NFL 2015 Week 4 Predictions

NFL 2015 Week 4 Predictions

Amos retains tie for the lead, and ESPN’s FPI flexes an additional win for the week to make it a three-way tie.

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 PredictionESPN’s Football Power Index (FPI) and Nate Silver’s Elo as benchmarks for comparison to Amos.

Week 4 Standings

Amos went 11-5 (69%). A notable pick by Amos this week was Philadelphia over the NY Jets, but missed picks on Baltimore, Miami, Dallas, Cleveland and Detroit.

ESPN’s FPI went 12-4 (75%). A one game week over Amos with the same Philadelphia pick over the NY Jets with an additional Denver pick. Other than that, the same picks as Amos.

Microsoft’s Bing went 11-5 (69%). Bing had a notable pick of Atlanta, where all other models had Dallas, but lost some ground with a NY Jet’s pick. Bing also missed picks on Cleveland, Miami, Baltimore and St. Louis.

Nate Silver’s Elo went 11-5 (69%). Elo had the same picks as Bing, without the Atlanta pick, favoring Dallas instead and gaining a pick with Pittsburgh.

Moving on to Week 4 picks, the models look to differentiate themselves. A handful of deviations are present.

Week 4 Picks

Amos has calculated some interesting picks this week. Amos is currently favoring Washington over Philadelphia (51%) and Houston over Atlanta (51%). Close games; the rest of the models favoring the opponent.

ESPN’s FPI is favoring New Orleans over Dallas (50.4%). Once again, a close game while the rest of the models have picked Dallas.

Microsoft’s Bing is going out on a limb and favoring Oakland over Chicago (63%) while the rest of the models have picked Chicago.

Elo is favoring the NY Jets over Miami (54%) with the rest of the models picking Miami.

Week 4 Overall Graph

Finally, from the performances of each of the teams, Amos has simulated the rest of the season. Big gainers this week are New England and Buffalo, collecting 4 additional wins throughout the remainder of the season. The  biggest loser this week is San Francisco, losing an additional 4 games.

See the results below and each team’s change from the last week.

Week 4 Remainder

Week 1: 11-5

Week 2: 9-7

Week 3: 11-5

Week 5: 10-4

Week 6: 10-4

Week 7: 9-5

Week 8: 10-4

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


NFL 2015 Week 3 Predictions

Week 2 is in the books and Amos had a great week. Sort of.

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 PredictionESPN’s Football Power Index (FPI) and Nate Silver’s Elo as benchmarks for comparison to Amos.

Week 3 Standings

Bing has now fallen behind from Week 1 and Amos and Elo are tied for first, but all performances are low due to a quite the occasion this past week.

Wins by Oakland, Jacksonville and Tampa Bay hurt all models. Surprisingly, all three models predicted Cleveland’s win over Tennessee, but only Amos was able to predict Washington’s win over St. Louis, coupled with a Denver over Kansas City pick, enabling Amos to pass Bing and tie with Elo.

Amos went 9-7 (56%) this week, moving up in the standings with an incredible differential prediction of Washington over St. Louis. Slightly less impressive, Amos also picked Denver over Kansas City. A big week for Amos to get back up with its peers while still a low accuracy week.

FPI went 9-7 (56%) as well, also predicting Denver over Kansas City and New England over Buffalo.

Bing had a terrible week, going 6-10 (38%). Bing’s performance poor performance was driven by misses all over the board with the losses of Chicago, Kansas City, Buffalo, and Indianapolis.

Elo also had relatively rough week, only getting to 8-8 (50%). The performance was driven by its misses of Kansas City and Indianapolis games.

Moving on into Week 3 picks, it looks to be a fairly tame week in the prediction arena. There is a large amount of consensus with only 4 games with differences.

Week 3 Picks

Deviations provide an opportunity for models to pull away from the others, with the risk of falling behind. This week we will see 4. Amos is favoring Detroit over Denver. A bold move as all other models are in agreement over Denver. Similarly, Elo is favoring NY Jets over Philadelphia while all other models a favoring Philadelphia. Bing is also squaring off against the models favoring St. Louis over Pittsburgh and Atlanta over Dallas. Finally, all four models are split between Philadelphia @ NY Jets.

