Pie For Dinner: Rufus Peabody’s NFL Player Prop Process

It’s the only menu out there that makes the Cheesecake Factory feel like a flash card. It’s the Super Bowl prop bet menu, and it’s your best friend for the next few days. Where do you even start with NFL player prop bets during the Super Bowl? With team-wide projections? With the players? With a game script? 

No, you start with pie.

Pie? Pie. 

At least that’s how Rufus Peabody does it: by divvying up the pie on player usage to form a foundation that informs the breadth of his Super Bowl player prop plays. A little flour. A little baking soda. A little Travis Kelce snap counts and target shares.

Get a Slice

Of the 1,020 minutes available in the 2022 regular season, the Chiefs held the ball for 514 minutes, 31 seconds. In that time, they ran 1,068 total offensive plays, not counting punts, kickoffs, etc. They attempted to pass 651 times, and they ran the ball 417 times. That works out to 61 percent pass plays and 39 percent running plays. 

Last year’s Eagles team squeezed in a few more, running 1,080 plays. Of those, 536 were pass plays and 544 stayed on the ground, for a nearly 50/50 split.

Those are numbers you can use to start to build a baseline of what a team’s offensive plan may be. They’re available to anyone on Pro-Football-Reference.com. You don’t need to buy special data packages or acquire API access to a database. 

And it’s by building those baselines that you can start to work on creating projections for NFL player prop odds.

Game State And Snap Counts for NFL Player Prop Bets

“I’m going from the team level to project the number of plays a team is going to run,” Peabody said. “From there, what’s going to be the run-to-pass ratio?  Which is going to be not just based on what their run-to-pass ratio was in the past, but relative to what their game states were in the past. Then I’m basically projecting out, based on this spread, this is what we expect the average game state to be for a baseline rushing, passing percentage.”

Game state is a critical part of the process. How often was a team playing from ahead or behind, and what did they do in those situations? When you start to subdivide run-to-pass ratio by game state – how often do teams run when playing with the lead, how much do they pass when behind, etc. – you lay the foundation for your player prop work.

Way Ahead, Way Behind or Tight Games?

In the Chiefs’ 44-21 crush job against the Cardinals on Sept. 11, 2022, they ran 23 plays when they were up three possessions. Not surprisingly, they rushed 13 times. In an average 23-play stretch, they’d run nine times given their 39 percent rush rate. (That may not be surprising to you, but it’s absolutely killing “Run the Damn Ball” hat-wearers.)

In a tight 27-24 game against the Chargers the following week, the Chiefs had 38 plays down one or two possessions. They only ran eight times. Game state matters, and you can use it to weight your projections. 

“I’m not looking at AJ Brown getting 5.2 catches per game in the past,” Peabody said. “I’m looking at what’s AJ Brown’s target share? What’s his snap percentage? And I’m looking at what percentage of the pie he’s had in the past because the pie for this game may be very different than the pie in previous games.

“I’m projecting out, based on the spread, this is what we expect the average game state to be for a baseline rushing, passing percentage,” Peabody said. “For individual player stuff, it really comes down to figuring out usage rates. Basically, trying to figure out what percentage of snaps is this guy going be on the field for? How is that going to break down run/pass? What’s his route percentage going to be? The real work is determining that sort of participation stuff. From there there’s a lot of simulation for distributions.”

Taking it to the NFL Player Prop Simulator

Once you’ve established what you expect participation to look like, you can take your numbers to the Props Simulator to see distributions and start pricing out lines. 

Coming into Super Bowl LVII, the Eagles averaged 32 rushing plays per game. Miles Sanders had 259 of Philly’s 544 carries, close to 48 percent. If we believed everything in the Super Bowl would hold true to the season-long average, we could have ballparked Sanders at 15.4 carries. If we thought he maintained his season-long 4.9 yards per attempt, he’d come in around 73.5 yards.

We can run that through the Prop Simulator and see the median is 69.5 yards, which is well above his 62.5 line at FanDuel.

It’s unlikely we’ve found that the market is wildly off just by using back-of-the-envelope math. Now what?

Number of plays and game state is where we can start refining our numbers. 

