Really appreciate the discussion on spread and ELO rating.
A few comments: I was originally intending to do spread picking as an additional contest on my own but the league wanted to change it up for this and so I was like ok sure!
I also agree that week 1 ELO is often not accurate, that is the nature of ELO. As kckolbe said week 15 it is amazing. Kckolbe also wanted a week by week comparison. Well, I dont have that exactly, but I discussed before, here are some details:
The only time ELO was not accurate as a whole was week 1 to week 4 of season 1, which makes perfect sense, there was no history. From week 5 to the end of season one, it was already 78% correct picks. And in season 2 is the same thing, 78.5% correct predictions. It was tied for 7th in the league in predictions in season 2. 7th out of 50 or so people. So as a whole it was much more accurate than the average user. And there was no true way to compare weeks. The only games it really got wrong were games that were close and could go either way, or complete shockers like the Yeti upset on the road.
As far as I saw, there was no real pattern to the correct predictions in season 2. It was just as accurate early as late, at least at first glance. Maybe a further analysis would show it being slightly more accurate later in the season.
Another thing to say is though, there was a lot more free agent and trade changes from season 2 to season 3, much more than from season 1 to season 2 I think. Maybe that will be an interesting thing to look at as well.
A few comments: I was originally intending to do spread picking as an additional contest on my own but the league wanted to change it up for this and so I was like ok sure!
I also agree that week 1 ELO is often not accurate, that is the nature of ELO. As kckolbe said week 15 it is amazing. Kckolbe also wanted a week by week comparison. Well, I dont have that exactly, but I discussed before, here are some details:
The only time ELO was not accurate as a whole was week 1 to week 4 of season 1, which makes perfect sense, there was no history. From week 5 to the end of season one, it was already 78% correct picks. And in season 2 is the same thing, 78.5% correct predictions. It was tied for 7th in the league in predictions in season 2. 7th out of 50 or so people. So as a whole it was much more accurate than the average user. And there was no true way to compare weeks. The only games it really got wrong were games that were close and could go either way, or complete shockers like the Yeti upset on the road.
As far as I saw, there was no real pattern to the correct predictions in season 2. It was just as accurate early as late, at least at first glance. Maybe a further analysis would show it being slightly more accurate later in the season.
Another thing to say is though, there was a lot more free agent and trade changes from season 2 to season 3, much more than from season 1 to season 2 I think. Maybe that will be an interesting thing to look at as well.
![[Image: card6.png]](https://cdn.discordapp.com/attachments/692421772183601193/1053345472418431016/card6.png)