Below outlines the picks by model, but also illustrates confidence differences between models.

Week 3 Overall Graph

Finally, from the performances of each of the teams, Amos has simulated the rest of the season. See the results below and each team’s change from the last week.

Week 3 Remainder

Week 1: 11-5

Week 2: 9-7

Week 4: 9-6

Week 5: 10-4

Week 6: 10-4

Week 7: 9-5

Week 8: 10-4

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


NFL 2015 Week 2 Predictions

Week 1 is in the books and now Amos has some new 2015 data.

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 and ESPN’s Football Power Index (FPI) as benchmarks for comparison to Amos. I will also be including Nate Silver’s Elo this week.

Amos went 11-5, performing well for the first week. Amos hit all of its picks over 70% besides Philadelphia’s loss. Bing went 13-3 outperforming the rest of the models by predicting wins at Atlanta and St. Louis. Elo went 12-4, falling into 2nd place. and ESPN’s FPI went 10-6, falling into 4th and par with home wins.
Week 2 Standings

Moving on into week two picks, there is a lot of consensus among the models. Check out all of the predictions below.

Week 2 Picks

There appear to only be four key games this week for the models.

First, Amos is heavily favoring Washington over St. Louis (60.1%) while all the other models are favoring St. Louis.

Second, Bing is favoring Chicago over Arizona (51.5%) while the rest of the models favor Arizona.

Third, the Denver vs Kansas City match up is split by the models, Amos and ESPN’s FPI favoring Denver while Bing and Elo favor Kansas City.

Fourth, the Buffalo vs New England match up is also split by the models, Amos and Bing favoring Buffalo while ESPN’s FPI and Elo are favoring New England.

Week 2 Overall Graph

Finally, from the performances of each of the teams, Amos has simulated the rest of the season. See the results below and each team’s change from the last week.

Week 2 Remainder

Week 1: 11-5

Week 3: 11-5

Week 4: 9-6

Week 5: 10-4

Week 6: 10-4

Week 7: 9-5

Week 8: 10-4

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


NFL 2015 Week 1 Predictions

With the first NFL game of the 2015 season less than 48 hours away, here are the Week 1 game predictions.

For those just joining, Amos is a statistical model created to predict the outcomes of each of 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 and ESPN’s Football Power Index (FPI) as benchmarks for comparison to Amos. Microsoft’s Bing Prediction was previously Microsoft’s Cortana, which was compared to the Bischoff model last NFL season. ESPN’s FPI is new to these comparisons.

First, before the predictions are displayed, it is important to note that predicting the first week within the NFL is a wildcard as only scarce or previous season’s data exists to feed into the model.

2015 Week 1

A few observations between the models are the confidence varies between them. The highest confidence of Amos is Green Bay over Chicago (94.9%); Bing is also Green Bay over Chicago (71.6%); the FPI is New England over Pittsburgh (67.2%). Amos appears to use the most confidence with its projections, however this also could stem from Amos’ heavy reliance on previous season data to power its model in Week 1.

There are also a lot of agreements among the models, including wins of New England, Green Bay, Denver and Dallas. Most of these are to be expected and should not come as a surprise. However, there are key differences within Week 1.

Deviating from Bing and the FPI, Amos is favoring Cleveland over the NY Jets (64.3%), Buffalo over Indianapolis (58.4%), Detroit over San Diego (65.6%) and Kansas City over Houston (50.0%). Specifically, Kansas City is being favored by just a hair, (approximately less than .02%). Bing deviates from Amos and the FPI with a prediction of Atlanta over Philadelphia (52.0%). The FPI does not disagree with both Amos and Bing in any one specific prediction.

What’s important to note about deviations from the two other models is that they indicate an opportunity for the deviating model to pull away from the other two models as more predictions are correct. Adversely, as deviating predictions do not come to fruition, the deviating model can lag behind the other two models.

