https://projects.fivethirtyeight.com/2020-nba-player-ratings/ Raptor apparently incorporates play by play data, player tracking data, on/off stats, and advanced efficiency stats like PER and WAR. Conclusions: 1. Eat your heart out Giannis. Harden is a net neutral to slightly positive on defense, while his offense easily carries him to Best Player status right now. 2. Capela and House are our next two most valuable guys. Way underrated defense from House, which again we can all see with our eyes. Screw the national media narrative: we are a GOOD defensive team. 3. Russell Westbrook has been a wash so far, but perhaps his latest game has been a sign of improving shooting numbers to come. 4. Eric Gordon lolol
I am curious why Giannis’s doesn’t have a higher rating per RAPTOR. Is it the poor outside shooting and free throw shooting? I think he’s probably underrated by this stat. The Bucks are really good, again, and he’s putting up crazy numbers, again.
It's probably because the East sucks, and the Bucks when Giannis is off the court have still been able to blitz teams almost as well as when Giannis is on the floor.
There's an article Nate Silver wrote which is somewhere on the site (I looked at it back when this 538 statistic/site was posted before). In it, he goes into the methodology of how they arrive at the scores for the players (and it's pretty involved). One of the things he mentioned is that it favored ball-dominant players and two-way players. I'm not sure it gets any more ball-dominant than Harden. Even in the examples it gave in that explanation that I read, it stated that RAPTOR inherently loved players like Harden, Kawhi, and Curry.
I think it's because Gianni's on/off stat is pretty mediocre(only +3.1 according to stats.nba.com). Just considering the net rating, with Giannis on court, bucks is the best team in the NBA, without him on court, they are a close second. Buck's offense and defense rating doesn't change much whether Giannis is playing or not. For Harden, when he is on court, we're +10.5 which is better than the overall net rating for any team in the NBA; when he is off, we're -9.4, which is only better than the Warriors. So Harden's on/off is +19.9, which is way better than Giannis's. I think RAPTOR's on/off measurement is a little more complicated by considering the affect of the rotation for both teammates and opponents, but even after that, Giannis's number is still much worse than Harden's.
It has to do with how good his team is. Theoretically, imagine a player that averages 35/10/10, but his team is exactly as good without him as they are with him. When he sits they have the same ortg, drtg, whatever. This imaginary player misses games and his team is still winning/losing at the same rate, his production is just evenly spread among the rest of his teammates. If this is the case, is that player really that valuable? He's soaking up production, but for what purpose? In this weird impossible hypothetical, that player deserves a +0.0 from this stat. Production for productions sake is not valuable. Anyway, I'm not saying this is what Giannis is doing, he's obviously valuable. But that's the kind of thing these stats try to suss out. If you look at the other players on the bucks, they have a lot of other players that are doing really well. Giannis can sit and the bucks increase their lead. Whereas with the rockets, Harden sits for 2 minutes and the other team goes on an 11-2 run. That gives someone like Harden a bigger share of the success. I think part of this is system based, it happens on a lot of Lebron teams too -- if the entire system is "let Harden/Lebron go to work" then they look like chickens with their heads cut off without that player in the game. So it gets a bit inflated for that reason. I think the Bucks are a lot more functional without Giannis than the Rockets are without Harden.
Maybe that’s it. Are the RAPTOR rating available for last year? Was it the same story with On/Off? Sometimes stats like this use multi-year data to smooth out the noise. I’m curious if RAPTOR does this. I skimmed over the methodology some weeks ago, and I don’t remember it mentioning that.
They do have some historical data: https://projects.fivethirtyeight.com/2020-nba-player-projections/james-harden/, but I don't think they use it only for future performance prediction but not for the current rating. I see a few noisy points, like Will Barton, Montrezl Harrell, who doesn't have a good historical rating, but is currently high on the list. So even if they do, the weight must be small.
Nate Silver said that the public Raptor model hasn’t taken into account strength of schedule yet and that in 5-10 games Giannis specifically will get a big jump on the model from good to MVP levels. So who knows how the rankings will look later. Lebron might get a boost too because of SOS.
I also remember him saying that the on/off numbers don't play as big a role in the ratings as the box numbers do. Aaannd... here we go, for anyone interested. How it works : https://fivethirtyeight.com/features/how-our-raptor-metric-works/
Starting with WB .... He's been pretty solid IMO. Yeah , his shooting needs to improve but he's added a different dimension to this team with his speed and attacking the rim early in the clock - Just as I stated he would in the first thread on the trade. Defense .... Anyone else notice how much they have improved since EG hit the trainers table ? Playing EG at the 3 spot was a big part of why this team struggled early on - House is a massive upgrade at the 3. EG's minutes need to come behind Harden and Westbrook , not at the expense of House or Tucker - Find a better option to back up those two spots please Moray. No , EG wasn't why they gave up so damn many open 3's early on .... but the lack of size was a .... Big issue. EG just aint Thikkkk like Tucker and able to defend much taller players effectively. Really too bad EG isn't tradable this year .... this team is so stacked at the guard spots , he hasn't even been missed with Rivers being a better creator and solid enough shooter. Damn , Harden as a post defender .... Chuckwagon 2.0 ?
I like checking the RAPTOR ratings from time to time. Some insights: 1) Harden is the best player in the NBA - and he is a plus defender (total ppp allowed he is #2 behind Matthews for wings, which supports this) 2) Capela has been really solid after a down or even year last year - this is the guy we need all year 3) Tucker is key to this team's success 4) Russ has not been maximized. I think part of this is because 1) RAPTOR was essentially created to devalue Russ's 16-17 season because they didn't believe the absurdly high BPM, 2) our bench kind of sucks, and when Russ plays with them, they still suck, 3) we've only seen Russ integrate himself next to Harden in the last 3 games. Whether this is due to chemistry or improving injuries, I don't know. Glass half empty - Russ is going to be an albatross and we will win in spite of him. Glass half full - the worst is over, the best is yet to come, and we are nowhere near our ceiling if Russ plays the way he has been. 5) McLemore has been a great find. This meets the eye test too. My KU-hating friend loved to rag on him, but I saw the improvement last year, and had a gut feeling he would pan out. I'm happy to see him thrive. Morey finally got his second draft win. 6) We desperately need someone besides Austin Rivers to score off the bench. Here's hoping Gordon comes back healthy and ready to go. RAPTOR has us ranked #1 in ELO and #3 in odds to win. Clutch City, we have a shot!
Man. Westbrook still below replacement level. Kinda thought these last 3 games would dig him out of that hole.
I would guess because they stay good or don't drop off as much when he is off the floor, but I haven't looked at it.
He's the third highest overall rated player in the NBA. That's a pretty good rating. It's just that Harden is completely at another level to everyone else. He's rated 50% more again than Lebron, Giannis and Luka (who are the only other guys over a 5 overall rating). Lebron also just edges him out right now due to his amazing hot start - a career high by far in assists per game so far. He may cool a bit and Giannis overtake him to be second.