Unfortunately for us daily fantasy football players, NFL coaches don’t always make rational decisions. If they did, I wouldn’t have to cry myself to sleep over Ladarius Green’s lack of usage each week.
But seriously, I cried myself to sleep this week over Ladarius Green’s lack of usage.
But seriously seriously, we can’t always make decisions based on player talent or how we think players should be utilized because so much of their fantasy production comes down to usage. I know for a fact that C.J. Spiller is a more talented player than Eddie Lacy, but I wouldn’t draft Spiller over Lacy in fantasy drafts this year, primarily because of usage.
Predicting workload is vital to fantasy success, and a major part of accomplishing that is understanding game flow—how a game will unfold in the most likely scenario. In-game decisions are strongly dictated by situations; coaches need to make decisions that (they believe) are optimal given the specifics of the game. If we can predict how games are likely to unfold, we can better grasp in-game decision-making, and thus better predict usage—the backbone of fantasy production.
How Game Flow Affects Usage
Before diving into the ‘how,’ I just want to show you some numbers from a study I recently completed on fourth-quarter passing stats. I charted attempts and YPA for quarterbacks in the fourth quarter, based on the score.
Quarterbacks who are leading by a touchdown or more typically record barely better efficiency than those down by seven points or more. The difference is negligible. Meanwhile, those down by a touchdown throw the football twice as frequently in the fourth quarter as those who are winning.
This is a pretty significant difference that suggests major shifts in play-calling based on the score. Many times, teams with the lead shouldn’t be so conservative with their play-calls. But they are, so we can benefit from that.
There’s a difference in touchdowns and interceptions, too.
Yes, losing quarterbacks throw more interceptions, but the harm that comes from an interception (-1 point) is just one-quarter as detrimental as the benefit that comes from a touchdown (4 points). Because quarterbacks down by a touchdown in the fourth quarter throw over twice as many scores as those leading, the net advantage is very positive for quarterbacks on trailing football teams.
If we add this together, we get a couple extra expected points for trailing quarterbacks in the fourth quarter. That’s from passing stats alone, with an extra boost to guys like Cam Newton who drop back and can take off on the ground.
So is all of this data evidence to target quarterbacks on losing teams? Not so fast.
Here’s the thing about losing football teams: they’re losing. Which means they usually didn’t score a lot of points. Which is bad. And typically, teams that are down in the fourth quarter are losing because of a lack of success through the air. Selecting a quarterback who has struggled all game because he’s going to see lots of fourth quarter attempts is like buying a piece-of-crap car because it has really effective windshield wipers (did that once, but I feel really safe driving in the rain now, nbd).
Maybe not all quarterbacks who are trailing are bad, though…
Using Vegas to Maximize Passing Stats
If we break down possible game outcomes into buckets based on the score, we’d get something like this: winning with lots of points, winning with few points, losing with lots of points, losing with few points.
The first two are of course best for actual teams, but for fantasy quarterbacks, we’d ideally like to see their teams score lots of points—yet fewer points than the opponent. That way, we get the benefit of (likely) early passing efficiency with the advantage of more late-game passing attempts.
So we’re basically looking for games that are projected to be close and high-scoring. Who in the world could possibly help us predict games outcomes like that?
Hint: It’s not my Uncle Bruce.
It’s Vegas! We can use the Vegas lines for all sorts of reasons—game props to predict player fantasy stats, the total to project overall scoring, and so on—but one underutilized action is using the lines and spreads to predict game glow.
Ideally, we’re looking for something like this: Broncos (-4) at Saints, Over/Under 64 points
In that scenario, we have a projected final score of 34-30 in favor of the Broncos. Both quarterbacks for these teams would likely be smart plays in such a scenario, but you might actually give the slight edge to Drew Brees and the Saints. They’re the team projected to score a ton of points, yet still be losing in the fourth quarter. Even if New Orleans is down only, say, 60 percent of the time, that’s a small edge that could end up tilting the scale in favor of Brees over Peyton Manning.
On DraftKings, it’s always important to keep in mind that every decision is cost-dependent. Maybe Manning is the best play in a vacuum in any given week, but he almost might eat up the highest percentage of your salary cap. If you can identify a cheaper quarterback who is likely to lead a team to a lot of points—perhaps one who has a high floor based on the game script projected by Vegas—that can sometimes be an even better situation to target.