We bet on victory.

And if I am the one placing the bets, I need two key factors to have the confidence to place the bet:

1. A model-based approach, or at least a quantitative approach. But almost all of the bets in this column are based on the results of models built by me or, occasionally, my colleagues at ESPN Analytics.

2. A less efficient market. NFL teams and totals are high-volume markets, and betting against those mainstream numbers is like betting on All-Madden. That’s why I often look at props (especially on defense!) and alternate lines. Less attention means less efficient markets, and therefore more opportunities to find value.

These two criteria form the core of this weekly column. Each week I will publish bets from our models in categories that are often less visible, with the simple aim of coming out on top. We will look at odds from the whole spectrum, from -1000 to 100:1 – for me, value is value, no matter the price.

I promise you one thing: we will follow the results – by betting category – no matter what happens. That’s kind of scary, because of course a loss is entirely possible. But I also think it’s really important. And that’s exactly what I did a year ago in my similar column, “Last Minute Bets.”

Last season, it finished with a 6.2% ROI on 466 bets. I can’t guarantee success based on that alone – this year will be more difficult, especially if sack props (my most profitable category) aren’t available, as they currently aren’t in Week 1 – but we’ll post our performance whether we win or lose.

To stay on the topic of transparency, here are the ground rules I set:

1. For the sake of simplicity, I place a flat rate of one unit per bet, regardless of the odds.

2. The story is submitted on Thursday and published shortly after. I review the lines right before submission and then consider them locked regardless of any movement between then and kickoff, in any direction.

3. I can add bets later, so if a bet I’m showing value for isn’t offered on Thursday but is added on Friday, I can submit that bet and it will be published shortly after with a note indicating which day it was added. The same odds movement rules apply.

4. I treat bets without action as pushes.

5. All odds are provided by ESPN BET. Of course, I advise you to always find the best odds possible.

Now that we’ve got all the organizational stuff out of the way, let’s talk football. It’s time for some bets for Week 1.

Alternative reception lines

Wan’Dale Robinson: 40+ yards gained (+200)

Because he tends to catch short passes, Robinson doesn’t necessarily have the profile of a receiver who will significantly exceed his prop line of 29.5 yards. And you can see in the chart below that Robinson is more likely to get 20, 30, or 40 yards than teammates Darius Slayton and Jalin Hyatt, but less likely to have an explosive play. At 40+ yards, however, that’s just a good price for Robinson: It pays 2-1 if he exceeds his over/under by just 10.5 yards.

The fact that Robinson had a very respectable 21.5% target share last year helps. And while it’s not factored into the model, I think the fact that Robinson runs quick routes, especially against the Vikings, who run a lot of blitzes under defensive coordinator Brian Flores, should be an advantage.

A new model I’ve been working on this offseason is a complete distribution of receiving yards for a single player in a single game.

Here are the Giants’ WRs in Week 1. The graph shows the probability (y) that each receiver will reach at least x yards. pic.twitter.com/zLipEJHQ2a

— Seth Walder (@SethWalder) 5 September 2024

KJ Osborn 30+ yards gain (+210)

One clue the model picks up is Osborn’s receiving yards props from a year ago, when he averaged an over/under of 30.4. Now, situation certainly plays a big role in those numbers and his current lines. But he can certainly be a receiver whose median line is around 30 under the right circumstances. The receiver roles in New England are a little unclear, so we’re using the uncertainty to our advantage by playing the alternate line. Plus, the fact that the Patriots are 8.5 points behind the Bengals means they’re likely to be behind — and throwing.

Also consider:
Jauan Jennings 35+ yards gain (+260)

Touchdowns anytime

OK, these aren’t exactly stealthy bets – I originally planned to focus mostly on non-touchdown bets – but they’re still backed by a model and therefore good enough to fit our criteria.

Nico Collins scores 1+ touchdowns (+185)

From a model perspective, there’s a lot of good about Collins here. He’s an outside receiver in an offense with a high implied team total (25.75), and he was nothing short of outstanding last season. One number the model likes to see: Collins ranked fifth in completed air yards per route at 1.91 (think: yards per route, but only in yards traveled through the air, since those are more repeatable than YAC yards).

Also consider:
Jaylen Waddle 1+ touchdowns (+145)
Tyreek Hill 1+ touchdowns (-125)
Tank Dell 1+ touchdowns (+260)

Multiple touchdowns

Tank Dell scores 2+ touchdowns (+2000)

It’s pretty clear that my Week 1 touchdown model is high on the Texans. I’m a little concerned that this might be because it doesn’t fully account for the additional target competition that Stefon Diggs brings, but at the same time, the model also doesn’t know — and I trust this — that Dell has been incredibly good in ESPN’s receiver scores, ranking eighth-best last year with an Open Score of 81. In general, the model likes Dell for many of the same reasons it likes Collins. Dell ranked sixth in air yards per route completed last year, exactly one spot behind Collins. I’m cautious with my touchdown models because they’re brand new, but I’ve rated Dell’s 2+ touchdown chances at +1265.

Alternative sum

Raiders at Chargers over 45.5 (+170)

This comes from a brand new FPI+ model developed by my colleague Dylan Mervis on the sports analytics team. FPI+ is a translation of our general Football Power Index into a model designed for use in the betting market, so why the alternative here? The combination of the Chargers’ offensive FPI rating (13th) and the Raiders’ offensive FPI rating (26th) leads the model to believe the median value should be a little higher than the current mark of 40.5. And looking at the range of possible values, the value above 45.5 is where we deviated from the market the most. FPI+ prices changing at +156.