Nate Silver's 538.com predictions: Portland 6th in West, 3rd in division

[PtldPlatypus]

http://projects.fivethirtyeight.com/2017-nba-predictions/?ex_cid=538twitter

 

[julius]

I know Portland way over performed last year (compared to how many people thought they would be), but whats with OKC still getting love?

Utah I can see, they're a solid team. It should be a battle between Portland and Utah for the division. I'm not sure how OKC can be thought of as an odds on favorite to win the division. They're basically a ball hog and 11 other players.

I don't get how they accumulative win average for Utah is higher than OKC..but OKC has a higher chance of making the playoffs.

I think Sports simulations are usually horse shit anyways. Way too many variables to take into account.

 

[e_blazer]

Nate's getting senile.

 

[PtldPlatypus]

Nate's getting senile.

 

[Pinwheel1]

Cleveland is the defending champs and they play in a weaker conference, yet he has the Warriors winning 68 and the Cavs only 57?

Maybe 60 for the Warriors or at the most 62. I just think the Warriors will think big picture a little bit more this year and rest their stars more.

He dropped SA down 16 wins from what they got last year......interesting.

 

[THE HCP]

Nah, don't know who this is and don't know how much cred this guy has. Not sure he's on with some of these predictions if you ask me. Jazz 51 WINS? Cavs only 57? Clippers only 48? Minny 46?

 

[BlazerDuckSeahawkFan94]

Nah, don't know who this is and don't know how much cred this guy has. Not sure he's on with some of these predictions if you ask me. Jazz 51 WINS? Cavs only 57? Clippers only 48? Minny 46?

I think the Cavs only won 56 last year iirc. LeBron teams don't really chase wins like they used to.

But Nate Silver is a horses ass.

 

[Boob-No-More]

Nah, don't know who this is and don't know how much cred this guy has. Not sure he's on with some of these predictions if you ask me. Jazz 51 WINS? Cavs only 57? Clippers only 48? Minny 46?

Nate Silver and his team are numbers geeks. They are very good at what they do, but they are only as good as the numbers they crunch and the formulas they use to crunch them. They use statistical modeling, not biased opinions, as the basis for their predictions. They update their results whenever new data is available. In the case of the presidential elections, that means they are getting updated several times a day as new polling results are released. In the case of the NBA, they rerun their models every night based on the results of that days games. As with any modeling/simulation, the results are only as accurate as your model and your data. But, their predictions are 100% model/data driven and have room for opinion in the calculations.

Last year was their first year using their CARMELO model, and even in its first year, it outperformed the Las Vegas odds makers. But, no model is perfect. Last year, they predicted the Blazers to win 36.2 games, compared to 26.5 wins by the Westgate Las Vegas SuperBook, compared to 23 wins by the HCP (that's who Nate Silver is FAMS!).

This is only their second year using this model and they've made some refinements (details here). Right off the bat, I notice they have SEVERAL roster errors. For example,they still have Chris Kaman, Brian Roberts and Gerald Henderson on our roster, and no Evan Turner (BOS), Shabazz Napier (ORL) or Festus Ezeli (GSW), but they do have Jake Layman. For UTA, they have George Hill, but no Boris Diaw (SAS) or Joe Johnson (MIA), but oddly they still have Trey Burke, who was shipped out in the same trade that brought them George Hill. There are multiple roster errors for most teams. Obviously, their database is not up to date. So, I wouldn't put too much stock in this first stab at their 2016-17 predictions. I'm sure that will get cleaned up soon.

But, even once they get the rosters up to date, there are still limitations to these types of models. They don't factor in things like team chemistry, injuries, etc. Again, comparing UTA and POR last year, they predicted 43.8 wins for UTA (actual 40) and 36.2 wins for POR (44 actual). We outperformed their model by nearly eight wins and UTA under performed relative to their model by nearly 4 wins. POR had great team chemistry, that resulted in more wins than expected (the sum was greater than the whole of the parts), and UTA had significant injuries. Last year, they also predicted HOU would win 52.6 games, but they only won 41. This is an example of bad team chemistry causing a team to under perform.

