NFL: Week 6 model projections from Jonathan Von Tobel

Week 6 NFL model projections and best bets

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

Week 5 Best Bets: 4-1 | Overall Best Bets 16-9
Week 5 Overall: 10-3-1 | Overall Record: 47-29-2 

Week 5 Recap

It was another successful week for Jon Von Model despite an inauspicious start to the card on Thursday night. The Commanders – who were projected to win by 14.81 points – lost by 20 points to the Bears to put the JVM SuperContest card in an 0-1 hole. Heading into Sunday the best bets were on a 1-5 slide, but all that changed with a 4-0 day to bring the card to 4-1 for the week.

That record in best bets was representative of the week overall for the model, which was 10-3-1 in all edges this week to improve to 47-29-2 ATS (61.8%) for the season. The contest entry that I share with Mitch Moss is now 16-9 and four points out of first place.

All of this is obviously fantastic. This experiment is doing better than I could have hoped, and I believe it is successful because the projections have been in lockstep with the market almost this entire season. That could lead to some volatile results – such as the 1-4 record in Week 4 best bets – but as long the projections stay in line with the market we should be about .500 or better for the season.

Before we move on though, I did want to expand on the Thursday night projection.

With the result in hand it does seem ridiculous to have Washington projected to win by such a margin, but now that I have had time to ruminate about the projection I have come around on standing firm with it.

Jon Von Model is now 3-1-1 in predicting the results of Chicago’s contests this season. It had spit out the Bears’ opponents as best bets the first three weeks of the season, and it had Denver as a 3.6-point winner in Week 4 where the line closed at a field goal. The model had a very good read on the Bears going into Thursday night, and it felt wrong to go against that success.

That game is also a small example of what I am dealing with as someone who has constructed this model. The success this model has had is built on small sample size, but the sample size is growing and the success is maintaining itself. While I believe this will even out by the time the season ends – 61.8% for all edges is not sustainable – I am at least at the point where I will not question what the model projects as the biggest edges.

Week 6 Adjustments

The concept of making adjustments on a model that is doing so well seems counterintuitive, but it has to be done. The original projections for this week were actually very similar to the betting market, but there were some – like New Orleans being a 9.31-point favorite – that warranted some changes.

One the biggest adjustments that needed to be made were to New England. JVM is not built on many defensive statistics, but the one which it weighs heavily – Adjusted Line Yards from FTN – seemed to really inflate the value of the Patriots. After updating the metrics on defense I made sure to go through and modify each of New England’s offensive metrics to account for the poor start this season. The one update which swung the Patriots’ rating back to where it should be was to EPA (RBSDM) and net EPA (NFElo). Once those were adjusted to represent this season’s output for New England we got a line that mirrored the betting market.

New Orleans was a similar case, but the adjustments that needed to be made were with Houston. The model – still using some metrics from last season – needed to be told how much better the Texans are this season. Updates to CJ Stroud’s output, as well as the other offensive metrics brought this number down a full six points. The Saints are still going to be included on this week’s card, but the margin is much tighter after JVM was updated.

The other two adjustments that are worth nothing this week are to Indianapolis and Dallas.

For the Colts there were many adjustments that needed to be made. First, Gardner Minshew is now the starting quarterback with Anthony Richardson now on injured reserve. There might be some who believe that Minshew is an upgrade over Richardson, and while that is statistically true the difference is less than a point, according to Jon Von Model.

Indianapolis also needed an upgrade as a team. The Colts are now 3-2 ATS and better than the market believed them to be. JVM’s initial projections had Indianapolis catching 11.32 points, but adding Minshew and updating the team metrics brought this line down to 2.51 points.

Dallas needed to be upgraded as well. The initial projections had the Cowboys winning by over five points, but updates to Dak Prescott as well as the team’s offensive metrics completely flipped the line and actually made the Chargers the biggest edge of the week.

Week 6 Projections

The first thing that I believe is worth discussing is how tight the margins are getting with Jon Von Model’s projections. As previously mentioned, the Chargers are the biggest edge this week and it is only a 2.17-point edge. The five best bets as of Wednesday morning have edges of 2.01 points or fewer. If you expand your scope to factor in the whole card there are seven games within a point of the current betting market and 13 games that are within two points.

This is a sign that the model is doing well with its projections, but it also shows just how tight these margins are in the NFL. It is great that JVM is on the same path as the market, but it is not great when we have to make five plays every week.

That being said, the contest card as of Wednesday looks like this for Mitch and I in the SuperContest:

Chargers (+2.17)
Saints (+2.01)
Lions (+1.64)
Falcons (+1.56)
Rams (+1.54)

The numbers in parentheses are what the edge is according to the current betting market. I included them to give you a sense of what the edges are, and how tight the margins are getting.

I do want to point something else out as well: This is the first week that the Chiefs are not included in the best bets.

Jon Von Model has been higher on Kansas City than the betting market has been, and it has paid off. In the four weeks we used the Chiefs on the contest card we went 3-1 when playing them. The betting market has finally caught up to the model’s rating on Kansas City it seems, as the projection for this week only gave the Chiefs a 0.54-point edge.

The growing fear on my part was that we were going to be on Kansas City every week, but that fear gets pushed aside for a week at least. It is also a great example for bettors to stick to their guns if they have an edge. JVM clearly found one in the Chiefs and it paid off. Sticking to an angle is something I have trouble with, and this process has helped me see the value in it.

With that, we’re off for another week of sweating out Jon Von Model’s projections. Obviously the hope is that the success continues, but you will hear me preach caution every week in this column until the season comes to an end.