What actually wins the Milly Maker: five years of winning lineups

One of the best ways to be successful in life is to study what the greats are doing. What are the winners doing in your field? DFS is no different. I've been studying winning Milly Maker lineups since 2020, looking at how DFS pros attack large-field GPPs, and hunting for actionable information we can use to predict future winning outcomes.

Whether you have the money to mass-enter the Milly Maker on a Sunday, or you just want to play the fifty-cent Mini Max or a free contest, these lessons carry across large-field tournaments. Keep in mind the winning Mini Max score often exceeds the winning Milly Maker score.

These are the winning trends over the past five seasons that we're taking into consideration when we set up the optimizer at Prediction Fantasy this season.

Scores: what it takes to win

The average score of Milly Maker winners across 2022–2025 was 236.5. The biggest swings were the 240.8-point high in 2022 and the 230.5-point low in 2024.

Typically a higher-scoring NFL season produces higher DFS scores, but that's not always the case. 2024 saw an average of 22.9 points per game, a full point higher than 2022, when the average Milly Maker winning score was 240.8.

What this really tells us is that a lot of chalk failed in 2024. There were multiple instances of the winning score being very low with the ownership reflecting it, like Week 7 of 2024, when the winning score was 185.36 and total ownership was 88.51%. Week 10 was similar: a low score of 196.88 on total ownership of 68.83%.

Ownership: the field got sharper

Total ownership in winning lineups averaged 116.1% across 2022–2025, with a high of 122.6% in 2023. Going further back, 2021 was as low as 96%.

What's interesting is the split. Between 2020, 2021 and 2022 the average total ownership was 104.3%. Look at 2023, 2024 and 2025 and the numbers are higher: 122.6%, 111.5% and 122%, an average of 118.7%. That's a 14.4% jump between the two data sets. The 14.1% increase between 2022 and 2023 looked like an outlier when it fell to 111.5% in 2024, but it jumped back up to 122% in 2025.

My theory is that the field got a lot sharper in 2023, and that's the boost. 2023 brought sims to the mainstream, with Stokastic offering NFL sims for Week 1. Simulating the contest thousands of times and telling a player their projected ROI levelled the playing field between casual players and diehard grinders.

So what happened in 2024, when sims already existed? A combination of factors, and number one was certainly chalk failing. We had multiple weeks with scores below 200 and an average ownership of 91% during those weeks. Remember, 2024 wasn't a historically low-scoring NFL season; it was 0.1 points away from being tied for the highest-scoring season in the sample.

I think chalk failed for several reasons. We had a very strong rookie QB class producing spike weeks at lower pricing and ownership than the Josh Allens and Lamar Jacksons who obviously draw more ownership. Their deflated pricing and rushing ability paved the way for more naked QB lineups winning, a QB without a stacked pass catcher, and most of the field just wasn't building that way. They still aren't, often, but sims have shown the field that you probably should be at times.

You also have guys like Josh Allen, who was difficult to stack week to week because there wasn't much consistency in the target share, but you just knew he could go for 30+ with his legs. As he did in Week 1 of 2024, when Diggs didn't produce a tournament-winning score at his salary to get into the winning lineup alongside him.

What we typically see in winning lineups is about 1–3 players under 5% ownership. This will undoubtedly be subject to variance as we'll get super contrarian lineups and chalky lineups, but with the average being 116% total ownership, that's the kind of construction you'll typically see.

The most popular stacks

• QB + 1
• QB + 2
• QB + 3
• Naked

Construction	2022	2023	2024	2025
QB + 1	5 · 28%	6 · 33%	3 · 17%	4 · 22%
QB + 1 + BB	3 · 17%	2 · 11%	2 · 11%	2 · 11%
QB + 2	2 · 11%	6 · 33%	7 · 39%	1 · 6%
QB + 2 + BB	5 · 28%	1 · 6%	1 · 6%	6 · 33%
QB + 3	1 · 6%	0 · 0%	0 · 0%	0 · 0%
QB + 3 + BB	1 · 6%	1 · 6%	0 · 0%	2 · 11%
Naked	1 · 6%	2 · 11%	5 · 28%	3 · 17%
Total	18	18	18	18

QB + 1 hit 26% of the time across 2022–2025, making it the single most common construction. Most of the time there was no bring back with it: QB + 1 with a bring back accounted for a further 12.5%, and never rose above 17% in any season.

Not surprisingly, QB + 2 wins more often than QB + 1. When you consider all the player combinations across 200,000+ lineups on any given Sunday, naturally there are going to be one or two games that go off, and stacking multiple players from those games is going to be optimal. QB + 2 won 39% of weeks in 2022, 39% in 2023, 44% in 2024 and 39% in 2025 (when we factor in bring backs). That is the steadiest construction on the board, and the reason it is the one to build around.

