Meet @robotke. @robotke is a naive betting robot. Ultimately, its supposed to be chatty but at the moment is a bit passive; it tells you which teams to bet on. Its a Bayesian machine. An experiment. It learn every single day. Someday, I hope, it will not be so naive. Use with caution.
We are a betting nation, and I might add, its the latest craze. I have seen everyone betting, from those who know nothing about football, to football fans turn analysts. Its like an open feast for vultures. Only the carcasses think they are the vulture. The vultures, the real vultures, the increasing numbers of betting companies, are swooping in for the kill. An orgy of a feast.
Now that everyone is in on this betting thing, I wanted to find out exactly how a greedy naive algorithm would do. This is a project to find out just how often the robot would win. Winning a single bet is great. But the robot is ambitious, and wants big wins. Jackpot maybe. The algorithm needs to be greedy, like all of us. Big wins, big odds. So multi-bets.
Building the robot
Talking with a few people who actively bet, I got a feel of how they scan for winners. I especially talked with non-football fans. Basically it breaks down to two processes. Finding the winning streak of a teams, then betting one with the longest winning, or losing streak.
So, I figured, the bare minimal AI for this robot needs to mirror this process. Check winning streak for the teams, calculate the probabilities of winning, and optimize for the best picks.
Follow @robotke for comments and insights.