Jump to content

Why do people take 538 NFL predictions seriously?


BITE ME

Recommended Posts

2 hours ago, BITE ME said:

Is it because it is Disney/ESPN owned?  Putting aside how terrible they are at picking elections, the preseason predictions for the AFC East are beyond bad.  What is even crazier is they had Dallas and the Eagles with better records than the Bills and Washington as the worst team in the NFL.

 

Patriots  9-7

Bills  8-8

Jets 7-9

Dolphins 6-10

 

https://projects.fivethirtyeight.com/2020-nfl-predictions/?ex_cid=rrpromo


Fivethirtyeight has been exceptionally good at picking presidential and congressional elections.  Statistically I don’t think any other group has been better over the last couple decades.  Eight years ago they called every state correctly for the presidential, senatorial and gubernatorial elections except one senate seat.  I guess that’s why people take their predictions seriously. 

Link to comment
Share on other sites

2 hours ago, BuffaloBobs said:

 

It still doesn't explain how they could only predict two teams above ten wins. That's historically completely unpredicted and suggests a major problem with their modeling.

At the start of the season they do some analytical work to have all teams "regress to the mean". I.e. the very worst teams are more likely than not to get better and the very best teams are more likely than not to get worse. So in aggregate all the records get drawn towards 8-8 to start. Regressing to the mean is not a statement on any one particular team, it's a blanket statement on how statistics work in a population. Also, when 538 makes those predictions, aside from the QB, they literally are only looking at game scores from the past. The more recent the game, the more weight it has. But in the start of the season the most recent games were from last season.  This approach is somewhat boring in a sense though...

 

Aside from Allen being the #1 rated QB on their site (all from statistics, not a judgement call), if you look at the historical ELO of every team it says that this Bills team is currently the best team in franchise history. The prior highest ELO was right after the 51-3 smackdown of the Raiders, prior to the first SB loss.  

Link to comment
Share on other sites

It's because their predictions, are in fact, accurate on a large scale. I actually emailed them for historical data about two years ago and its uncanny. Games where they predict the winner of the match is very accurate given a large sample size.  I broke it out into ranges of 51-55,56-60, 61-65, 66-70, etc. All of those hold true. Meaning if a team they predicted had a 68% chance of winning it would fall in my range of 66-70% chance. 

 

Long story short the whole point was to find an edge when books set moneylines wrong. You just need a HUGE bank roll to ensure it works. 

  • Like (+1) 1
Link to comment
Share on other sites

On 1/5/2021 at 4:43 PM, Reader said:

Yes...let us bash the website that lists Allen as the best QB in the league.

 

538 QB rankings

 

Even more ridiiculous.  Josh Allen is not the best QB in the NFL.  Any serious Bills fan knows this.  Mahomes and Rogers are clearly better quarterbacks.  It is fair to say he is the 3rd best quarterback in the NFL and even that is debatable.  Mahomes is clearly the best QB in the NFL.  It really is not debatable.

Link to comment
Share on other sites

41 minutes ago, BITE ME said:

 

Even more ridiiculous.  Josh Allen is not the best QB in the NFL.  Any serious Bills fan knows this.  Mahomes and Rogers are clearly better quarterbacks.  It is fair to say he is the 3rd best quarterback in the NFL and even that is debatable.  Mahomes is clearly the best QB in the NFL.  It really is not debatable.

 

Without researching it, I suspect that their QB ELO rankings have a large weighting toward recent games.  Josh definitielly had a more productive Dec/Jan.

Link to comment
Share on other sites

×
×
  • Create New...