Research On The Competitive Balance Between Seasons In The NBA Using A Bayesian State-space Model

As an important part of culture industry,the professional sports industry which occupies an important position in economic life,is taking an active part in the driving of other industries such as sports culture,sports apparel,spors equipment and the modern media industry,making an huge contribution to the development and boom of modern economy.Many economists begin to study every aspect of professional sports league for this reason and push the research of professional sports league to a new high of economic research.To the economists' opinion,the competitive balance is a key factor to the rapid development of professional sports for a long time and reaearchs on competitive balance have increasingly became an important research project in the study field of professional sports.An important strand of the competitive balance literature is the measureing of competitive balance between seasons which is also the hot and difficult point of competitive balance literature.The mostly used methods by researchers to solve this problem is competitive balance index.This literature thoroughly researched those methods and found two limitations.Firstly,those index methods only focus on the variance of team strengths but ignored the season-to-season changers in team strengths.Secondly,they didn't take a thorough view on the peculiarity of each team when measuring competitive balance.In the context of above two limitations this literature make use of Bayesian model and MCMC methods,highlight both the variance and autocorrelation when analyzing competitive balance between seasons.By leading the hole frame into the state-space architecture,modeling the variance and autocorrelation in a time regression framework we are able to simultaneously compare the variance and autocorrelation in team strengths over seasons which not only made a better measurement of the change of competitive balance between seasons but also captured the evolution of team strength between seasons.Team strength were measured in the individual game level.Taking an advantage of hierarchical bayesian model,we are able to make dynamic posterior sampler to the data of individual game level.The dynamic posterior sampler in Hierarchical Bayesian model can fit the data to an maximum extreme,and take a thorough view on the peculiarity of each team when measuring team strength.This literature make an better break through to the limitations of competitive balance index,and well reflect the full view of competitive balance between seasons in the NBA from 1995 to 2014.According to the analysis results,the variance in NBA team strengths has been little change and decreased very slimly over seasons,while the autocorrelation in team strengths have a little rising trend over seasons.