Also, things like margin of victory factor into their statistical model, and that's another area where POR outperformed the model and UTA under performed. To me this shows that POR is more clutch and UTA not so much. UTA lost a LOT of really close games, a reflection of the fact that they did't have a true go to guy they can give the ball to in the closing seconds of a close game. In a nutshell, we have Damian Lillard and they don't. If you look at basketball-reference's Expected W-L for last season, the Blazers (based on strength of schedule and margin of victory) should have won 43 games and UTA should have won 46 games. Keep in mind, Expected W-L is not a prediction, it's an after the fact calculation based on actual results. Again this is indicative that UTA under performed in close games and we outperformed expectations.

Their models also doesn't take into account WHEN a player racks up his stats. For example, last season Russell Westbrook was second in the entire league in Wins Above Replacement Player at 18.3, trailing only Steph Curry (21.7) and firmly ahead of both LeBron James (16.7) and Kevin Durant (14.1). Their model actually thinks Russell Westbrook wins more games for his team than LeBron James or Kevin Durant. It also thinks he was worth more than twice as many wins as Damian Lillard (8.4). Anyone who has seen Westbrook play more than a couple times knows he racks up god like stats over the first three quarters and then becomes a hero ball playing, turnover machine, chucking up bad shots and freezing out his teammates in the 4th quarters of close games. There is a reason OKC lost 15 games during the regular season when leading in the 4th quarter. There is also a reason why they choked away a 3-1 lead over GSW in the WCF. In both cases, that reason's name is Russell Westbrook. Honestly, I know we're biased here, but who would you rather have with the ball in their hands with the game on the line, Russell Westbrook or Damian Lillard? Westbrook's epic choking explains why OKC under performed both fivethirtyeight's prediction (58 wins) and basketballl-reference's Expected W-L (59 wins) last season (55 actual wins).

So, the model needs further refinement. There are reasons why teams outperform and under perform their statistical model. Many of those factors weigh in the Blazers favor (team chemistry, go to scorer who is big in the clutch) compared to UTA (no go to scorer) and OKC (poor team chemistry and a leader who chokes big time in the clutch).

So, take these predictions with a grain of salt, or at least understand how they are calculated so you understand their conclusions and possible sources of error that could cause the actual results to vary significantly from their predictions.

BNM

 

[Chris Craig]

I don't like this Nate guy. The Jazz and Thunder ahead of us in the standings? Sixth seed? I don't know what this guy has been smoking, but I don't want any. The Blazers will be at worst 4th, but probably 3rd, below Golden State and the Clippers and ahead of San Antonio. Also, 1st in the NW...the champs are here...the champs are here...you can't see us.

 

[Chris Craig]

Nate Silver and his team are numbers geeks. They are very good at what they do, but they are only as good as the numbers they crunch and the formulas they use to crunch them. They use statistical modeling, not biased opinions, as the basis for their predictions. They update their results whenever new data is available. In the case of the presidential elections, that means they are getting updated several times a day as new polling results are released. In the case of the NBA, they rerun their models every night based on the results of that days games. As with any modeling/simulation, the results are only as accurate as your model and your data. But, their predictions are 100% model/data driven and have room for opinion in the calculations.

Last year was their first year using their CARMELO model, and even in its first year, it outperformed the Las Vegas odds makers. But, no model is perfect. Last year, they predicted the Blazers to win 36.2 games, compared to 26.5 wins by the Westgate Las Vegas SuperBook, compared to 23 wins by the HCP (that's who Nate Silver is FAMS!).

This is only their second year using this model and they've made some refinements (details here). Right off the bat, I notice they have SEVERAL roster errors. For example,they still have Chris Kaman, Brian Roberts and Gerald Henderson on our roster, and no Evan Turner (BOS), Shabazz Napier (ORL) or Festus Ezeli (GSW), but they do have Jake Layman. For UTA, they have George Hill, but no Boris Diaw (SAS) or Joe Johnson (MIA), but oddly they still have Trey Burke, who was shipped out in the same trade that brought them George Hill. There are multiple roster errors for most teams. Obviously, their database is not up to date. So, I wouldn't put too much stock in this first stab at their 2016-17 predictions. I'm sure that will get cleaned up soon.

