WATCH: Adam Levitan, Evan Silva review the NFL offseason

The Fantasy Football Millionaire — aka the Milly Maker — is an extremely unique tournament. It often has close to 200,000 entries, roughly 33 percent of the total prize pool goes to first place, and someone wins an absurd $1 million on a $20 buy-in.

Because this tournament is so unique, optimal strategy strays significantly from the norm. It’s very easy and lazy to simply say we need to pick the best players. Of course we do. But there’s a lot more strategy we can employ beyond that in terms of ownership, roster construction, flex usage and stacking. By optimizing a strategy beyond “pick the right players,” we can give ourselves a repeatable process that hopefully accentuates our chance — albeit very small — to win.

With the help of Fantasy Labs’ Contest Dashboard tool, I deconstructed the winning Fantasy Football Millionaire lineups over the past two seasons. In all, there were 29 of the $20 or $27 buy-in variety. It’s a smallish sample, and the highly variant nature of the tournament brings luck into play. But I do believe some viable trends emerged among the Milly Maker winners.

100% had at least one player who was less than 5 percent owned

I often ask myself how wild we have to be to reach the top of the Milly Maker. Can we win it by rostering all solid plays, or do we need to get really weird?

The data shows it certainly helps to get on a low-owned play. Fifty-one percent of Milly Maker winners had at least one player less than 2 percent owned, and every single one had at least one player less than 5 percent owned. To take it a step further, 10 of 29 Milly Maker winners (34 percent) rostered a player under 1 percent owned.

I’m not sure I’ve ever rostered a player less than 1 percent owned, and I’m sure many people reading this haven’t, either. It’s not necessary to win the Milly Maker, but it is a reminder that we need to be conscious of including some far-off-the-board plays in our lineups for this specific, unique tournament.

45% spent $5,900 or less on QB

We’ve talked for years about why spending down at quarterback makes sense. The full-PPR format, the insane depth of the position and the predictability of QB opportunity all are contributing factors. So I’m not surprised that 45 percent of Milly Maker winners spent less than $6,000 on their quarterback.

The key stat here is that just 23 percent of total Milly Maker entries spent less than $6,000 on quarterback in this sample. So by finding upside for cheap at the position, we really can gain an edge on the field.

It was somewhat rare that a Milly Maker winner used a high-priced quarterback, as just 28 percent of them spent $6,600 or more. The overall field used a $6,600-plus quarterback at a much higher rate — 54 percent. Put another feather in the cap of Team Spend Down at QB, the close cousin of season-long’s Team Wait on QB.

Only 7% used TE in the flex

Of the 29 Milly Maker winners, 16 used running backs in the flex (55 percent), while 11 went with wide receivers (38 percent) and two (7 percent) used tight ends. If we zoom out and look at all 6,595,643 rosters entered in those tournaments, 47 percent used RB in flex, 38 percent used WR and 14 percent used TE.

The number that stands out here is tight end, where the double-TE strategy was employed by a surprisingly high number of people but wasn’t effective in getting the $1 million. It makes sense, as tight ends typically don’t have the ceiling necessary to beat a 200,000-entry field. Last season, Travis Kelce (4), Rob Gronkowski (3) and Jared Cook (2) were the only tight ends to have multiple 100-yard games. Meanwhile, 26 different wideouts got that bonus.

86% used at least $49,800 of the $50,000 salary cap

Having a duplicate lineup as someone else in the Fantasy Football Millionaire is a disaster. In the event we do run pure and beat this massive field, we can’t afford to chop the prize with one or (god forbid) multiple people. I don’t mean this from a monetary perspective — I’d do inhumane things for $500K. But from a game theory and expected value perspective, having a duplicate lineup is unacceptable: Every time we give ourselves a 0 percent chance of winning the top prize before the games even start, we lose a ton of equity.

So for that reason, the theory that you should always leave money on the table is a popular one among sharp players. After all, 76 percent of all entries in the Milly Maker use at least $49,800 of salary. So in theory, a simple way to differentiate yourself is to leave at least $300 on the table. However, none of the 29 winners I looked at had a duplicate lineup, even though 25 of them spent at least $49,800. Note that 18 used the full $50,000 and six used $49,900. We should be more worried about cumulative ownership (see below) than punting the cap.

72% spent $3,100 or more on D/ST

Scoring at D/ST is unpredictable — even more so than at the other positions. After all, you likely will need a pick-six, scoop-and-score or return touchdown in order to win the Milly Maker. So I suspected that paying down at D/ST –- saving salary at a highly variant position in order to spend up at others — would make sense.

But a majority of Fantasy Football Millionaire winners (72 percent) spent at least $3,100 on their D/ST, a far higher number than the total field (51 percent). Also note that 38 percent of Milly Maker winners spent at least $3,500 on their D/ST while the total field was at 28 percent. There could be a few reasons for this, including short-term variance. But I suspect the higher-priced defenses often are under-owned relative to their expectation, especially in weeks when the salary cap is tight. So spending up at D/ST creates leverage from both an ownership and equity perspective.

79% used a QB stack

By now, almost all DFS players are aware of the mathematical advantages of stacking. You don’t have to understand Python or machine learning to realize that if a quarterback has a big game, at least one of his pass-catchers also will have a big game. So it’s not a surprise that 79 percent of Milly Maker winners used a QB stack — it’s actually more surprising that 21 percent got there without a stack (aka naked QB).

If we dig deeper into the winning stacks, we see just 17 percent used at least three players from the same team (QB/WR/WR, QB/RB/WR, QB/WR/TE, etc.). Double stacks were overused relative to effectiveness. But 31 percent included a tight end in their stack, which is interesting because of how infrequently winners used tight end in their flex (see above).

83% rostered at least one player who was more than 20% owned

There’s a common misconception that we can’t roster any “chalky” plays if we want to win the Fantasy Football Millionaire. It’s just not true. The popular plays often are the best plays, especially in the case of an injury-induced value, role change not reflected in price or uniquely awesome matchup.

Note that 45 percent of Milly Maker winners rostered at least one player who was more than 30 percent owned. The key to roster differentiation and upside is to ensure we don’t have a full roster of the chalk. That’s where cumulative ownership comes in (see below).

86% had average cumulative ownership below 15 percent

We can calculate our ACO (average cumulative ownership) simply by adding up our total ownership and dividing by nine (the number of roster spots). We know it’s more than fine to roster guys at above 20 percent owned, and we also know rostering guys over 30 percent often works. But we need to balance that out with some lesser-owned plays so our cumulative ownership doesn’t creep too high. A whopping 25 of 29 Milly Maker winners had ACO below 15 percent, proving how crucial this number is.

I am a promoter at DraftKings and am also an avid fan and user (my username is adamlevitan) and may sometimes play on my personal account in the games that I offer advice on. Although I have expressed my personal view on the games and strategies above, they do not necessarily reflect the view(s) of DraftKings and I may also deploy different players and strategies than what I recommend above. I am not an employee of DraftKings and do not have access to any non-public information.