However, the bring back with QB + 2 is very hit or miss. In 2022, 28% of winning lineups had it. In 2023 and 2024 it happened just once each season, 6%. It jumped back to 33% in 2025.

Simply put: QB + 1 wins almost a quarter of the time, but using a bring back with QB + 1 is probably negative EV. QB + 2 wins much more often, and in some instances the bring back is needed, but it's not always necessary. It's great having an optimizer where you can get some game stacks without forcing one into every lineup.

Team stacks versus game stacks. Pure team stacks were most common in low-scoring slates. 2024 saw the highest rate of pure team stacks or no-game-stack builds, which also produced the most naked QB builds at 28%, the high across the sample, consistent with bring backs almost disappearing that season.

Game stacking appeared in 60–70% of winning lineups, but the key differentiator is that it wasn't always the primary stack. Often it was a secondary stack: a WR and RB in the same game from opposing teams. There's back-and-forth scoring, the RB is getting touchdowns on one side, and the opposing WR is accumulating yards and receptions for a tournament-winning score relative to his salary, without the touchdowns that would bring his QB with him. In Week 1 of 2025, Ricky Pearsall had 17.8 and JSN had 23.4; a naked QB build with Fields scoring 29.52.

Stacking, and the bring back

A bring back means you have one or more players from the opposing side of a game you're stacking. Say you're stacking the Ravens against the Bengals: you may have Lamar Jackson with Flowers, and on the other side of the game you want Chase or Higgins.

The bet you're making is that if Lamar and Flowers are needed in the winning lineup, the Bengals are forcing the Ravens to respond by scoring points; the shootout scenario. There are all kinds of variations: Lamar + 2 with one bring back, Lamar + 2 with two bring backs. The lineup is a reflection of the story the game is telling. We've had instances where a game goes for 80+ points and you need six players from it to win that week. When you stack, you're making a bet on the game and needing to get fewer things right, rather than hitting a nine-leg parlay.

Prior to 2023, if you were stacking a QB with at least two of his pass catchers, you almost always forced a bring back in large-field tournaments. Between 2023 and 2024 winning lineups used one far less; 12% and 5% respectively. In 2025 it jumped way back up to 44% of the time. So why the massive changes? Is it more descriptive than predictive?

During 2023–2024 we saw the real start of naked QB builds taking down large-field tournaments. Before 2023 it was rare; one week in all of 2022, 6%. In 2023, naked QB builds won 11% of the time. In 2024 it was 28%, five of the eighteen weeks, and the high across the sample.

Again, I think sims showed that naked QB builds can be optimal when the QB runs a lot and the receiver room is difficult to project week to week. We've seen it with Josh Allen, Patrick Mahomes, Bo Nix.

Obviously if naked QB builds win better than a quarter of the time in a season, we can expect bring backs to be down that year. You don't need the game stack if you're relying on the QB's legs. Think Josh Allen or Jalen Hurts rushing for multiple touchdowns.

There are still instances of a QB doing a lot on the ground, not bringing a pass catcher, but the other side of the game pushing the pace. Think the Lions putting it on a Justin Fields–led team. Fields threw for only 167 yards in Week 10 of 2022 but had 147 rushing yards and four touchdowns on his way to 48.5 DK points. Amon-Ra St. Brown had a great day in the same game with 25.1. The Fields naked build with the Amon-Ra bring back was the winning stack.

While a naked QB build is void of a primary stack, it doesn't mean that the lineup isn't correlated. In fact, every single winning lineup was correlated in some way. For example, in Week 12 of 2023, a Josh Allen naked QB build won to the tune of 43.6 FPTS from Allen; there was a mini game stack of Nico Collins and opposing receiver Calvin Ridley in another game. Every winning lineup across all five seasons had some correlation in it. Stacking remains king.

The flex has changed hands

• WR
• RB
• TE

If you've been playing fantasy football the past couple of years it's probably no surprise that RB has surged in the flex, even on a full-PPR site like DraftKings. Wide receiver dominated most years with a 42% average, peaking at 60% in 2022, but RB hit the flex in 56% of winning lineups in 2025. TE hit 28%. 2025 was a big down year for WR in the flex at only 16%.

There are multiple reasons. One of the big ones is that NFL defenses are playing more high safeties to cut down on big plays, and offenses are responding by running the ball more. We've also seen far fewer injuries the past couple of years to the league's big-time backs: JT, CMC, Saquon.

Prediction: we will see more WR in the flex this year than we did in 2025. 16% is not a repeatable number. I'm not sure we'll return to the 42% average, but I expect closer to 30% than the teens.

Vegas totals tell you how to build

There's a lot of discussion in DFS about Vegas game totals, and the chalk often centers on the highest-total games. So how accurate are they at predicting the environments that produce winning lineups?

