APBRmetrics

Hello all,

I am new to the forums, and excited to make my first post! In addition to this being my first post, this is also my first ever basketball (or any other sport, for that matter) analytics project, and also my first ever self-directed Python project. I am very open to any and all criticism. I hope I am able to start some sort of discussion or provide some insights! Now, onto my analysis:

There’s been a lot of talk recently about how NBA teams are starting to abandon efforts at getting offensive rebounds in order to play transition defense. I wanted to explore this, as well as how defensive and total rebounds have changed from the past, through numbers. Specifically, I wanted to find out if teams are having less players try to grab rebounds (offensive, defensive, and total) now than they did since 2005 with their increased emphases on transition play. So, I thought I should examine this by seeing how the ratio of team’s top rebounder’s (for each of the top 3 categories respectively) rebounds per game to total team rebounds per game has changed (or not changed) from the past. I also wanted to see this ratio for the second best rebounder from each team, and also for the sum of rebounds per game of the best and second best rebounders on each team. Specifically, I looked from the 2004-2005 season up until the 2014-2015 season (it was easiest to work with this subset of data as there was no change in the number of teams within this period).

To clarify, I picked out the best offensive rebounder, defensive rebounder, and total rebounder from each team on a seasonal basis since the 2004-2005 season, and divided each of those player’s seasonal ORPG, DRPG, and TRPG, respectively, to the team’s total ORPG, DRPG, and TRPG, respectively.

I realize these ratios are somewhat flawed in that there could be variability on each team due to factors such as new draft picks/players, trades, and the variability in free/easy defensive rebounds for the team. Further, the ratio in itself may be somewhat arbitrary. Nevertheless, my hope is that a significant change in these ratios should mark some sort of change within the team’s rebounding strategy.

I wrote the code to do this entirely in Python, and it can be found at: https://github.com/jgandhi714/Rebound_Ratios. A PDF with all 94 plots that my code produced can be found in this GitHub repository, as well as a folder with each plot saved as a separate PNG file. I was initially disappointed to find out that league-wide averages for the ratios afore-mentioned have remained relatively unchanged (first 4 plots in PDF), but many of the individual team plots show significant changes in the ratios (I'll leave it to you to take a look for yourself). Please let me know what you think (and critique my code and help me make it more efficient/extensive)!

I wonder if for this topic it might be worth looking at trend on rebounding rates for the third highest guy, the raw rates and his share of the total for the 3rd to 5th. You say the rates for 1st and 2nd didn't change much on average. If there is any notable change maybe it is the 3rd guy doing more from the perimeter in the place of one or both of the other perimeter guys? A plausible theory worth checking I think.

Crow wrote:I wonder if for this topic it might be worth looking at trend on rebounding rates for the third highest guy, the raw rates and his share of the total for the 3rd to 5th. You say the rates for 1st and 2nd didn't change much on average. If there is any notable change maybe it is the 3rd guy doing more from the perimeter in the place of one or both of the other perimeter guys? A plausible theory worth checking I think.

Perhaps I was unclear. Although the league-wide average ratios remain relatively unchanged since 04, many of the ratios for the individual teams have significant, even drastic, changes. This leads me to believe that the positive changes in the ratio happen to balance out the negative changes in the ratios across the league. I wanted to post images for the plots that do show significant differences in the ratios, but to be honest I'm not really sure how to do so in these forum posts! If you take a look at the pdf, you can find many teams that do have pretty significant changes in the ratios.

I believe I understood your point about the averages vs. individual teams. If you could cluster teams "not sending many to boards / getting back on D" versus not, maybe there is a change for group 1 compared to past or group 2 that just happens to be cancelled out by the league being nearly equally divided. And group 1 perhaps getting more press than group 2.

Did you understand my suggestion? You didn't respond to it. Coaches and the media may talk about only sending 1-2 to the boards but is a third guy on average still going in a fair amount? Has his load decreased, stayed the same or increased? That is new ground worth examining imo, especially since theory 1 wasn't supported. If #3 on average can pick up most of what was previously provided by #4 and #5, it is easier to justify not sending them in.

I haven't looked at the plots yet. Wouldn't load on my phone. You could try to copy and paste here between img buttons (provide on top of message post panel)

[img]linktoimage[/img]

If that doesn't work, Mike G knows more about posting procedures.

I understand your point now, thanks for the input! That theory does make sense to me, and it does seem worthwhile to check the ratios for the third best rebounder (and maybe even 4th and 5th while I'm at it). I will update the code and post my results soon.

I tried to post the images that way in my original post using the GitHub links but it did not work.

Good. I may not have explained sufficiently the first time. Look forward to hearing what you find.

I am not good at posting graphs or pics here. Some do make it work though. In this post viewtopic.php?f=2&t=8964&start=90 a graph got posted. using opening image bracket then http://imageaddress followed directly by closing image bracket. Note the syntax for possible clues. Did you use the http heading?

I added in new charts for the ratios of 3rd best rebounder rebounds and 4th best rebounder rebounds to team rebounds, in addition to cumulative charts (i.e. top 3 rebounder rebounds and top 4 rebounder rebounds to team rebounds). The top 3 rebounders rebounds to team rebounds chart shows that the ratio for DRPG and total RPG have dropped about 5 percentage points. Unfortunately, I am still having trouble posting the plots - I assume this has to do with the fact that they are uploaded to GitHub.

I am not sure if 5% is a significant change or just arbitrary. The team charts did once again reveal huge changes in ratios, and can be found on the originally posted GitHub link.