NFL 2015 Week 8 Predictions
Amos and Nate Silver’s Elo tie for the week, and Bing regains some lost ground.
For those just joining, Amos is a statistical model created to predict the outcomes of each NFL game. Amos currently predicts games straight-up. Additionally, Amos then predicts the remainder of the season to compute the ending record of each team.
Amos also has peers within the NFL prediction field. While there are a number of sources for game predictions, I have identified Microsoft’s Bing Prediction (Cortana), ESPN’s Football Power Index (FPI) and Nate Silver’s Elo as benchmarks for comparison to Amos.
Amos went 9-5, having misses of Buffalo, Indianapolis, Pittsburgh, San Diego and Dallas.
ESPN’s FPI went 8-6, having misses of Buffalo, Tennessee, Indianapolis, Detroit, San Diego and Dallas.
Microsoft’s Bing went 10-4, which is only its second time finishing a week above the other models, having misses of Buffalo, Indianapolis, Pittsburgh and San Diego.
Nate Silver’s Elo went 9-5, having misses of Buffalo, Indianapolis, New Orleans and San Diego. Unfortunately, Elo did not have a pick for the Minnesota and Detroit game which will be counted against it.
Moving onto Week 8 picks, ESPN’s FPI is calling for some upsets. However, the main toss up is Cincinnati vs Pittsburgh this week.
Amos’ picks this week are no surprises except Amos continues to have faith and is favoring Dallas by less than 1% (actually 50.008%).
ESPN FPI is predicting some upsets this week with favoring Detroit over Kansas City (53%), Chicago over Minnesota (55%), and Indianapolis over Carolina (56%). FPI is also on the fence over Dallas vs Seattle, but is leaning slightly to Seattle.
Microsoft’s Bing has two deviations from the other models which are Cincinnati over Pittsburgh (52%) and Oakland over NY Jets (53%).
Nate Silver’s Elo has the same predictions as Amos and Bing, however Elo, much like Amos, has predicted right at 50% for the Cincinnati at Pittsburgh game. Elo has taken the Pittsburgh pick.
Finally, from the performances of each of the teams, Amos has simulated the rest of the season. The big gainer this week is Kansas City, who was able to pick up 2 more wins for the season. The biggest losers this week are Dallas and San Diego, which should not be surprising after the loses this week.
See the results below and each team’s change from the last week.
Week 1: 11-5
Week 2: 9-7
Week 3: 11-5
Week 4: 9-6
Week 5: 10-4
Week 6: 10-4
Week 7: 9-5
Week 9: 7-6
HALFTIME REPORT & Week 10: 5-9
Week 11: 10-4
Week 12: 9-7
Week 13: 11-5
Week 14: 9-7
Week 15: 14-2
Week 16: 7-9
Week 17: 9-7
Wildcard: 3-1
Divisional: 2-2
Conference: Predictions
Update: A reader pointed out that Nate Silver’s Elo does indeed have a pick for the Cincinnati at Pittsburgh game. This has been added and updated in all charts and graphs.
Update: Bing changed its pick from NY Jets to Oakland. This has been added and updated in all charts and graphs.
8 Comments
Nate
October 27, 2015 at 10:46 pmELO did have a pick for the Minnesota-Detroit. If you check their picks against the spread for last week, you’ll see they had the game as a pick ’em but had Detroit selected. So they would have lost anyways, but they did have a pick.
L Daniels
October 27, 2015 at 11:02 pmI hadn’t read your message yet when I posted my own. I do apologize for any redundancy.
Best,
-L
Nate
October 27, 2015 at 11:19 pmAdditionally, this week, they have Pit as the Pick ‘Em choice for the ATS picks so I would give them the edge for the sake of your comparison.
TrevorBischoff
October 28, 2015 at 8:48 amThanks for taking the time to point this out. I’ve added Pittsburgh as Elo’s pick for this week.
I will also go back and review the previous weeks to make sure Elo’s standings are accurate.
Trevor L. Bischoff
L. Daniels
October 27, 2015 at 11:00 pmHello,
I just wanted to comment that I believe your categorization of an ELO designation of even as an incorrect selection is a biased application of statistical comparison.
Your Amos rating is given the benefit of the doubt when it chooses the Cowboys by 0.000008. ELO is not given the benefit of the doubt.
Under this comparison system, if we were predicting coin tosses, ELO would call a pick ’em on every toss. Your analysis wold mark every call a loss, and they would get zero correct.
There are several ways to remedy this:
1. Even when ELO announces a pick’em, they are favoring one team fractionally above the other. You can determine this by looking at the ELO interactive graphic. By clicking on point spread, it lists which teams are favored and by how many points. When a team is favored by less than 0.5 points, they list PK rather than a number. However, you can determine which team is marginally favored by looking at which team the PK designation is next to. For example, next week, ELO favors Pittsburgh, as can be determined from 538’s website. It should be noted, to be fair to you, that ELO incorrectly picked Detroit last week and would have lost anyway. If you are going to award yourself eligibility for a correct prediction for a 50.008% confidence pick, the most fair comparison would allow Nate Silver the same opportunity.
2.You could award him a half win.
3. You could leave the pick out of the analysis altogether.
4. The truest evaluation of the prediction accuracy over a season would not evaluate wins and losses in a binary manner. It would add up the percent chance of victory that each prediction service gave to the winning team. In this method, if you gave team A a 80 percent chance of winning, you would accumulate .8 for a Team A victory, and .2 for a Team A loss. This would eliminate your dilemma of determining pick’em predictions, as it would simply add .5, no matter who wins. I understand this method might not fall within the purview of your analysis, put I felt it was worth noting.
The easiest fair solution is option 1. It would take 30 seconds of research to determine who 538 favors when they report 50-50 odds. This would put their evaluation in line with your own.
And thank you for your site. I value it greatly in the comparison of NFL predictive models. That is why I feel it is important that ELO gets the same playing field, so as to show a true comparison of their accuracy.
Thank you for taking the time to create your site, and to read my message. The stats nerd in me loves pouring over your tables each week.
Best,
L
TrevorBischoff
October 28, 2015 at 8:47 amThanks for taking the time to offer your suggestions and keep up with my blog. I think these are all great ideas.
I wasn’t aware that Elo did in fact have a pick when the teams were at a 50-50 prediction. I wasn’t aware that the PK designation was actually toward one or the other team. This is great news. I’ve added the pick for this week and I will also look over the previous weeks to ensure Elo is on equal grounds with the rest of the models.
Thanks again for taking the time to keep up with my blog.
Trevor L. Bischoff
Gluta Panacea
January 7, 2016 at 6:17 pmAwesome write-up. I’m a regular visitor of your blog and appreciate you taking the time to maintain the nice site. I will be a frequent visitor for a long time.
wooden greenhouses
February 8, 2016 at 2:44 amMay I simply say what a comfort to discover someone who genuinely knows what they are talking about over the internet. You actually know how to bring a problem to light and make it important. More and more people should look at this and understand this side of the story. I can’t believe you aren’t more popular since you most certainly possess the gift.