DFS GPP & MME Strategy: How to Build Tournament Lineups

Step 1: Start With the Contest, Not the Optimizer

A 20-max tournament, a 150-max contest, a small-field single-entry GPP, and a massive-field lottery-style tournament do not call for the same portfolio.

Before you build, define:

• how many lineups you are actually entering
• field size
• payout concentration at the top
• how duplicated popular lineup constructions may be
• whether the slate rewards stacking or other forms of correlation

Your optimizer settings should come from those contest conditions, not the other way around.

Step 2: Build a Player Pool You Are Willing to Lose With

MME does not mean using everybody. A wide player pool can create the appearance of diversification while quietly adding combinations you never actually wanted.

Start by deciding who belongs in the pool. Use projections, Research Stations, coaching notes, slate analysis, injuries, starting lineups, weather, and other sport-specific information to decide which players and teams fit your tournament thesis.

A useful test is simple: if a player appears in one of your final lineups, can you explain why that player was in your pool? If not, the optimizer is making a strategic decision you should probably make yourself.

Step 3: Use Stacking to Build Correlated Lineups

In many DFS sports, the best tournament lineups are not collections of unrelated median projections. They benefit from game environments and player combinations that can succeed together.

Stacking rules are therefore a portfolio tool, not just a lineup-construction shortcut. The exact stack depends on the sport:

• NFL:quarterback/pass-catcher combinations and possible bring-backs.
• MLB:correlated hitters from the same offense.
• NHL:line and power-play correlation.
• NBA:game environments and role correlation can matter, but rigid stacking rules are usually less central than in NFL or MLB.

Do not force the same stacking template onto every slate. Correlation is the principle. The implementation changes by sport and contest.

Step 4: Set Unique-Player Rules With a Purpose

Unique-player settings control how different generated lineups must be from one another. Too little variation can leave you with a portfolio that depends on nearly identical combinations. Too much variation can force weaker plays into lineups simply to make them different.

Use more uniques when you want broader lineup diversity and fewer when a small slate or concentrated player pool makes heavy overlap unavoidable. There is no universal “correct” number.

DFS Army’s public MME examples have used unique-player rules as part of the build, but those settings changed with slate context. Treat them as examples of process, not permanent presets.

Step 5: Treat Exposure as Risk Management

Exposure is the percentage of your lineups containing a player, stack, or other lineup element. It determines how much of your tournament portfolio depends on one outcome.

Maximum exposure can keep a favorite play from appearing in nearly every lineup. Minimum exposure can force a player or stack into enough lineups to matter. Neither setting tells you whether the underlying take is good.

The goal is not to avoid risk. GPPs require risk. The goal is to choose where your portfolio takes it.

Step 6: Use Ownership as Context, Not a Rule to Fade Chalk

Projected ownership helps estimate how the field may allocate its lineups. That information is useful because tournament value depends partly on what your opponents are doing.

But “low owned” is not the same thing as “good.” A player can be unpopular because the projection, role, matchup, or ceiling is weak. Likewise, a popular player can still be correct when the salary and opportunity are too strong to ignore.

Think at the lineup level. A lineup can contain popular players and still be differentiated through stacks, secondary correlations, salary construction, or lower-owned ceiling plays elsewhere.

Where the current Domination Station supports ownership controls, use them to shape the portfolio you already intended to build, not to mechanically force every lineup below one arbitrary ownership total.

Step 7: Randomization Should Add Range, Not Replace Analysis

Randomization can prevent the optimizer from treating small projection differences as absolute truth. That can be useful in tournaments because projections are estimates, not certainties.

The tradeoff is obvious: too much randomization can turn a researched player pool into noise. Use it as a controlled way to widen outcomes, not as a substitute for deciding who belongs in the pool.

Step 8: Generate More Than You Need, Then Review the Portfolio

The optimizer output is not necessarily the finished entry set. Generate the build, then inspect it.

Check:

• player exposures
• stack exposures
• how often the same core combination repeats
• ownership concentration
• salary patterns
• whether a player or stack appears far more or less than you intended
• whether late-news players have entered the pool correctly

If the portfolio does not represent your strategy, change the rules or player pool and run it again. Do not keep a bad build just because the optimizer completed successfully.

Step 9: Make Late News Part of the MME Process

Late news can change projections, ownership, starting roles, weather, and the value of an entire game environment. The larger your entry set, the more important it is to understand how one change affects the whole portfolio.

Before lock, recheck the current DFS Army projections, late-breaking news, sport-specific coaching, starting-lineup information, and weather where relevant. When late swap is available and currently supported in the applicable workflow, use it deliberately rather than treating the original build as untouchable.

Step 10: Backtest the Process, Not Just the Winning Lineup

Post-slate review is where an MME process becomes repeatable. Looking only at the winning lineup creates hindsight bias. Review what your build actually did.

DFS Army has published weekly MME backtests using the Domination Station to review settings, exposures, stacks, and resulting lineups. Those articles are most valuable as examples of the review loop. Their exact percentages should not become permanent defaults.