But, even once they get the rosters up to date, there are still limitations to these types of models. They don't factor in things like team chemistry, injuries, etc. Again, comparing UTA and POR last year, they predicted 43.8 wins for UTA (actual 40) and 36.2 wins for POR (44 actual). We outperformed their model by nearly eight wins and UTA under performed relative to their model by nearly 4 wins. POR had great team chemistry, that resulted in more wins than expected (the sum was greater than the whole of the parts), and UTA had significant injuries. Last year, they also predicted HOU would win 52.6 games, but they only won 41. This is an example of bad team chemistry causing a team to under perform.

Also, things like margin of victory factor into their statistical model, and that's another area where POR outperformed the model and UTA under performed. To me this shows that POR is more clutch and UTA not so much. UTA lost a LOT of really close games, a reflection of the fact that they did't have a true go to guy they can give the ball to in the closing seconds of a close game. In a nutshell, we have Damian Lillard and they don't. If you look at basketball-reference's Expected W-L for last season, the Blazers (based on strength of schedule and margin of victory) should have won 43 games and UTA should have won 46 games. Keep in mind, Expected W-L is not a prediction, it's an after the fact calculation based on actual results. Again this is indicative that UTA under performed in close games and we outperformed expectations.

Their models also doesn't take into account WHEN a player racks up his stats. For example, last season Russell Westbrook was second in the entire league in Wins Above Replacement Player at 18.3, trailing only Steph Curry (21.7) and firmly ahead of both LeBron James (16.7) and Kevin Durant (14.1). Their model actually thinks Russell Westbrook wins more games for his team than LeBron James or Kevin Durant. It also thinks he was worth more than twice as many wins as Damian Lillard (8.4). Anyone who has seen Westbrook play more than a couple times knows he racks up god like stats over the first three quarters and then becomes a hero ball playing, turnover machine, chucking up bad shots and freezing out his teammates in the 4th quarters of close games. There is a reason OKC lost 15 games during the regular season when leading in the 4th quarter. There is also a reason why they choked away a 3-1 lead over GSW in the WCF. In both cases, that reason's name is Russell Westbrook. Honestly, I know we're biased here, but who would you rather have with the ball in their hands with the game on the line, Russell Westbrook or Damian Lillard? Westbrook's epic choking explains why OKC under performed both fivethirtyeight's prediction (58 wins) and basketballl-reference's Expected W-L (59 wins) last season (55 actual wins).

So, the model needs further refinement. There are reasons why teams outperform and under perform their statistical model. Many of those factors weigh in the Blazers favor (team chemistry, go to scorer who is big in the clutch) compared to UTA (no go to scorer) and OKC (poor team chemistry and a leader who chokes big time in the clutch).

So, take these predictions with a grain of salt, or at least understand how they are calculated so you understand their conclusions and possible sources of error that could cause the actual results to vary significantly from their predictions.

BNM

Well Nate and his team better take another look at the numbers

 

[Boob-No-More]

Well Nate and his team better take another look at the numbers

The first thing they need to do is update their database so the rosters are current. We signed Evan Turner 3 and a half months ago, and they still have him on the Celtics roster. And, that's only one of dozens of roster errors I spotted.

BNM

 

[Strenuus]

Can it just be Tuesday so we can get to the regular season? We're so close!

 

[Wizard Mentor]

The first thing they need to do is update their database so the rosters are current. We signed Evan Turner 3 and a half months ago, and they still have him on the Celtics roster. And, that's only one of dozens of roster errors I spotted.

