If one item outside of hiring Alvin Gentry has dominated the reasoning behind the New Orleans Pelicans improving next year, it has been continuity. The simple thought that a team that keeps a majority of its players from one year to the next will be inherently better due to "chemistry." "Chemistry" is impossible to quantify, but there is a statistical way to quantify roster continuity and see if there is a correlation between that continuity and a team’s success. That statistical correlation can be employed to model the change in winning percentage as continuity changes, and that model can give one hell of a prediction for the Pelicans record this year. So, follow along and see where it takes you.
First and foremost, for this piece, continuity is defined as the percentage of a team’s minutes played by players who were on the roster at the end of the previous season. To get an idea of continuity percentages, all historical data for team’s continuity can be found here.
Breaking down that continuity data a little bit, the league average for continuity over the last 10 years is 66%, or roughly 2/3 of the minutes from year-to-year are taken up by a player who was on the team the previous year. The Pelicans have struggled in this department over the last 10 years with an average continuity percentage of only 59%, and in the basement years of 2010 to 2013, the continuity percentage did not make it above 50%. Putting all of this in perspective, the Spurs over the last 10 years have enjoyed a continuity percentage of 83%, and the NBA championship over the last 10 years has averaged a 77% continuity rate. On the surface, continuity seems to be a very good thing.
Going deeper though, a statistical model examining the correlation between continuity and winning percentage makes the relationship even clearer. In statistics, the correlation between two sets of numbers is defined by an R-value. A positive R-value means a positive linear correlation, and a negative R-value means a negative linear relationship. R-values fluctuate between -1 and 1, but statisticians usually consider an R-value above .5 or below -.5 as an indicator of a legitimate correlation. For the last four years, the R-value for correlation between win percentage and continuity across the whole league averaged 0.57. A positive relationship exists between continuity and winning.
At this point, I did not feel as if we had enough numerical base to set up any type of model. All I saw from the above study is that the numbers back up the common sense that as a team plays more together, they should perform better. For this study of continuity and winning, I needed to directly examine the effect of choosing to remain continuous over making major roster changes. I needed to see if there was any relationship in the change in continuity from one year to the next and the change in win percentage. Again, over the last four years across the entire league the R value between the change in win percentage and the change in continuity was .44. What this means is that there is a positive correlation between choosing to remain continuous and winning. For everyone who says "rolling it back" is a bad idea, statistics say you are mistaken.
Now, at this point, the numbers look promising for the Pelicans decision, but I took one more step to ensure that these numbers were legitimate. I ran the p-values across all of the above relationships. The p-value is a checking mechanism to ensure that this correlation is not random. Its job is to give the percent chance that one thing does not directly cause another thing. For the increase that I saw in continuity and winning percentage, a 1 in 1000 chance exists that the correlation does not actually exist. For the statistics on choosing to "roll it back," the chances that that correlation is random is exactly 1.3%. Ladies and gentlemen, we have a legitimate statistical model!!
Now that we have this model, how does it directly apply to the Pelicans? The Pelicans have found an even stronger correlation between continuity and winning than the league average. Over the last 10 years, the correlation between continuity and winning percentage looks like this graphically.
The dashed line in the center is the trend line of win percentage and continuity based on data from the last 10 years. Using this trend line, one can predict the winning percentage for the Pelicans based on the continuity percentage. This prediction is defined by the equation on the graph where ‘x’ is the continuity percentage and ‘y’ is the win percentage. Conservatively estimating the continuity percentage for the Pelicans at 85%, this model predicts the Pelicans to have a record of 56-26. Woah!!
Now, at this point, I can understand anyone who is saying that a 56 win prediction from this model is silly. However, let me offer one more offer of proof that takes into account possibly our greatest addition, Alvin Gentry. Over the last 10 years, only 16 teams have maintained an above league average continuity, but changed coaches. The average change in win percentage for those 16 teams was a 25% increase. A 25% increase in win percentage again equates to a 56 win season.
Teams are aware that there are more ways to win in the NBA than simply going out and signing free agents who may or may not be better than an existing player. One way to do that is allow players to build that mythical "chemistry." Couple that chemistry with a more suitable style and a little better health, and a 56 win season may just be possible.