Floor vs ceiling: when boring wins

Take two receivers who finish with identical season totals. One delivers 11 to 15 points every single week. The other alternates 25-point eruptions with 3-point disappearances. Same total, same end-of-year rank, and if you treat them as the same player you'll be wrong in both directions: starting the volatile one in the week you needed safety, benching the steady one in the week you needed a miracle.

Season totals average away the thing that actually decides fantasy weeks. Here's how we measure it, and when each profile genuinely wins.

How we measure floor

Our player metrics include a consistency score built from weekly game logs, adjusted to weight the floor side of a player's distribution, then expressed as a percentile within his position (a 90 means steadier week-to-week than 90% of players at the spot). Two design choices worth knowing:

• It's percentile-ranked within position, because positions have different natural variance. A "volatile" running back and a volatile tight end are different animals, and comparing raw variance across positions mostly measures the position.
• It requires a real sample. Below a handful of played weeks we don't pretend to know; in contexts where floor matters most, our draft engine charges a small flat penalty for that unknown rather than guessing, because variance you can't price is still variance.

An honest limitation: consistency is computed from a completed season's weeks, so it describes an established pattern, not a guarantee. Roles change, and the metric follows with a lag.

When floor wins

Guillotine leagues, first and loudest. You're eliminated for the lowest weekly score, so a zero-week is fatal and ceiling above "not last" buys nothing. This is the one format where our engine prices floor directly into draft rankings: up to a 10% bump for the steadiest profiles, up to 15% off for boom-bust, with the penalty deliberately bigger than the bonus.

Elsewhere, floor wins whenever you're the favorite. Projected ahead in a head-to-head week, your job is to not lose, and every point of variance you start is a gift to the underdog. It also wins at fragile roster spots: the lineup slot where a dud cascades (your only healthy RB2, a must-win week in a tight playoff race) wants the 11-to-15 guy, not the lottery ticket.

When ceiling wins

Flip the matchup math. When you're the underdog, variance is your friend; a safe 120 loses to a favorite's 135 exactly as hard as a volatile 95 does, but the volatile lineup sometimes prints 150. Playoff weeks against stronger seeds, best ball formats where the software auto-starts your best scores, and any week the projections say you need to get lucky: those are ceiling weeks, and the boring receiver is quietly the wrong pick.

The discipline is matching the profile to the week on purpose. Most managers start their "best" players every week and never ask which kind of best the week requires.

FAQ

Is consistency a skill or just noise? Some of each. Role stability (target share, snap count, red-zone usage) is real and persistent; game-script luck is noise. That's why we build the metric from a full season of weeks and rank within position, to keep as much signal and as little noise as a public stat allows.

Should I draft for floor or ceiling? In guillotine, floor, and our engine enforces it. Everywhere else, draft mostly on value and use floor versus ceiling as the tiebreaker at close calls, then manage it week to week with your start/sit decisions.

Does a small sample make a player risky even if he looked great? Priced honestly, yes. A five-week wonder might be a new permanent role or a blip, and until the sample settles you're holding unpriced variance. Charge for it mentally the way our engine does, a modest penalty, not a disqualification.