NFL: Week 2 model projections from Jonathan Von Tobel


NFL Quick Links:
Super Bowl | Betting Splits | Odds | Picks | Matchups


Week 2 NFL Projections

This is part of a recurring series on in which VSiN host Jonathan Von Tobel tracks the progress of a model he created for the NFL season

Week 1 Best Bets: 4-1 ATS | Week 1 Overall: 12-4 ATS | Season Overall: 12-4 ATS

The first week of Jon Von Model’s life could not have gone better.

First, the five biggest edges that the model had – Jacksonville, Miami, Green Bay, Philadelphia and Buffalo – ended up going 4-1 ATS with the lone loss being the Bills on Monday night. Second, if a bettor decided to trust the entire model and play every edge then they would have gone 12-4 ATS in the first week of the season.

It is fantastic that this experiment is off to the start it is, but it is obviously only one week of results. The real work begins this week.

I am now faced with quite a few challenges on my end. How do I incorporate the data from the first week? Do I build up projections based off of just one week of data, or do I continue to work off of the model projections that were used for the first week? My answer is somewhere in the middle.

My inspiration for the model this week came from Billy Walters, via Eric on X who said that Walters recommended in his book to keep one’s ratings at 90% old and 10% new. So what I decided to do was to then adjust my numbers on certain quarterbacks that I did not have true data on coming into the season.

Teams that received adjustments were Tampa Bay, Green Bay, Las Vegas and Atlanta, as well as the teams which started rookie quarterbacks. Other teams with returning starters I kept the same for this week, and will update their numbers as the sample-size of data grows for their respective signal-callers.

This also gives me time to find a suitable replacement for the Football Outsiders metrics that have disappeared – something we discussed last week – and that could be beneficial, as Aaron Schatz did announce the database at FTN is being worked on to incorporate “everything” that was up on Football Outsiders. Hopefully that means access to the drive success statistics they had in the past.

If I cannot get access to that drive success rate then I have decided to use the success rate statistic offered at RBSDM as a substitute despite the lower correlation when I run my regression testing.

With that, let’s move on to next week.

Week 2 Projections

After running the matchups for Week 2 I am very torn on how these projections came out. Eight games are within two points of the betting market and if we expand the parameters to three points then 11 games are in line with the betting market. When a majority of games are in synch with the betting market it gives me some confidence that the projections are on the right path.

Having said that, the five games which have the biggest edges have some massive ones.

For example, Jon Von Model projects a massive 10.89-point victory for Tampa Bay this week over Chicago. If you were to tell me that there was an edge on the Buccaneers this weekend I could get behind it.

I believed the Bears to be the most overvalued team in the league when we were heading into the regular season, and the result on Sunday reinforced that. However, if you were to tell me you had a 7.89-point edge on the Buccaneers in that game I would tell you that you might have to make a small adjustment in your calculations.

The problem is, that no matter what adjustments I made with what statistics I had at my disposal the margin only got wider between the two teams. In other words, Jon Von Model hates the Chicago Bears, and I cannot blame it.

The other massive edge that I do not like is what Jon Von Model is projecting for Monday night between Cleveland and Pittsburgh. Not only does JVM have the Steelers favored, it has them favored by 4.29 points which equates to a 6.79-point edge. Once again, an edge that is over a touchdown, and something that tells me there might be an issue somewhere.

So, this leaves me conflicted.

Part of me wants to continue to toil away and find a way to adjust these numbers so that the larger edges are more in line with the betting market. The other part of me thinks that this model was successful in the first week, so why not believe in it for one more week and see what happens?

We will be going with the latter approach for this week. That means that the five biggest edges for the SuperContest entry that Mitch Moss and I have will be as follows:

Buccaneers (-3)
Steelers (+2.5)
Chiefs (-3)
Cowboys (-9.5)
Bengals (-3.5)

Before we wrap it up, I do want to point out two more projections that Jon Von Model has that I whole-heartedly agree with.

Green Bay was not as impressive as the final score indicated against Chicago, but defensively they looked to be an above average group. They should match up very well against Desmond Ridder and Atlanta. I did not like what Ridder did at quarterback in Week 1 and bet the Packers immediately on Sunday. That is one where my personal thoughts line up with Jon Von Model.

The same could be said of Carolina at home against New Orleans. Outside of two terrible interceptions, Bryce Young played better than what his final line would indicate. The Saints also did not look like a team which should be laying a full field goal on the road against a divisional foe. JVM makes the game a near pick, so there is value in getting a key number.

With that, we sign off until next week. I obviously hope this week goes as well as Week 1 but I cannot help but feel an uneasy feeling of dread heading into the card.