Skip to Main Content
Betting, Fan Engagement, Integrity

Super Sunday: Breaking Down Player Prop Lines Using Stats Perform’s VQ

By: Stats Perform

VQ Product Owner Bryan Shumway (BFS) outlines a sample of Stats Perform’s VQ predictions and probabilities for Super Bowl LIV, and Kyle Cunningham-Rhoads (KCR) from our data science team considers why they match or vary from the consensus market line.

BFS: VQ uses our deepest historical  proprietary data feed (NFL X-Info*), and our experienced research team, to train an AI machine learning model. VQ generates predictions about how specific  players are expected to perform against certain teams in particular situations based on past performance or performances of similar players against similar opposing players and styles. 

Kansas City Chiefs’ Patrick Mahomes scrambles during the first half of the NFL AFC Championship football game.

Patrick Mahomes’ Rushing Yards 

Consensus Market Line: 30.5
VQ AI Prediction: 20.5

KCR: Mahomes averaged just 3 rushes for 15.5 rushing yards during the regular season for the Kansas City Chiefs, but he’s averaged 53 yards on 7.5 carries in the postseason. 

In the divisional round, the Houston Texans played almost all man coverage, and in the conference championship, the Tennessee Titans split between man-to-man and zone coverage. 

It’s easier for QBs to run against man coverage because the defenders’ backs are usually turned and they’re trailing receivers downfield.

The 49ers, however, play at one of the highest levels of zone coverage in the league, which means there will be four or five defenders just 10 yards off the line of scrimmage on almost every play. That’s not a lot of empty space for a QB to exploit. 

BFS: Our VQ API can also be used to derive the probability of other lines. Inputting the consensus gives the 30.5 a 69.65% probability to the under.

Patrick Mahomes’ Pass Attempts 

Consensus Market Line: 35.5
VQ AI Prediction: 36.5

KCR: The Chiefs are remarkably consistent with their game-scripts. Though they’re pass-first team, they run the ball enough to keep the defense honest.

Mahomes has averaged 36.2 attempts this season, including the playoffs but excluding a game against the Denver Broncos in October that he left early due to injury. Mahomes also had two matchups with the Los Angeles Chargers, who played at the league’s slowest pace. 

A standard game-script is expected for the Chiefs, and both VQ and the market say Mahomes will sit in a range of 35 to 37 pass attempts. 

Kansas City Chiefs’ Tyreek Hill catches a touchdown pass during the first half of the NFL AFC Championship football game.

Tyreek Hill’s Receiving Yards

Consensus Market Line: 77.5
VQ AI Prediction: 74.5

KCR: The last time Hill had over 72 yards was all the way back in Week 10, so it’s questionable to have such a high number for a game against a stingy defense. Put plainly, however, Hill is very good. Stats Perform tracks receiver/defender matchups, and Hill gets open more than 50% of the time when covered one-on-one. So why doesn’t he produce more? The average receiver will face a double coverage about 4% of the time, but Hill sees double coverage more than 19% of the time. 

Why is this relevant? The 49ers almost never use double coverage. Their Cover 3 defense relies more on using their speed to close off the passing lanes that the QB wants to throw through than shutting down individual receivers. Yes, Hill will get some exposure to shutdown corner Richard Sherman, but he’ll also see nickelback K’Waun Williams and right cornerback Emmanuel Moseley. Williams is a good slot corner, but Moseley is a below-average outside corner. The prediction and the market say Hill should be able to outwork the two of them for some big receptions. 

BFS: *The NFL X-info feed is the result of professional analysts studying every player position and role in every game regardless of  whether they’re in possession of the ball or involved in the play. The more trusted and deeper the data source, the better we can train our machine learning models. Put simply, AI models work better the more data they have to analyze. Instead of generalizing, they can create personalized probabilities for specific players and situations. 

Read more about Stats Perform VQ here.