I had previously determined that there is virtually no difference in WATS (Wins Against The Spread) percentage between favorites and underdogs in the NBA from 2009-2013. One might naturally assume that this finding could be extrapolated to mean that there is also no difference in WATS percentage between winning teams (who would be favorites more often) and losing teams (who would be underdogs more often). While this might be a solid starting point for a prediction, there is certainly not enough of a correlation here to rule out testing this hypothesis, as of course good teams can be underdogs against better teams, and bad teams can be favorites against worse teams.
Rather than simply splitting teams up by whether they had a winning or losing record over the past 5 years and comparing the average WATS for these groups, I decided we could paint a more accurate and detailed picture by running a Pearson Correlation between regular wins and wins against the spread. What would it mean to find a statistically significant correlation here, either negative or positive? Finding a positive correlation would indicate that the more regular wins a team amasses, the more likely it is to beat the spread. This finding would also seem to suggest that bettors generally either underestimate good teams or overestimate bad teams. Finding a negative correlation would indicate the exact opposite, that the more regular wins a team has the less likely it is to beat the spread, and that bettors typically overestimate good teams and underestimate bad ones. As for my predictions going in, I was expecting to find that the null hypothesis would prevail, just as it had in my analysis of favorites vs. underdogs.
Results: I found that Regular Wins from 2009-2013 were significantly correlated with WATS from 2009-2013. The Pearson Correlation for this data set was .771 (see graph below), which is significant at the .01 level and is a high enough significance for the null hypothesis to be defeated.
Analysis: The results indicate a significant positive correlation between Regular Wins for a given team and WATS for that team, meaning that for this data set, the more wins a team had the more likely it was to beat the spread.
What does this mean in terms of results I have found in previous analyses? For one, I realized in retrospect that perhaps my analysis for Favorite WATS % vs. Underdog WATS % was flawed in that I failed to include the magnitude of the spread in each of these cases. It seems plausible that if I had run a correlation of the average magnitude of a team’s spread with its chances of beating the spread, I might have uncovered a significant result. Something to consider for the future. Secondly, I immediately began to wonder if I could somehow use this finding in conjunction with another analysis I had ran earlier, showing a significant bias towards Away teams beating the spread, in order to develop a winning betting strategy. For the time being however, would it be possible to construct a winning strategy based solely on the findings above?
In the 5 seasons between 2009 and 2013, there were a total of 394 games played for each team in the NBA (rather than expected number of 410 due to the 2011-2012 season being shortened by lockout to only 66 games). As we have established earlier, in order to overcome the vig charged by casinos, a profitable betting strategy would have to beat the spread more than 52.4% of the time, or in this case accumulate more than 206 WATS in the given time frame. So if we were to employ a betting strategy where we only bet on the top 5 winningest teams, would any of them meet this threshold? As it turns out, all of them would. The five winningest teams from our given time period are San Antonio (281 wins), Miami (272 wins), Oklahoma City (271 wins), Chicago (246 wins), and Dallas (238 wins). These teams had WATS totals of 259, 249, 252, 225, and 230 respectively.
This is a very encouraging finding, as it seems that this would be a simple and effective way to consistently turn a profit betting on NBA games. However, there is one caveat: This is a long-term strategy, as Miami, Dallas, and Chicago have all recorded individual seasons between 2009 and 2013 in which their WATS totals were below a profitable level (52.4%). The challenge in this case would be to how to predict which teams would have the highest win totals 5 years from now.