APBRmetrics

Did a little project yesterday, was looking into how often players get their FG misses rebounded by his team compared to teammates misses, PPP on their missed shots second chances vs teammates second chances, and got their POE by OREB frequency compared to teammate average, Second Chance PPP on their misses compared to teammate average, and both together, and also have a version using per 100 possessions.

I would say its a nice potential underlying factor if O-RAPM are a tad different than expected. Not to say its all noise, some of the names at the top clearly make sense (Zion, Drummond), but more so in the sense that some of the negative POE markers are interesting to look into, Pels AD, Nets KD, some years of Embiid were some interesting ones.

I would say its not necessarily noise if a player's shot doesnt get rebounded much, but I would say it often can be, or at least isnt a fundamental flaw in that player's offensive approach or anything, we see KD's % of own shots OREB by team shoot up after he leaves the nets to be about average compared his nets tenure, AD shoot up in 2019 and 2015 in comparison to 2016-2018 where it was incredibly low, Embiid too compared to 2021 and 2022. If a mark is really high or really low maybe that's something to look into without necessarily being indicative of the player himself, if anything. Numbers were rounded.

Note: This does not include FT misses, as that feels like a completely different thing and would just lead to high FTA players being arbitrarily low in misses rebounded

Results on teammate or Own shot OREB% is slightly different than PBP stats, maybe something to do with me including team rebounds as well? they are similar though.

https://docs.google.com/spreadsheets/d/ ... =320727342

The total possessions are the season total, so players on multiple teams will have two rows with their season total, so their per 100 Poss POE will be slightly deflated, and adding together would get their actual season mark. Getting it to be Player + Offensive Team possessions shouldn't be difficult though, but I want to figure out a way to host this on a better place where its easier to access without downloading or it being slow to load. DSMok gave me some advice to look into TABLEAU, might try that out.

https://docs.google.com/spreadsheets/d/ ... =928721534

With offensive possessions per team instead ^

This is an important topic, little or never presented. Rebounding and pts off rebounding are significantly different. Randomness and attention to when to shoot to get best residuals. Shared responsibility with others but initiated by shooter. How many think about it much, act to maximize? How much do coaches harp on it?

I'd look at this data far more if a team was listening.

Figured out how to sort by team and year with second link using column drop down menus. Not obvious. Could use a mention of any such details needed for use. Even simpler display could help. But access at all is a big deal.


First link didn't have way to my knowledge to sort. Not sure who would use data from more than 3 years ago. I guess could if studying historical team construction or full player evolution but I'd think nearly all use would be fairly recent. Crop the database for less intense loads and load times? That is something that could be considered.

Crow wrote: ↑Sat Oct 05, 2024 3:55 pm This is an important topic, little or never presented. Rebounding and pts off rebounding are significantly different. Randomness and attention to when to shoot to get best residuals. Shared responsibility with others but initiated by shooter. How many think about it much, act to maximize? How much do coaches harp on it?

I'd look at this data far more if a team was listening.

Figured out how to sort by team and year with second link using column drop down menus. Not obvious. Could use a mention of any such details needed for use. Even simpler display could help. But access at all is a big deal.


First link didn't have way to my knowledge to sort. Not sure who would use data from more than 3 years ago. I guess could if studying historical team construction or full player evolution but I'd think nearly all use would be fairly recent. Crop the database for less intense loads and load times? That is something that could be considered.

I suddenly realized I can use Rshiny

https://timothywij123.shinyapps.io/OREB ... yerMisses/

The search bar you can look up either player or name, the rest you can organize by season or minimum possessions and stuff.

For Thunder, Giddey misses were far more effectively followed up than normal. Wiggins and Wallace, modestly above expected.

SGA and Holmgren misses saw horrible followup results.

This will be an area to check with Hartenstein involved.


4 main Celtics have poor miss followup activity. Tatum and Horford better than expected.

Aaron Gordon and Jokic miss followup is very strong. Because of themselves and the other. These could be a central argument for the 2 big strategy.


Followup after Doncic shots is just average. Brunson, mildly negative. Giannis moderately positive, Lillard mildly negative.
To mention a few.

James strong positive, A Davis very negative. Andrew Wiggins, strong negative, D Green mild negative, Curry mild positive. Leonard and Harden strong positives. P George strong negative. Embid mildly negative.

Results sometimes but not always predictable. Could be partly random but shooting with strong rebounders in position can be helpful.


For a team version, could want to see data with a particular non-shooter on the court or not. Not asking for it, just mentioning the possibility.

Full dataset is great for the intense user. A slim version with only a couple of outcome columns might be a cleaner, simpler solution for more of the audience (from a design review perspective, fwiw). Even if keep all columns, name and total impact closer together could help.


Good wishes in pursuing team interest. Appears you have strong chops to do useful work. Just keep up efforts to make things easier / sinpler for the user.

Having an interesting database is the necessary start. Using it hard is the next step. Getting decision-makers including players to pay attention is the step after that. Updating, review and refinement of database and actions after that.