Although finding correlations and analyzing past data is interesting, it is ultimately a meaningless endeavor unless it can predict future trends. Having found a strong link in the 2009-13 seasons between how many straight-up wins a team achieves and how many wins against the spread (WATS) they accrue, I wanted to check to make sure that this was trend holding true for the current NBA season (2014-15). I figured that the all-star break would be a natural point for this analysis.
I employed the exact same method for this midseason analysis as I did for my analysis of the 2009-13 data, using a Pearson Correlation to see if there was a significant interaction between the number of straight up wins a team had at the all-star break and the number of WATS a team had at the same point. Seeing as I had already discovered a strong positive correlation between these two variables, I expected to find the same results this time around.
I found that Straight Up Wins at the 2014-15 all-star break were significantly correlated with WATS at the 2014-15 all-star break. The Pearson Correlation for this data set was .42, which is significant at the .021 level and is of a high enough significance for the null hypothesis to be defeated. As you can from the visualization of the data in the graph below, this correlation was not quite as strong as the one that I uncovered in the data from 2009-13. However, this correlation is still quite strong, and the difference can partly be chalked up to us using a smaller sample size.
When comparing the 2009-13 analysis to the midseason 2014-15 analysis, an interesting issue presents itself: The teams that have historically fared best against the spread over the past 5 seasons, are not the top teams against the spread so far this season. If we take a look at the top 5 ranking teams against the spread over the 2009-13 seasons, we come up with Oklahoma City (252 WATS), Miami (249), San Antonio (249), Dallas (230), and Memphis (227). How are these teams doing rank-wise in WATS this season? They are 20th, 22nd, 23rd, 13th, and 14th respectively.
This becomes a problem if you are attempting to predict a teams future success (both against the spread and straight up) based on past performance. In many ways this isn’t surprising at all, as these teams have all changed their personnel over the course of 6 years (especially Miami). However, if you asked me (an admittedly non-fan when it comes to the NBA) at the beginning of the season which teams would have the best records both against the spread and straight up at the all-star break, I would have picked San Antonio, Cleveland, and Chicago. Many pundits would have agreed with me, but these teams have all underperformed to some extent thus far, and are currently ranked at 22nd, 18th, and 17th respectively in WATS. Additionally, the above teams have a collective win rate against the spread of only 45.5% as of this post.
Two points emerge from this midseason check-in: One, relying on preseason predictions and historical data can lead us into making some incorrect assumptions. Secondly, on a more optimistic note, it may be possible to take advantage of preconceived notions about a team in order to have success against the spread; as of course if we had bet against the above teams thus far this season, we would have had a success rate of 54.5%.