The key assumption of the model is that there exist a relation of the teams with higher winning percentage to the teams that cover the bet. And that there is a relation of the teams that lose the most (with the lowest winning average) to the teams that no cover the bet.

In this context, cover the bet means that the team satisfies the betting line points or point spread and wins the bet. On the other hand, no cover the bet, means that the team does not satisfies the betting line points and loses the bet.

First, let’s define the betting variables:

W/L
	- win	Win a game match
	- lose	Lose a game match
C/N
	- cover	Win a bet by covering the betting line points
	- no cover	Lose a bet by not covering the betting line points
	- push	No action, no cover neither no_cover
O/U
	- over	Over the total points line
	- under	Under the total points line
	- push	No action, no over neither under

The relation of winning a game match to cover the bet is essential for answering the question: does the winning teams always cover the point spread?. This is because, it is easier to pick which team will win a game match instead of which team will cover the bet. The reason, is that cover the point spread can be applied to a team that loses the game match.

In this sense, sports magazines and sportsbooks predict and make available their winning ranking of teams before the NBA season starts. Hence, if the proposed relation exist between the variables win and cover, this factor can help the bettor to counteract the house edge.

The data for the presented study consists of all the betting lines and NBA game results since the 1990-1991 season to the 2013-2014. These records were collected from local sportsbooks in different countries, Mexico, US, UK and Austria. The data comprises hundreds of games divided in 24 seasons.

The SPXS - Sports Picks Expert System is used for implementing a combinatorial regression to identify underlying correlations between the betting variables in the dataset. The expert system analyzes the three main variables that results from a game match: W/L (win/lose) the game match; C/N (cover/no cover) the betting line (point spread); O/U (over/under) total points line (TPL) and their differences or error with respect to the casinos betting lines: delta w is the difference of points from the final score in a game match between two teams, delta c is the error in points with respect to the favorite team to win the game match (ATS) and delta o is the error in points with respect to the TPL.