"Stack your lineup" means something different in every DFS sport, and treating it as a single generic rule leads to worse builds in at least one of them. We compared our own MLB GPP dataset (~76,000 opponent lineups, covered in our MLB study) against a large DraftKings NFL Classic contest — 277,000 entries — to isolate what the correlated unit actually is in each sport, and how differently winners built around it.
The two datasets weren't built the same way — the MLB numbers come from 24 contests spanning 11 slate nights, while the NFL numbers come from one very large single-slate contest — so this piece leans more on comparing the shape of the correlation effect than on treating every percentage as equally validated. Where that distinction matters, we call it out below and again in the caveats at the end.
MLB: the correlated unit is the team
In MLB, correlation comes from the batting order itself — hitters who bat near each other score together, because a runner on base for one player is frequently driven in by the next. Our MLB research found that stacking the slate's #1 run-scoring team produced a top-10 finish rate roughly 35x higher than stacking a team ranked 7th or worse, and that a genuine second stack (2+ hitters from a second team) roughly doubled the conditional top-10 rate. The unit worth optimizing around is the team, and specifically the team with the best Vegas-implied scoring environment on the slate.
NFL: the correlated unit is the game
NFL doesn't have a batting order, and a single team's passing offense only correlates with itself up to a point — a QB's big game doesn't guarantee his RB1 also has one, since rushing and passing production trade off against each other on the same play count. What NFL games do have is a total: when a game gets high-scoring, points get spread across both offenses, not just one.
In the 277,000-entry NFL Classic contest we reviewed, QBs from the slate's single highest-scoring game appeared in 71-80% of top finishers, compared with just 13% of the field at large. That gap — top finishers overwhelmingly concentrated in one specific game environment, the field spread thin across all of them — is the NFL version of the MLB team-selection effect, except the unit is the game, not either team individually.
That distinction matters practically. An MLB player deciding which team to stack is making essentially one decision. An NFL player deciding which game to build around is implicitly making a decision about two teams at once — and, as the next two findings show, about which specific players on both sides of that game to roster.
Bring-backs mattered more than we expected
A "bring-back" is a player from the QB's opposing team in the same game — betting that if the game goes high-scoring, both sides score, not just your stack's side. Bring-backs were present in 71-80% of winning lineups, versus 42% of the median field lineup. That's a real, replicable construction signal: winners weren't just picking the right game, they were rostering both sides of it.
This is arguably the cleanest NFL-specific finding here, because it has no real MLB analog — there's no equivalent to "rostering the opposing pitcher" in a baseball stack. It's a construction move that only makes sense once you've accepted that the game, not the team, is the unit you're betting on.
QB + 2 pass-catchers beat the field's default shape
The field's default NFL stack is a QB plus one same-team pass-catcher. Top-100 finishers skewed heavier: 36% of top-100 lineups carried a QB plus two same-team pass-catchers, well above the field's baseline rate of building around just one. More passing-game correlation, not less, showed up disproportionately among winners.
Read alongside the bring-back finding, a clear winning shape emerges: QB with two of his own pass-catchers, plus a player from the opposing offense. That's a three- or four-player correlated block built entirely around one game, which is a meaningfully more concentrated bet than the field's typical QB-plus-one-guy default.
Forcing a same-team RB with the QB was not worth it
We also checked whether pairing the QB with his own team's lead RB carried an edge, the way a hitter-pitcher pairing might in some formats. It didn't. Winners rostered a same-team QB+RB pairing at a 0.30-0.37 rate, barely above the field's 0.20 baseline — a real but weak gap, nowhere near the size of the game-selection or bring-back effects. Treat QB+RB as optional, not a construction rule.
This makes structural sense once you think about how a passing offense allocates production. A QB's ceiling game is driven by throwing volume and efficiency, which more directly correlates with his pass-catchers than with his own team's running back — a big passing day can just as easily come at the running game's expense as alongside it. The RB pairing isn't wrong, exactly; it's just a much weaker correlation bet than the field's own instincts might suggest.
Winners carried more ownership, not less
Same finding as MLB: NFL winners were not contrarian. Summed lineup ownership among winners averaged 136 versus a field median of 93 — winners were more concentrated on popular players than the average field entry, not less. Picking the right game environment and being on the field's consensus players inside it beat trying to be different for its own sake.
Duplication: classic vs. showdown is not close
Format matters as much as sport here. In single-game showdown contests, we found brutal duplication — 37 to 52 identical lineups inside a single top-200. In the full-slate NFL classic contest, by contrast, winning lineups were nearly unique, topping out at 4 copies of any single build. If you're playing multiple entries and want each one to have a genuinely independent shot, classic formats give you that; showdown formats can leave a big chunk of your entries splitting a payout with strangers who built the same lineup you did.
The mechanism is straightforward: a showdown slate has one game and a small captain/flex player pool, so the field's "obvious" build converges on a handful of combinations. A classic slate has a full player pool across many games, so even a field that broadly agrees on which game to attack has far more ways to fill out the rest of the roster. If duplication risk matters to how you allocate a multi-entry bankroll, that's a format decision as much as a construction decision.
What this means for building NFL lineups
Chain the findings together and the construction order for NFL looks different from MLB's team-first approach: pick the game first, using the same kind of implied-total ranking that drives MLB team selection, then build the QB stack toward both sides of it — your own pass-catchers plus a bring-back — rather than toward same-team correlation alone. RB pairing is a minor, optional lever, not a required piece. And because winners in both sports carried market-level ownership rather than avoiding it, the "right" construction in NFL, like MLB, is usually the popular one applied to the correct game, not an unpopular one applied for its own sake.
Side-by-side comparison
| Dimension | MLB | NFL |
|---|---|---|
| Correlated unit | Team (batting order) | Game (both offenses) |
| Why | Runners scored by the next hitter | High-scoring games spread points across both sides |
| Primary tilt | Highest Vegas implied team total | Highest Vegas implied game total |
| Second stack | 2nd team stack (2+ hitters) | Bring-back from the opposing team |
| Spend-up target | Real starting pitching | QB in the slate's top game |
| Ownership vs. field | Winners at market, not below it | Winners above the field median |
Practical takeaways
- In MLB, pick the team first — a single-team stack is the primary lever. In NFL, pick the game first — a QB stack only works if it's attached to the slate's best scoring environment.
- In NFL, build in a bring-back by default; it showed up in the large majority of winning lineups and only a minority of the field.
- Don't force a same-team QB+RB pairing in NFL — the data shows a weak effect at best, not a rule worth building around.
- In both sports, don't manufacture contrarian plays for their own sake. Winners in both datasets carried ownership at or above the field median.
- Prefer classic-format multi-entry over single-game showdown when duplication risk matters to you — showdown fields duplicated far more aggressively in our sample.
The NFL findings here come from a single 277,000-entry DraftKings Classic contest, not a multi-week sample like our MLB dataset — treat the specific percentages as directional evidence for a correlation pattern with strong theoretical backing (game-total variance driving both offenses), not a fully validated multi-slate rule the way our MLB findings are. We'll update this piece as more NFL contest data comes in.