BNM

He's also saying that, in essence, CJ's year last year was basically a fluke, and that he will "regress to the mean".
Last year: CJ = rotation play (which was wrong)
This year: CJ = average starter (we'll see)

I guess I just don't agree, but we'll see.

http://projects.fivethirtyeight.com/carmelo/cj-mccollum/

 

[Boob-No-More]

Well, after last night's win over UTA, we're still projected to be 6th in the West, but our chances of making the playoffs jumped from 72% to 75% and our CARMELO rating moved up 8 points to 1574. UTA had a corresponding 8 point drop, but we still trail their 1590 CARMELO rating by 16 points. After just one game, we cut their CARMELO lead in half.

The big news was, as a result of SAS's blowout win over GSW, the Warriors saw their chances of winning the NBA championship drop from 55% to 38%. No mater how convincing SAS's win, that's a huge change in odds for one game in late October.

BNM

 

[Natebishop3]

Have people forgotten that OKC missed the playoffs two years ago without Durant? Why is this team going to be THAT much better without him now?

 

[BlazerDuckSeahawkFan94]

Have people forgotten that OKC missed the playoffs two years ago without Durant? Why is this team going to be THAT much better without him now?

They aren't better than us but their squad is pretty legit, especially defensively. They're a team you'd like to avoid in the playoffs, I put them on a Grizzlies tier.

 

[Pinwheel1]

Have people forgotten that OKC missed the playoffs two years ago without Durant? Why is this team going to be THAT much better without him now?

They started out 4-12. Remember Westbrook was also injured during that stretch. So if he is healthy this year plus Adams, Kanter Oladipo....and of course the great Sabonis Jr.......

 

[Natebishop3]

They started out 4-12. Remember Westbrook was also injured during that stretch. So if he is healthy this year plus Adams, Kanter Oladipo....and of course the great Sabonis Jr.......

I don't think it's really that much different... definitely not better than us anyway.

 

[Pinwheel1]

I don't think it's really that much different... definitely not better than us anyway.

No I don't think they are better than us. But with a healthy Westbrook that year...they would have made the playoffs. And right now he is healthy. So they are better than they were then.

 

[Natebishop3]

No I don't think they are better than us. But with a healthy Westbrook that year...they would have made the playoffs. And right now he is healthy. So they are better than they were then.

So the Thunder went 40 and 27 in the games that he played in, so they won 59.7% of their games with Westbrook in the lineup.

They won 45 games overall. He missed 15 games. 59.7% of 15 is 8.95 so roughly 9 more wins with Westbrook, but they had already won 5 of those 15, so I'd give them another 4 wins with Westbrook. That would put them at 49 wins that season. That would be good enough for 8th in the west that year, and then they would have been smushed by the Warriors.

 

[Wizard Mentor]

Well, after last night's win over UTA, we're still projected to be 6th in the West, but our chances of making the playoffs jumped from 72% to 75% and our CARMELO rating moved up 8 points to 1574. UTA had a corresponding 8 point drop, but we still trail their 1590 CARMELO rating by 16 points. After just one game, we cut their CARMELO lead in half.

The big news was, as a result of SAS's blowout win over GSW, the Warriors saw their chances of winning the NBA championship drop from 55% to 38%. No mater how convincing SAS's win, that's a huge change in odds for one game in late October.

BNM

One reason we might be lower in the CARMELO rankings than we think/hope/feel we should be:

Travel. Teams are penalized based on the distance they travel from their previous game. For a long leg such as Boston to Los Angeles, the traveling team loses about 16 CARM-Elo points, and its odds of winning are cut by roughly 2 percentage points. These travel penalties are calculated linearly, so a 2,000-mile travel leg is twice as bad as a 1,000-mile travel leg.

We log the most miles of any team, if I remember correctly. Getting a team back in Seattle would help this.

 

[Wizard Mentor]

One more quick note:
We had the largest disparity between home and road record in the league, if I heard things right during the broadcast. This could be attributed to our travel miles and youth.

 

[SlyPokerDog]

One more quick note:
We had the largest disparity between home and road record in the league, if I heard things right during the broadcast. This could be attributed to our travel miles and youth.

And being forced to spend hours and hours on a plane with @HCP.