Every season, the winning stack's game total was above the league average. Higher totals did correlate with higher winning scores.

• Winning stack's game
• League average

Winning stack's game matched to nflverse game data for all 72 weeks, 2022–2025.

As expected, bring backs are more likely when the totals are higher, though not perfectly, with 2024 being the exception. When the winning stack totals were 46.14 in 2022, we had a 50% bring back rate. In 2023, when the winning stack totals fell to 44.6, the bring back rate declined to 23%. However, we've talked about how 2024 was quite the outlier: when the winning stack score rose back up to 46.3, the bring back percentage did not climb with it; it was only 17%. In 2025 we saw the highest stack score of the sample at 47.44, and the bring back percentage was the highest we had seen, 56%.

Favorites, underdogs, and home field

Between the 2022 and 2025 seasons, the favorite team's QB, and the stack, was in the winning lineup 67% of the time. This was overwhelming between 2022 and 2024, when it averaged 74%. 2025 was the outlier here, as underdog stacks saw a huge resurgence and were in 56% of the winning lineups. Favorites only won 44% of the time that year.

The key to playing underdog stacks is doing it with low-owned quarterbacks in high game total environments. What was the key differentiator was whether you did it at home or on the road. Underdog quarterbacks hit on the road just as much as away favorites did, at 24%. However, underdog home quarterbacks were in the winning lineup only 10% of the time. Easily a fadeable spot.

Winning stack's game matched to nflverse game data for all 72 weeks, 2022–2025.

43%

Home + Favorite

31 of 72 weeks

24%

Away + Favorite

17 of 72 weeks

24%

Away + Underdog

17 of 72 weeks

10%

Home + Underdog

7 of 72 weeks

Winning stack's game matched to nflverse game data for all 72 weeks, 2022–2025.

Home teams appeared in 53% of winning lineups. This isn't a strong signal in itself; the signal is home favorites, who were in 43% of winning lineups, almost twice as many as away favorites and away underdogs, both at 24%. Home underdogs only won 10% of the time and were the easiest fade.

Where the salary cap is going

2021–22 saw the most winning QBs priced between $5,500 and $7,500. By 2025, 89% of winning QBs were under $6,500 and a third were under $5,500. This is something I often see studying DFS pros; they aren't afraid to go for the cheap contrarian stack, because when it hits, the lineup massively leverages the field as they're able to pay up for floor and ceiling elsewhere.

RB salary is rising and so is cap allocation, climbing every year from $5,740 in 2021 to $6,625 in 2025. In 2025, 53% of winning RBs cost $6,500 or more. That money used to go to QB and TE. One caution: if we start to see a plethora of injuries at running back, this will change, and the salary will flip back to other positions as RB savings open up the rest of the lineup.

WR salary has always balanced stars and scrubs: 35% cheap, 31% mid, 17% high, 17% elite. TE went from elite Travis Kelce pricing to a stable $3,500–$4,500 streamer, which is no surprise; I always recommended streaming at tight end when I did cash content, as it's the highest position of variance next to DST. DST average salary rose from about $2,800 to $3,100 by 2025, and the $3,000–$3,500 tier is the most common price point in winning lineups. A differentiated, well-priced defense is a great decision; they typically provide some of the best points per dollar in a winning lineup.

5 Key Takeaways

Lean into the Vegas game totals. Winning lineups featured stacks from games that ran above league average in all the seasons sampled; this ranged from 1.75 to 2.57 points. With the exception of 2024, higher totals correlated with higher winning scores across all seasons.

Target favored QB stacks. Much like the game totals, the game spreads gave us a real sense of the players to target in these stacks, as the Milly Maker winners featured 67% of favored QB stacks.

QB + 2 is the most common stack in winning lineups. QB + 1 of his teammates hits about 25% of the time, but QB + 2 was by far the most dominant stack with an average of 40% across the sampled seasons. You'll get the occasional onslaught stack of QB + 3, and naked QB builds are on the rise, but if I was forced to pick one way to build my lineups to take down the Milly Maker or a contest of similar size, I would stick to QB + 2.

Naked QB builds are on the rise. Blame it on Sims, blame it on offenses that spread the ball out, or blame it on offenses that cater to mobile quarterbacks who can win the game on their legs, but naked QB builds are on the rise. Since 2023, we've seen naked QB builds win the Milly Maker 18% of the time.

RB is back. Whether it be due to two high safeties, positive injury variance, or a combination of the two, running backs are hitting the flex more often and RB salary is on the rise in winning Milly Maker lineups; seeing a five year high of hitting the flex 56% of the time in 2025, and also climbing to an average salary of $6,625 in 2025.

Stay tuned to Prediction Fantasy as we continue to track the winning trends of Milly Maker winners as we head into the 2026-2027 NFL season.