Possible projects for someone good at coding

Home for all your discussion of basketball statistical analysis.
Post Reply
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Possible projects for someone good at coding

Post by Crow »

Here's one:

Look at personal and team defensive results when someone goes high usage on offense.

Like anytime a player uses 3 of 5 consecutive possessions or something like that.

Do they get tired or slack on D? Do their teammates get slack? Does it matter what the offensive results of those possessions were?

Look at matchup and team data. Shot defense and overall. Defensive shot charts too.

Anyone able and interested?

Short of doing this, one could look at shot defense and overall of players across the league with usage over 25-28-30% versus everyone else or lower bins. And team data with them on court vs. not. Can do it easy for individuals but don't know the trends until you look for them.
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: Possible projects for someone good at coding

Post by Crow »

Lineup data could be looked at too. I saw someone say today that they saw a trend of high usage players being in lineups with poor defense but it was a brief mention and I forgot to link it.
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: Possible projects for someone good at coding

Post by Crow »

Another project could be a sortable trade database. Sortable by give / get position, salary or total contract range, level of draft assets, combinations of the above for specific GMs and league as whole. Inter-conference, intra conference. By age, upcoming free agent status, positional depths and age / contract status within that, etc.

Could try to build a machine learning model off it.
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: Possible projects for someone good at coding

Post by Crow »

Another project that wouldn't necessarily involve coding would be looking at four factor profiles of the team which GMs to be got their most training vs. what their own teams looked like as GM.

An example: Presti.

Looked at his 7 yrs with Spurs vs. 6 prime early years in OKC after 2 transition seasons.

Avg. Team relative offensive rating for prime Thunder was almost 300% of Spurs avg. during his stay there. Relative defensive rating only about 1/3rd as good.

At factor level avg. on offense, Thunder were ranked 5 spots worse on team efg% but still above league average. OR%, 13 spots worse and near bottom. Own turnovers, 8 spots better and above lg avg. On FT/FGA, 10 spots better and elite.

On defense, it was 6 spots worse on opp. Efg%, -7 on DR%, -12 on opp. TO and -13 on fouling.

So not much similarity. Less good at everything basically except just taking the ball to the hoop.
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Possible projects for someone good at coding

Post by josecarlos »

Hi!!!

I would like to help you in your projects. But, I am not totally able right now and sadly I have no knowledge about machine learning.

If you know about Machine Learning and know how I can start to learn it I would be glad it to do it!!!
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: Possible projects for someone good at coding

Post by Crow »

Plenty of machine learning books and courses out there.

Undertake if / when you want at your own pace.

A trade model might be too much for a first project. Thanks for the interest though.
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Possible projects for someone good at coding

Post by josecarlos »

Crow wrote: Tue May 26, 2020 2:43 am Plenty of machine learning books and courses out there.

Undertake if / when you want at your own pace.

A trade model might be too much for a first project. Thanks for the interest though.
Thanks for the recommendation and ideas!!!
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: Possible projects for someone good at coding

Post by Crow »

Does being married and / or having kids have discernable impacts on player free agent behavior? Maximizing guaranteed years early vs risk to make more later? Willingness to move? Might be valuable analysis for teams. Would need to do a lot of leg work to get a complete / accurate household database. Has any team done it? I'd guess no to few on comorehensive basis. Assume it is known info on own players and key targets. That knowledge is good but comprehensive analysis is the way to learn and leverage trends.
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: Possible projects for someone good at coding

Post by Crow »

Trade database could identify what it takes to get targets on average and then you could look at teams that set up for such a trade possibility. Coincidentally or perhaps explicitly.
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Possible projects for someone good at coding

Post by josecarlos »

Thanks for the recommendations.

I came to recommend a video conference but I have seen already there is a post about it. :)

http://apbr.org/metrics/viewtopic.php?f=2&t=9855
Crow
Posts: 10533
Joined: Thu Apr 14, 2011 11:10 pm

Re: Possible projects for someone good at coding

Post by Crow »

List of research topics.

https://twitter.com/bballstrategy/statu ... 61155?s=20

Coding probably helpful for many.

Starts with time, interest and motivation.
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Possible projects for someone good at coding

Post by josecarlos »

How this post is dedicated to coding, I have found this github repository that I post it here for those who may be interested:

https://github.com/anpatton/basic-nba-tutorials
Post Reply