Start/sit: a framework for 11:40am on Sunday

Twenty minutes before lock, two names, one lineup slot, and a browser with nine tabs of contradictory advice. Every fantasy manager knows this exact misery, and most handle it the same way: keep reading until some site sounds confident enough, start that guy, and blame the site by 4pm.

The spiral has a structure, and the structure has an exit. This is the decision order we built our start/sit assistant around; it works just as well executed by a human with a clock.

Step one: is there actually a decision here?

Check the projection gap first. When one player projects meaningfully ahead, say more than a couple points a week at the same position, the decision is already made and everything else you might read is noise shopping for a reason to overthink. A surprising fraction of Sunday agonizing happens over gaps that were never close; the fastest start/sit improvement available to most managers is simply trusting clear gaps.

Our assistant sorts every call into one of four leans: start, flex, toss-up, sit. The first job of that scale isn't the advice, it's the triage. Most decisions aren't close, and finding out fast is worth more than a deeper take on a settled question.

Step two: if it's close, pick the right kind of player

Now, and only now, context earns its keep. The question stops being "who's better" (the projections just told you: nobody, really) and becomes "which distribution does this week want?"

• Projected to win comfortably? Start the floor: the steady usage, the locked-in role, the player whose bad week is 8 points. Favorites lose to variance, so start as little of it as possible.
• Projected to lose? Start the ceiling, deliberately. A safe loss and a volatile loss count the same, but the volatile lineup occasionally steals the week. Underdogs who start their safest players are choosing a dignified defeat.
• No idea? Role beats matchup. Snap counts and target share persist week to week far more reliably than "tough defense" narratives, which are built on samples of five games and priced into everyone's head anyway.

Step three: respect the toss-up

Some calls are genuine coin flips, and here's the part almost nobody says out loud: a true toss-up means either choice is defensible and further deliberation has negative value. We label toss-ups as toss-ups instead of manufacturing fake confidence, because pretending 52/48 is a strong take is how tools train you to distrust them. When you hit one, pick the player whose profile fits the week (step two), close the tabs, and spend your remaining twenty minutes on something that matters.

What you're owed by any advice source, human or machine, is the reasoning and an honest confidence level. What you're not owed is certainty that doesn't exist.

The traps, quickly

Benching a stud after one bad week (his role didn't change; your mood did). Chasing last week's breakout into your lineup before the role is real. Overreacting to weather forecasts that mostly affect kickers. And the classic: making the bold call for its own sake in a week you were favored, which is starting variance exactly when it helps your opponent.

FAQ

How big does a projection gap need to be before I stop thinking? As a rough rule, a gap bigger than a couple of points at the same position is a settled question, and the closer you get to lock, the more that threshold should shrink. Deliberation has a cost too.

Should I start my studs no matter the matchup? Almost always. Elite players keep their roles against good defenses, and the bench-your-stud call goes wrong far more often than it goes right. The exceptions are injury-related usage limits, not matchup fear.

Is it wrong to just follow one site's rankings? Rankings are one input with default settings baked in. Your scoring, your week's win probability, and which kind of variance you need are things a generic list can't see, and they're exactly what the close calls turn on.