Due to several requests, I am posting the data I used when researching the impact of shot locations on rebounding rates. Here's a key that will help you understand the tables.
Net Miss (Defensive Rebounding Chart)
The number of misses from close range (0-9 feet) subtracted from mid-range misses (16-23 feet). The league average for this value was then subtracted from each team. A positive number indicates more mid-range misses, which should lead to higher then average defensive rebounding rates (DRR). A negative number naturally means the opposite.
Net Miss (Offensive Rebounding Chart)
Pretty much the same, but mid-range shots were subtracted from close range shots, since a high number of close range shots equates to a higher offensive rebounding rate (ORR). Therefore, a positive number should lead to a higher then average ORR, and a negative number lower then average.
Net DRR and Net ORR
Simply each teams offensive and defensive rebounding rate for the 2011-2012 season, with the league average rates subtracted from them. This gives an idea of who is under/over rebounding.
Skill (Offensive and Defensive Chart)
A team's Net Misses subtracted from their Net Rebounding rates (either defensive or offensive). This subtracts the "system" (shot locations) from the rebounding rate, which leaves us with "skilled" (difference) rebounding numbers.
(System+Skill)=Output or Output-System=Skill
Net Miss (Defensive Rebounding Chart)
The number of misses from close range (0-9 feet) subtracted from mid-range misses (16-23 feet). The league average for this value was then subtracted from each team. A positive number indicates more mid-range misses, which should lead to higher then average defensive rebounding rates (DRR). A negative number naturally means the opposite.
Net Miss (Offensive Rebounding Chart)
Pretty much the same, but mid-range shots were subtracted from close range shots, since a high number of close range shots equates to a higher offensive rebounding rate (ORR). Therefore, a positive number should lead to a higher then average ORR, and a negative number lower then average.
Net DRR and Net ORR
Simply each teams offensive and defensive rebounding rate for the 2011-2012 season, with the league average rates subtracted from them. This gives an idea of who is under/over rebounding.
Skill (Offensive and Defensive Chart)
A team's Net Misses subtracted from their Net Rebounding rates (either defensive or offensive). This subtracts the "system" (shot locations) from the rebounding rate, which leaves us with "skilled" (difference) rebounding numbers.
(System+Skill)=Output or Output-System=Skill
Net Misses and Net ORR/DRR were the two variables regressed. The correlation between shot locations and offensive rebounds was 17.4%, which I rounded to 20%. The correlation between shot locations and defensive rebounds was originally 26.4%, but after double checking, the value came out to 22.3%. I will round this up to 25%.
Why round? Because the shot location info at Hoopdata while useful, doesn't match-up perfectly with Counthebasket's shot locations regarding rebounds. If the shot location data was exact, and every type of shot was able to be accurately analyzed, its reasonable to assume that the correlation constant would increase. Therefore, I bumped up the R2 values from 17.4% to 20% in offensive rebounding, and the 22.3% to 25% in defensive rebounding.
Another puzzling aspect of the study may have been the skill difference. How could I simply subtract missed shots from rebounding rates? Isn't that mathematically blasphemous? Well based on the chart in my first post, close range and mid range shots (the ones I included in the study), have VERY high chances of being offensive and defensive rebounds respectively. Its not known "how" high, but its clear the value is close to 100% (1). Assuming this the case, we can subtract these values from Net Rebounding Rates, as a change in rebounds of x = a change of rebounding rate of x. The results aren't perfect, which is why I only classify 2/3 of teams as truly "skilled/unskilled" rebounders, so as to minimize error.
Quick recap. My original correlation for defensive rebounding and shot location was 30%. That was incorrect, the new value is 25%. The offensive rebounding correlation is still 20%.
Comments, suggestions, critiques, and praise are welcome.
Why round? Because the shot location info at Hoopdata while useful, doesn't match-up perfectly with Counthebasket's shot locations regarding rebounds. If the shot location data was exact, and every type of shot was able to be accurately analyzed, its reasonable to assume that the correlation constant would increase. Therefore, I bumped up the R2 values from 17.4% to 20% in offensive rebounding, and the 22.3% to 25% in defensive rebounding.
Another puzzling aspect of the study may have been the skill difference. How could I simply subtract missed shots from rebounding rates? Isn't that mathematically blasphemous? Well based on the chart in my first post, close range and mid range shots (the ones I included in the study), have VERY high chances of being offensive and defensive rebounds respectively. Its not known "how" high, but its clear the value is close to 100% (1). Assuming this the case, we can subtract these values from Net Rebounding Rates, as a change in rebounds of x = a change of rebounding rate of x. The results aren't perfect, which is why I only classify 2/3 of teams as truly "skilled/unskilled" rebounders, so as to minimize error.
Quick recap. My original correlation for defensive rebounding and shot location was 30%. That was incorrect, the new value is 25%. The offensive rebounding correlation is still 20%.
Comments, suggestions, critiques, and praise are welcome.
