Winning in the Age of AI

An optimizer takes your projections and salary constraints and returns the mathematically highest-projected lineup. A simulator runs the slate tens of thousands of times — using your projections, ceiling distributions, and projected field ownership — to evaluate which plays and combinations perform best across a wide range of outcomes.

Simulators help answer questions that used to be impossible: Is Blake Corum a good play at 60% ownership with Kyren Williams out? Should a Justin Herbert stack include Ladd McConkey and Quentin Johnston, or does one of them have to go? Does that stack need a bring-back, and at what ownership level does it stop making sense?

Before sims, you made those calls on instinct and experience. Now you can test them. That matters.

But neither tool works without good inputs. And that's where most players get the relationship backwards.

Simulators are built on your projections. The slate is simulated based on your projected points, your ceiling outputs, your ownership estimates. If those inputs are wrong, every simulation built on top of them is wrong too. You can run a million iterations on a bad projection and get a very confident, very wrong answer.

The same is true for optimizers. Put bad projections in, get bad lineups out — optimized perfectly to a flawed foundation.

This is why projections are still the most important thing in DFS. Not the tools that consume them. The tools amplify whatever is in the inputs. Sharp inputs get amplified into an edge. Weak inputs get amplified into losses at scale.

Every analytical tool in this space — optimizer, sim, ownership model — is downstream of projection quality.

What makes a projection system genuinely useful isn't the math alone. It's what goes into the math.

Strong projection systems balance analytics, market expectations, matchup context, and football intelligence. That structure keeps you honest — you don't want to be blindly overconfident when the market knows something you don't.

But the projections also have to reflect football intelligence: defensive matchup reads, utilization tendencies, game plan context, personnel packages, situational usage patterns. A number that doesn't account for how a cornerback is deployed, or how an offensive coordinator manages late-game leads, has a hole in it.

Analytics and football intelligence aren't in competition. The best projections combine them — and the best DFS players apply both when evaluating what the model is telling them.

Here's how the decision tree actually works once your projections are set.

You can use sims to help build lineups. They'll give you direction. They'll surface combinations the field is underweighting. They'll tell you whether a high-ownership play still makes sense given your projected exposure on that game. That's all real value — use it.

But you can also do something the sims can't: look at your projections and identify where the industry is wrong.

Maybe the field sees a backfield as an even split — but your film study and usage tracking tell you one back is about to absorb the workload. The industry hasn't moved there yet.

The sim doesn't know what you know.

It's running on consensus ownership projections like everyone else. When you find a spot where you're genuinely right and the field is wrong, you don't need a sim to validate the play. You need lineups structured to maximize what you win when your read is correct. That's a different question — and a harder one — than what the simulation recommends.

Finding those spots, building the right structure, and playing it with conviction when the evidence is strong: that's the highest edge available in DFS. Not because it's secret. Because it's hard, and most players won't do the work to get there.

Putting it together — a few principles that hold across any slate:

Projections lead, tools follow. Your football intelligence read on the slate has to be built in before you touch the optimizer or run a sim. Weak projections make every downstream tool worse.

Use sims to pressure-test, not decide. Let your read lead. Let the tool challenge it.

Know your differentiation before you lock. Every lineup needs a clear answer to: where am I gaining leverage, and why? Lineups that are different by accident aren't differentiated. They're just random.

Build around a scenario. The best tournament lineups reflect a specific view of how the slate plays out. When you can articulate what has to happen for your lineup to win, you've built something. When you can't, you've assembled players.

The players who win consistently in large-field tournaments aren't the ones running the most simulations. They're the ones who understand what the tools can and can't do, feed them better inputs than the field, and still trust their own read when the evidence points somewhere the consensus hasn't gone.

Optimizers and sims are tools to help you build better lineups.

They are not answers to the test.

The test is still yours to take.

See the game differently.