degrees but learned a fair amount of stuff along the way.

We have found over the years that our analysis has been appreciated by people who like to bet on or handicap sports, either for fun or for a living. But the projections of our models could be dead-on or dead wrong. If you bet based on the recommendations of our models and lose, you do not get a refund.

We certainly don't encourage people to gamble on sports, especially without realizing how long the odds are of winning consistently. At the same time, we realize that a lot of people out there are fully aware of the risks and still want to play, either professionally or among friends, for fun and hopefully for profit too. If those people are willing to pay us for our research, we are happy to provide it to them

One thing we will never compromise is our objectivity. Like anyone else, we have "gut" instincts on games too, but those opinions are never built into our algorithmic models. Also, you will notice the complete absence of ads for sports books, handicappers, or 900 lines on our site. We are indebted to no one in the gambling industry.

How to Use Betting Mode

We've tried to make betting mode as simple to use as possible while still providing a lot of valuable analysis. The short story is this:

- We have built multiple algorithmic models that make betting-related predictions. Their results do not always agree, which makes sense because they use different methods and analyze data in different ways.

- Some of our models predict odds (e.g. Team A has a 57% odds to cover -7.5) and others predict points (e.g. Team A is expected to win by 8.7 points). You can get all the detailed raw predictions on the Predict Games pages for each individual game.

- On the Picks pages -- Win Picks, ATS Picks, and O/U picks -- we provide an overview summary of the picks and relative confidence of all of the models for all games on a given day. For ease of comparison, we have mapped raw predictions to star values. For example, 51% odds to cover is a 1-star ATS pick, while 61% odds to cover is a 3-star pick. Likewise, a margin of victory prediction of 8 against a spread of -7.5 is a 1-star ATS pick, but a margin of victory prediction of 14 against the same spread would be a 3-star pick.

- A one-star pick is by no means a recommendation; it's much closer to a toss up. A one star pick means that the projection is within the margins of error we expect from our models, and for all we know, the game could actually be a one-star pick for the other team or side.

- Four-star predictions are extremely rare and 5-star predictions are all but nonexistent. Again, we are realists and not touts. Anyone who tells you a team has, say, an 85% chance of covering the spread in a given game is most likely smoking something, and we don't mean cigars. Professional gambling syndicates would have beaten that line down before you even got a peek at it.

- Fair value money line calculations, for now, are only on the individual game pages in Predict Games. To get fair money line values, we translate the projected win odds from our models to values that would represent an expected return of 0 if you bet them. So, theoretically, if you can get a money line number better than the fair value money line, then you should have the odds in your favor.

- When we track the results of our models, which we will do over the course of the tournament, we take a snapshot of the lines and our corresponding predictions between 15-45 minutes before each game starts and measure our performance on that. In most cases this is the closing line but not always.

Key Things to Know

- All data-driven predictive models have margins of error involved in their calculations. Don't read too much into differences in predictions of 2-3% or 2-3 points, and remember that 1-star picks (less than 55% odds or within 2.5 points of the current line) are far from strong calls.

- There are some factors that data is not well suited to modeling effectively. For basketball, one of the most important ones is recent injuries or key players returning to a team after being out for an extended absence. Our models don't explicitly account for such an effect, so we have no choice but to say you need to make a subjective adjustment in these cases. UNC in 2009 is a good example, with Ty Lawson. Who knows if he'll play, what shape he'll be in when he does, and whether other teammates will pick up the slack or not.

- Another thing to watch out for is games with uncommon lines, especially for the Similar Games model. Since that model compares to historical games to make predictions, if it can't find many historical games with similar lines, the relevance decreases. This typically happens in the cases of huge spreads, extremely low or high totals lines, or when a key recent injury causes a distortion in the expected line for a team.

Summary

Above all, our goal is to make our Betting Mode customers more informed. Beyond the algorithmic predictions, the Predict Games pages for every game include tons of data for you to review and use to uncover potential matchup characteristics that may warrant more consideration or adjustments to the algorithmic predictions.

As a basic strategy, we believe that over the long term, solid algorithmic predictions can help to consistently achieve wagering performance that is at or slightly above breakeven at standard vigorish, with a some solidly profitable seasons now and then. Add some successful subjective analysis for injuries or other more intangible factors -- or some inside information! -- for each game and results should improve from there.

If you want to bet on sports for more than just entertainment purposes, just know that like any casino game, the odds are stacked against you. As they say, Vegas didn't pay for all those bright lights by selling $4.99 breakfast buffets. But if you're going to wager, hopefully the information BracketBrains provides will help inform your decisions in a positive way.