APBRmetrics

For the top 6 teams on SRS (I excluded the Suns from that group for now), the average team performance is about 108 offensive efficiency and 101.5 on defensive efficiency (with 3 above and below on each). That is what teams most favored to make conference finals are doing right now. Will change some. Not sure how much, how significant.

Interesting. With the league running at 106.7 ORtg, these top teams are much more distinguished on defense than offense.
Five of the top 6 DRtg teams are among these elite. The Bucks are weakest on D at #10.

On ORtg, the elite 6 are spread from #1 Mil to #26 Utah. Very little correlation with success.

So it appears NBA.com has blacklisted my IP address which prevents me from generating the matchup files for RAPM. One location option in my VPN worked for a couple days, but that seems to be blacklisted now as well. I've tried several other location options on my VPN, but have yet to find another one that works. So I won't be able to run RAPM again until I get this resolved.

NBA recently added a "security" feature that messed up pretty much all code using their API. I'd reach out to Ryan Davis (https://twitter.com/rd11490) because he has a solution that he can explain far better than myself.

I found a twitter thread that appeared to be discussing this problem. It said the solution was to set a Referer header value of stats.nba.com. A while back I had to add a Host header value to fix a similar issue where my requests hung indefinitely. So I just changed the header parameter from Host to Referer and it seems to be working again. Thanks for the feedback to help lead me in the right direction.

Here's an update for RAPM through games played on 11/13:

https://docs.google.com/spreadsheets/d/ ... ingle=true

Select early RAPM estimate notes:

Tatum at #1 by a comfortable margin.

Leonard a tiny bit ahead of Siakam.

Beverley 6th despite lousy but insufficient in scope boxscore stats.

Many other early perhaps not sustainable surprises in top 25 including Brunson and Hardaway.

Harden 32, Westbrook 368.

Doncic 278, Porzingis 426.

Jaylen Brown 39, Kemba Walker 53, Embid 54, Saric 56, J Butler 63, Drummond 90, Coney 101, Irving 127, Fultz 129, B Simmons 157, Aaron Gordon 196, SGA 240, Capela 256, Grant Williams 273, Lonzo Ball 282, PJ Tucker 298, T Young 310, Morant 311, DeRozan 325, Hield 338, Aldridge 358, Beal 365, I Thomas 367, M Turner 372, Randle 406, Huerter 402, LaVine 412, B Ingram 414, Barrett 419, Draymond Green 421, Marcus Morris Sr. 425, Reddish 427, D Russell 429.

The actual estimates matter more if the focus is on own team. Look them up if you want.

Clippers and Jazz off to weak starts on the road. You'd want better from the Sixers and Raptors but they are ok. The best are Lakers, Celtics, Rockets, Nuggets and Bucks. Magic and thunder with no road wins yet.

Lakers, Celtics, Heat, Clippers and Bucks with the most 10 plus point wins. Nuggets with the most close game wins among good teams. Hornets and Griz have enhanced their records greatly with close wins.

shadow wrote: ↑Thu Nov 14, 2019 7:17 pm I found a twitter thread that appeared to be discussing this problem. It said the solution was to set a Referer header value of stats.nba.com. A while back I had to add a Host header value to fix a similar issue where my requests hung indefinitely. So I just changed the header parameter from Host to Referer and it seems to be working again. Thanks for the feedback to help lead me in the right direction.

Here's an update for RAPM through games played on 11/13:

https://docs.google.com/spreadsheets/d/ ... ingle=true

Seems like the date is set to always reflect the previous day. Is the data also being updated on a daily basis (I would presume not)?

I manually change the date when I update the data. So it should always be in sync. I was running it on almost a daily basis recently to make sure the header fix was still working. I will probably only be updating it once every couple weeks going forward. It doesn't change much on a daily basis.

It is probably typically higher early in season than at season's end, but the league's average to% right now is higher than it has ended up over last 23 years.

Does Second Spectrum or anyone else count total screens set? My eye says there is a big increase but would want to verify.

On screen assists, the number of players with 4 or more per game has gone from 8 in 2016-17 (first year of data at nba.com) to 18 right now.

PIPM estimates (1 yr):

Trae Young, barely overall neutral. Offensive impact almost completely offset by defensive impact.

D Hunter, modestly weak on both sides of the court.

Reddish sucks.

Tatum with more impact than Walker and 4 times Brown.

Carter and Dunn lead the Bulls. LaVine worst impact.

Sexton, Garland, Porter all moderately negative.

Porzingis barely positive. Would not be surprised if he ends up asking for a trade in 8-20 months.

Draymond just short of neutral here.

PJ Tucker, mild negative.

Kuzma -1.

SGA -1.

A Gordon -0.7.

DeRozan -1.7.

Conley improved to 0.6.

Rondae Hollis-Jefferson has played well for Raptors... as a SF who does shoot 3s much. Terence Davis also doing very well. Raptors very much in the thick of contention.

I was more favorable to Mikal Bridges than Miles Bridges at drafting and I still am.

Hawks, awful offense, awful defense. -10 pts / 100 possessions with Trae Young and slight worse without.