Finally, Amos has simulated the entire regular season on available preliminary data. These predictions are expected to change as we move through the season and Amos interprets game outcomes week by week on team’s ending record.

Week 1 Records

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 8: 10-4

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


2014 NFL Season Review & 2015 Updates

Football season is upon us. September 10th at 8:30 PM EST New England and Pittsburgh will open the NFL regular season. This season I will continue to predict NFL outcomes, which you can find from this site every week. However, as we enter into the 2015 football season, I thought it appropriate to review last season and then introduce some changes that I’ve made.

Last season I created a model that only took into account three variables which included which team’s offense (1) and defense (2) was better, and also considered home field advantage (3). To be clear, the model was intended to be just a simple experiment. That model was used to predict each regular season game and then compared to two other models, Nate Silver’s Elo and Microsoft Bing’s Cortana, and then went on to compete in the post season. Let’s take a look at how it fared within the regular season.

Ending Record

Nate Silver’s Elo came out on top with a 176-80 (69%) record. The Bischoff model came in 2nd place out of the three with a 172-84 (67%) record. And finally, Cortana came in third with a 169-87 (66%) record.

Breaking down the numbers further, the following graph shows how many correct predictions were made each week, by model.

By Week

Elo was the most consistent model, having its lowest performing percentage be 56% in weeks 2 and 13. The Bischoff model had a little more volatility, which had its lowest performing percentage be 44% in weeks 2 and 16. Another notable week was number 5, where the Bischoff model performed at 93%. This was the highest any of the three models. Cortana had the lowest performance, having its lowest performing percentage be 38% in week 4. It’s important to note weeks 8-11, where all three models had almost identical performances before falling into more volatility in the last stretch of the season.

In 2014, there was no clear indication of which model captured the season better, besides Elo’s four predictions over the Bischoff model. I am excited to see what the 2015 season brings.

For the 2015 season, I’d like to introduce my new model: Amos. Amos now takes into account more variables than the original, expanding to eight different variables. Amos also incorporates a percentage associated with each win. This should help communicate how strong the model predicts the win to be. Therefore, the higher the percentage indicated by Amos, the greater the probability the predicted team has of winning.

As any sports analyst will tell you, underdogs always have a chance of winning. This is captured by the remainder (the compliment) of the percentage. For example, if Amos predicts Green Bay has a 73% chance of winning its week 1 game against Chicago, then Chicago has a 27% chance of winning.

Additionally, I will compare Amos’ predictions (and results) beside other models where I am able to. As of now, I have not heard anything on whether Elo will be coming back or if Cortana will be making a return either, however I will try to identify other contenders to gauge the performance of Amos. What fun would this be if there was no competition?

Finally, and probably the biggest addition to the 2015 season, Amos will be simulating the remainder of the regular season each week, and predicting every team’s ending records. The first 2 weeks will be highly volatile in outcomes as more data is fed into the model and simulation.

We can expect high volatility at the beginning of the season, but should begin to see Amos predict better throughout the season. Week to week, we should also be able to tell how Amos interprets a team’s performance on a team’s ending record.

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 8: 10-4

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


NFL Picks Superbowl

All three models correctly picked the Conference Championship winners as New England and Seattle get set to play in the 49th Superbowl. However there is less parity between the models this week on who will be taking home the trophy on February 1st.

For those just joining, the Bischoff model, Microsoft’s Cortana, and Nate Silver’s Elo are all statistical models created for predicting NFL games. During the regular season, Elo led the models with correctly predicting 176 games, Bischoff came in 2nd with 172 games, and Cortana predicted 170.

Both teams have had a fantastic season, both going 12-4. The composition of those wins are very different however.

New England racking up 468 points on their opponents while Seattle didn’t even break the 400 mark with 394, a differential of 74 points. While impressive, New England only sits in 4th place in most points – trailing behind Green Bay Denver and Philadelphia.

On the defensive side, New England let opponents score 313 points on them, placing them 8th in the league. Seattle on the hand only allowed their opponents to score 254 points during the season, wildly boasting 1st place, with 2nd place being 27 points away, pushing the defense point differential between the two teams to 59.