“In projecting, I’m figuring out how much was noise and how much was signal,” Peabody said. “Just because a team averaged 66 plays a game this season doesn’t mean I’d project them to.”

Form Follows Function, Props Follow Game State

If we thought this was the kind of game where the Eagles were going to run more or less than normal, we could adjust. The more plays you think a team will run in a game, the more player projections will deviate from season averages, and vice versa 

There’s some art to usage, too. Kenneth Gainwell out-touched Sanders 11-1 in the second half against of the NFC Championship against San Francisco. Was there an advantageous matchup there? Did that hint at Gainwell getting more carries moving forward? Was Sanders just sitting because the game was in hand and his coaches didn’t want to risk an injury? Do Hurts’ second-half rushes with a bad shoulder discount that “protection” theory?

Some game state data is widely available. You can see how many plays teams ran while ahead, behind or tied, and what individual players did in those situations.

But some data isn’t as conveniently compiled. For that, you may need to comb box scores and drive charts. Is the juice worth the squeeze? For instance, if you wanted to know what the Chiefs were doing about injuries, one way you could draw parallels wass by looking at game states where they were in blowouts and starters came out of the game.

Accounting for, and Using, Injuries

No one’s really healthy by this point on the football calendar. But then again, not everyone had 240-pound Jacksonville linebackers falling on their ankle, either.

Mahomes’ high ankle sprain suffered in the Divisional Round was just part of the equation. JuJu Smith-Schuster, Mecole Hardman and Kadarius Toney were all injured in the AFC Championship game. Because of that, prop lines hadn’t been set for the bulk of the Chiefs receivers just a few days before the Super Bowl. 

Not only do you have to get creative about how those injuries might play out for those players, but you have to consider how they might affect the other pass catchers in the game.

“I can’t project (Marquez Valdes-Scantling) without knowing whether I expect JuJu to play and what role he’s going to have,” Peabody said. “I have to make a judgment call. In Mahomes, is he going to be scrambling at the same rate as normal? You have to think about how that is going to affect his past attempts and completions as well. If he’s less likely to scramble, he is going to be throwing. It’s making a judgment call and going off the information I have and, and trying to have a framework to look at it where you look at other quarterbacks that have been hobbled.”

Drawing Parallels for NFL Player Prop Bets

This is partly where attention to game state can come in because you can see how teams have played in certain situations that offer parallels to injury. When the Eagles were beating up the Giants in the 2022 Divisional Round, for example, A.J. Brown only played about 60 percent of snaps compared to his usual 84 percent.

If Brown was to get injured in practice this week and be ruled out for the Super Bowl, we could go back to those blowout situations and see who was most likely to pick up Brown’s targets. 

Keep in mind, though, that the Super Bowl is unique as the last game on the calendar, so when you’re divvying up the pie based on game state, you have to make some adjustments.

“You might have a situation where a guy gets rested in the fourth quarter (in blowouts),” Peabody said. “(Travis) Kelce might not be playing in the fourth quarter as much as he would be in a close game, but (teams) aren’t leaving any bullets in the holster. Your star players are going to be getting more snaps, at a slightly higher percentage than the season average.”

The Waiting Is the Hardest Part

If you’ve put it all together, you’ve got an idea of how many snaps each team will see and how they’re going to be spreading the ball over those snaps. You’ve used our simulator to feel out what your player distributions look like. After that, there’s nothing left to do but sit back and wait out your number.

The conventional wisdom holds true: bet Overs early and Unders late. Once the lines are released, sharp action and arbitrage players scoop up the Overs. Still, there’s plenty of value left to be had. Peabody himself will be waiting it out as long as he can before he ships the bulk of his action.

“When you get closer to the game, you’re going to start seeing public action drive these numbers up,” Peabody said. “That’s when most of my volume’s going to be this year. It’s going to be as close to the game as it can.”

But before you can kick back and wait, you need to know what you’re waiting for. With NFL player props it starts, like a cartoon hobo eyeing a country windowsill, with pie.

Determine the number of plays you expect in the game, distribute targets and carries, factor in game state to fine tune your projections and you’ll be slicing up the Super Bowl pie like you’re Marie Callender.