oh no it’s almost 2024 and i’m still thinking about RAPM!
people have been running RAPM models for like 20 yrs now. i think it’s mostly great, but it does get shit on quite a lot, often for “lacking context.” mostly, this is bullshit...
what jumps out?
that i used to think this was aNaLyTiCs when i knew nothing, but now i know a little and realize all the issues...
- no scale on axes!!!
- arbitrary cutoffs (why 15 games?)
- inconsistent text labels
- NET capitalized like it's an acronym when it actually isn't
some news:
- im graduating in 3 weeks wtf
- today was my last day doing bball analytics for the
@brooklynnets
. worked with fantastic people and learned a ton!
- in the fall, i'll be going back to
@nextdoor
full-time as a software engineer. excited!
here goes an analytics vibes thread without any actual numbers
all of these XYZ Over Expected (CPOE, RYOE, etc) metrics are just methodologically WRONG! how can we attribute the entire residual to a single player?! reading residuals is like astrology for pff nerds lol
It’s official:
@DiamondKinetics
x
@sidelinehd
🎉
Today, Diamond Kinetics announced the acquisition of sidelineHD, the youth sports media platform specializing in livestreaming and video highlights, in a move that accelerates innovation in youth sports technology.
automatically extracting players (and refs 😏) from nba video. definitely want to label the players with their identity first. what can I do with this after that? soliciting ideas!
using the matchup data on the nba stats site, i created a graph with nodes being players and edges weighted by how much players guarded each other. i ran louvain community detection algorithm on it and it found 4 communities. this means there’s “naturally” 4 positions!
@javavvitch
@devuxer
no you lol. id rather fire the people who agree with your tweet than those who use a productivity enhancing tool. wouldnt be a hard decision either!
⬇️ THREAD ⬇️
To chart our progress and to avoid going back to square one, we are establishing a new COVID Alert System run by a new Joint Biosecurity Centre.
That COVID Alert Level will be determined primarily by R and the number of coronavirus cases.
#StayAlert
Alright I’ve had enough of Kirk Goldsberry. There is no way he made this video (and others like it) but has the gall to take credit for it, putting his name before the person who most likely actually made this - Vanna Bushong who is clearly the graphics expert here.
The Rockets Took 61 3s in a 48 minute game!
They scored 63 points on 3-pointers
They scored 30 points on 2-pointers
They scored 27 points at the line
They beat the team with the best record in the league
Someone in South Carolina invested $2 into a Mega Millions ticket at a gas station, according to local news.
Current valuation: Over $1.5 billion
Buying lottery tickets FTW
Stitch Fix is using something called eigenvector decomposition, a concept from quantum mechanics, to tease apart the overlapping “notes” in an individual’s style. Using physics, the team can better understand the complexities of the clients’ style minds.
@richardludlow
@___Colb___
some are theorizing aggregate money spent on contracts will go down because of extremely punitive 2nd apron and a couple other things
suffering is voluntary for me but not everyone is that fortunate
think this is rad? donate here which goes to
@eji_org
and put my activity link in the donation comment
i classified nba players into offensive roles. will probably do a write up sometime but happy to answer any qs here!
(app made with
@_hex_tech
which is basically better shiny but for python chads)
ingles is not just a catch and shoot guy! time to dispel the myth that inefficient bench “volume” scorers like clarkson should automatically be 6th man of the year. clarkson hurts the team, ingles helps. not that tricky, Ingles 6MOY!
i HATE that masks vs no masks became left vs right. the "own the libs" mentality is so strong I wonder what would've happened if some smart left-leaning people concocted some reason to be anti-masks so that owning the libs would actually be wearing masks. is this 4D chess?!?!
@spakhm
if they're there by accident, then why is it independent of class size?
also, Berkeley is a state school.
when you come with the spicy🌶 takes, make sure they're not shitty 💩 first!
NBA play-by-play data can be infuriating! These are x, y coords of shots near the rim by season (1 unit = 0.1ft). Quite inconsistent throughout the years.
03-09: all shots are in center of basket!
10-13: yay diversity!
14-16: ??? back to dark ages
17-18: basket moved back???
Supposed 'god of visualization' produces this perplexing / useless viz. The lines on the court give us enough info on distance of shot - no additional information is conveyed by the colors. Sad!
I had fun talking to
@uditranasaria
about his team’s Big Data Bowl project. I loved the idea that tracking data can help us explore counterfactuals, quantify things that didn’t happen. Also, their stunning visuals were sold short in an audio format so I’ll include one here.
In this latest episode I was joined by
@uditranasaria
. We discuss his group’s novel way to measure passing value via physics-based models for ball and player trajectory. Then some big picture talk on the state of tracking data & public football analytics
👏 corner 3s aren't more efficient because they're closer 👏
the reason they're higher % is that they're way less likely to be off the dribble and way more likely to be open. there is some public tracking data floating around out there and you can verify this!
ran into the funniest "anti scraping" measure ive ever seen today. this site literally created their own font that renders characters as other characters lol...so like you'll see "a" in the html but visually you will see "b" -- security by obscurity yay!
fucking disgusting. asian american community needs to do better and recognize the rampant anti-blackness. stop pitting minorities against each other :(
@conorsen
change in acceptance rate over time doesnt mean all that much if you dont also look at something like applications per student (surely going up)