Superbowl

Once again, both teams accomplished a lot this season and the Superbowl is promising to be quite the show. Let’s see what the models have picked:

Bischoff: Seattle

Elo: Seattle

Cortana: New England


NFL Week 20 Picks

Due to the all three models having identical predictions on Week 19, all the models came out even at predicting 3 out of 4 games correctly. The only upset being Indianapolis’ fierce performance against Denver. It is also important to note Green Bay’s win over Dallas was a rocky one, with split views of the official’s call of an incomplete catch of Dez Bryant late in the game. Nonetheless, it helped the models predictions ratings.

For those just joining, the Bischoff model, Microsoft’s Cortana, and Nate Silver’s Elo are all statistical models created for predicting NFL games. At the regular season, Elo led the models with correctly predicting 176 games, Bischoff came in 2nd with 172 games, and Cortana predicted 170.

Here is the full view of last week’s performances:

Week 19 Outcomes

 

Moving on into the Conference Championships, the NFC Championship will feature Green Bay at Seattle. For the AFC Championship, New England will host Indianapolis.

Just like last week, each of the three models have picked the same outcomes. It comes as a little bit of a surprise that none of the models have picked up Indianapolis at all, however that is probably due to the inconsistent regular season they had.

Here are the picks:

Week 20 Picks


NFL Week 19 Picks

The Bischoff model had a rough starting week for the playoffs, predicting only half of the games with the model picking up Arizona’s past success in the season and the upset at Pittsburgh.. The two peer models, Microsoft’s Cortana and Nate Silver’s Elo, both had the same predictions and lost the Baltimore upset over Pittsburgh as well.

Playoff standings are:

3 – 1 (Elo)

3 – 1 (Cortana)

2 – 2 (Bischoff)

However excitement between the models continues to be low as this week all three models pick the same winners and all are home teams. Here is the breakdown:

 

Week 19 Predictions


NFL Week 18 Picks

The Bischoff Model ended with great a season, correctly picking 172 games for the 2014 year. Comparing itself for its main peer, Nate Silver’s Elo, it was just 4 games behind with Elo predicting 176 correct games. Microsoft’s Cortana lagged a little behind with only correctly predicting 170 correct games. To Cortana’s credit, it did a great job the last 3 weeks, tying the Bischoff model in Week 15 and leading the pack the Week 16 as well as Week 17.

Here are the final outcomes of the regular season.

Week 17 Outcomes

I was not planning on going into the playoffs with predictions, however due to everyone’s interest (not that it took much convincing), I figured I’d continue and see how the Bischoff model stood up to the other two models.

Microsoft’s Cortana and Nate Silver’s Elo agree on all of the games, however the Bischoff model disagrees on the Arizona at Carolina, which is probably due to Arizona’s superb success at the beginning of the year that the Bischoff model is still picking up.

Here are the full predictions.

Week 18 Predictions


NFL Week 17 Picks

The Bischoff model came out of Week 16 with its worst week of predictions since Week 2, predicting just 7 out of 16 games (43.8%), nearly half from the previous week where it predicted 13 out of 16 games (81.3%).

Week 16 OutcomesNonetheless, the Bischoff model still has continued to keep up with its peers. Nate Silver’s Elo came out with 9 correct and Microsoft’s Cortana, on top, boasted 10 correct predictions. Marking only the 2nd week that Microsoft out predicted both models.

 

Getting an overall look leading into the final week, Elo leads the pack with the Bischoff model following by 3 games while Cortana lags behind.

Season To-Date Picks

However, given the predictions for this Sunday, the Bischoff model poses a threat to tie Elo. Bischoff and Elo disagree on 3 games which allows a variety of outcome possibilities to finish the 2014 regular season. Cortana only disagrees with Elo on 1 game, which Bischoff also agrees with Elo on.

The three games deciding the fate of the Elo and the Bischoff model are:

San Diego at Kansas City

Philadelphia at NY Giants

Cincinnati at Pittsburgh

Below are the full predictions.

Week 17 Predictions


What People Are Saying