EuroLog #3: Inequality Analysis
And D. Hall should have been the MVP.
Fener (FEN) won the Final Four. I think it was well deserved — I watched both games and they mainly looked like the better team compared to their opponents. With that being said, I don’t understand why the MVP went to Nigel. D. Hall, throughout the both games, really stepped up. In the table below, we have players’ TS% change in final four (FF) compared to regular season (RS), alongside the USG% change. Keep that in mind, these are aggregated through the both games. 3rd place match was more of a scrimmage which adds to the noise that comes from the low sample size.
Devon Hall’s shooting performance is apparent here and I believe he should have taken the MVP award. I’d be OK to Marko taking it as well, since he scored during times where FEN seemed a bit off.
Anyway, congrats to Fener.
Gini From Economics in Basketball?
I listen to SocratesDergi’s EuroStep if I have the chance (mostly while going to work). Throughout the episodes, mainly İbrahim Kutluay kept mentioning how everybody can step up in Fener’s squad. This gave me an idea to check the inequality via gini, which I have learned before.
Gini is mainly used for checking the inequality of income for countries. I came across it in statistical models from social sciences where gini is included as country-level variable. The value is large if small fraction of the population has big portion of the wealth — in extreme, if a single person has all the wealth. The value is small if the distribution is relatively equal (e.g., everyone has similar wealth). We are going to carry this idea to basketball, in respect to points scored (but it’s possible to carry it out to any variable of interest).
The plot above is called Lorenz Curve. On the y axis, we have fraction of the total points scored while on the x-axis, cumulative share of players (from the lowest scorers upward). It tells you how concentrated is points among players. I think the following tables make it more clear:
CUM_PTS stand for cumulative points scored, CPTS% give you the cumulative percentages. You can compare the both teams by looking at nth row in both tables: For example, for Efes, 7 players account for 81% of the points while for Fener, 75.2%.
Let’s check the whole league:
In the y-axis, I included points per game instead of total points since teams that made into the post-season played more games, hence had the chance to score more points.
It seems like teams that made into post-season have more of an inequality which makes me think about somewhat related discussion, strong-link and weak-link sports. Another blog post topic has appeared :)
I thought about another way of looking at it, because maybe İbrahim Kutluay meant it per game basis. So, how does the distribution of most scorers per game look like for the teams? Let me show you how it looks like for Efes and Fener, since Efes seems to be the one with the least inequality when it comes to PTS.
I only used regular season for this one and also, in a tie scenario I pick randomly so these are slightly off.
Does this mean Kutluay is wrong? Not necessarily. Because we are looking at whole points or whole games, but changing the analysis to certain lineups, certain time periods within the game etc. may yield different results. I may actually check that in the future, for possessions with ‘high’ and ‘very high’ importance.