How Does SmartBracket Work?
The central question SmartBracket is designed to answer is “How do I win my pool?”
One might answer by saying, “Just make all the correct picks!” This, of course, is true. If you correctly predict the entire tournament, you will win. However, making every correct pick is realistically impossible. Highly complex and intricate models have been developed to predict which teams will advance to each round of the tournament. These models are largely quite good. If you follow their recommendations, you will likely do well in your pool.
However, you don’t just want to do well. You want to win.
There is a more correct answer to “How do I win my pool.” That answer is “By getting more points than anyone else!”
It’s not important that you make all the “correct” picks. What matters is that you score points when your opponents don’t, and break even when they do.
Expected Relative Score
SmartBracket crafts a strategy to maximize your score relative to the scores of the other brackets in your pool.
It does so by drawing on two sources of information: The chance that each team advances to each round and the percentage of your opponents who picked each team to advance to each round.
With this information, SmartBacket calculates the expected relative score of choosing each team to win each round.
The usefulness of this information is best illustrated by an example:
Say Kentucky and Florida are playing in the first round, Kentucky has a 90% chance of winning, and 100% of brackets have picked Kentucky to win.
We have two choices. 1) Pick Kentucky to win, and 2) Pick Florida to win.
If we choose Kentucky, 90% of the time we’ll be right, and 100% of the time we’re right we’ll tie with everyone else (because they also picked Kentucky). Both we and our opponents will get 1 point. This would result in us getting a relative score of zero.
However, we can’t forget that 10% percent of the time, we’ll be wrong (Florida wins). Luckily for us, 100% of the time we’re wrong, everyone else is too. So we’ll tie again, getting a relative score of zero points.
So regardless of what happens, if we chose Kentucky, we’ll get a relative score of zero points. Therefore, our expected relative score for choosing Kentucky is zero points.
Lets see what happens if we choose Florida instead. 10% of the time we’ll be right, and 100% of the time we’re right, our opponents will be wrong (because they all chose Kentucky) and we’ll come out 1 point ahead. This leaves us with a relative score of 1 point.
Now 90% of the time we’ll be wrong (Kentucky will win) and 100% we’re wrong, our opponents will get one point, and we’d have a relative score of -1 points.
Summing up, if we pick Florida, 10% of the time we’ll get 1 point, and 90% we’ll get -1 points. Our expected relative score will reflect both the chance we lose relative points and the chance we score relative points, because, either could happen.
Our expected relative score for picking Florida comes out to be -0.8 points. We’d like to avoid a negative result, so in this case we would choose Kentucky, because our expected relative score for choosing Kentucky is zero and we are likely to break even.
SmartBracket calculates the expected relative score of picking each team to advance to each round. (Which includes the expected relative score of picking that team to win each round and all rounds previous.)
Then, using the Rapid Recursive® method, it determines which full bracket strategy will maximize your expected relative score. SmartBracket weighs the pros and cons of each pick and each round, and crafts a holistic strategy to help you get more points than anyone else in your pool.
The Recursive Method
SmartBracket is powered by the patent pending Rapid Recursive® method. Recursive methods are designed to evaluate future risks and opportunities by breaking large and complex problems, (such as filling out an entire bracket) into a series of smaller, more addressable problems (such as which team to pick in a given game).
The Rapid Recursive® method enables SmartBracket to evaluate all nine-quintillion possible brackets in less than a second. Instead of evaluating every possible path individually, recursive methods evaluate each step of a path forward and string these steps together to create full paths. This significantly more efficient and powerful method allows SmartBracket to distill nine-quintillion possible brackets into just sixty-three one-game choices.
The Rapid Recursive® method has demonstrable applications in a range of industries, notably, Marketing Strategy and Customer Life-Time Value, Inventory Management, Policy Risk, and Theoretical Economics. To learn more about the Rapid Recursive ® method, click here.