The Draft Desk

Stack Your Lineup. The Data Insists.

When a quarterback has a big day, his top receiver almost always has one too. The tight end comes along half the time, the running back barely notices, and the fear of the opposing ground game is a myth.

Stacking, drafting a quarterback together with his own pass catchers, gets treated as a tournament-bro exotic. It is the most data-supported roster idea in fantasy football, and the play-by-play record is blunt about which stacks are real and which are stories.

Start with the lock. When a quarterback scores 20 or more PPR points, his top receiver clears 15 in 82 percent of those weeks, against a 59 percent base rate. The two positions are wired together for the obvious reason that they score off the same completions, and the wiring has held in every measured season: quarterback and top receiver land on the same side of their respective league medians in 68 percent of all team-weeks, where independence would say 50. That is a correlation you have to actively avoid, not one you hunt for.

Ceiling weeks are stacked weeks

The tournament argument in one number: among the team-weeks in the top tenth of combined quarterback-plus-top-receiver scoring, 94 percent feature both players spiking at once, versus 23 percent of all weeks. Part of that is arithmetic, the top decile of a sum naturally favors both parts being high, but the totality is the finding: the single-spike week, one player carrying a quiet partner to a monster total, essentially never reaches the ceiling tier. If you are chasing the week that wins a tournament outright, it is a stacked week or it is nothing.

The tight end is half a stack

The tight end version works at half strength. In quarterback spike weeks the top tight end clears 12 PPR points 52 percent of the time against a 36 percent base, a real lift, but his spikes are disproportionately touchdown-driven, which makes the week-to-week co-movement lumpier than the receiver link. Price it as a discount stack, not an equivalent one.

The lead back is an island. His quarterback's day moves him five points of probability, and the opposing grinder moves the quarterback less than folklore says.

The running back island

The lead back is close to independent of his own quarterback: 68 percent spike rate in quarterback spike weeks against a 63 percent base, and most of that thin residual comes through pass-catching backs collecting the same completions. Even the feared pass-heavy script barely touches him in PPR formats: in 40-plus-attempt games the lead back beats his usual scoring 48 percent of the time, a hair under the 50 percent baseline. The reception points largely refund what the vanished carries take away. Stack your quarterback with his receivers; roster his back on the back's own merits.

The opposing-RB myth

The folklore says a grinding opposing ground game starves your quarterback of possessions, so avoid quarterbacks facing a bell cow. Measured: when the opposing lead back logs 20 or more carries, the quarterback spikes at 22 percent against a 28 percent base. A dent, not starvation, and even the dent is partly backwards, since 20-carry games disproportionately happen when the grinder's team is already leading. Meanwhile the correlations fans should actually respect run the other way: quarterbacks spike more often when the opposing quarterback also spikes, 37 percent against 27, because shootouts feed both sides.

One last excuse, removed. The stack does not switch off against good defenses: facing a top-8 scoring defense, the top receiver still spikes in 78 percent of his quarterback's big weeks against a 52 percent base inside those same matchups. The coupling is the offense's property, not the matchup's.

How to draft with it: pair your quarterback with his top receiver where the price allows, treat the second receiver as a real stack piece rather than a hedge, since he rides the same wave at 66 percent versus 43, take the tight end version at a discount, and stop avoiding quarterbacks who face a workhorse back. The data has been running this experiment weekly since 2009, and it keeps returning the same answer.