Reading the card
How to read a start/sit card
A Decide Football start/sit card is a stack, not a slogan. Listed status comes first. The mean is the ranking number. Floor and ceiling are a model range around that mean. Certainty is a grade of how hard the math can lean, not a chance of winning the week.
This page is how to read that stack. How the numbers are built is the methodology. What the product is, and is not, is the about page.
Listed status, before any number
OUT, IR, and INACTIVE are not starts. The model sets that row to zero. It does not invent a return date, a practice window, or a sentence about when someone might be back. Questionable and doubtful apply a discount. They are not a cleared player, and they are not a ruled-out player. The designation is an input someone listed. The projection does not vote a new one into existence.
If the status and the mean disagree, believe the status. A healthy-looking mean on a player listed OUT is a zero for the week. The card is telling you the gate fired.
Mean is the ranking number
Start/sit and the weekly rankings sort on the mean. That is the single estimate the desk uses to order a call. Floor and ceiling sit beside it as a range: the low end and the high end, given the sample and the status the model was handed. A wider range means the sample is thin or the status is dirty.
The range is not a promise of points and not an official projection. Treat the mean as the rank. Treat floor and ceiling as the width around it.
Floor or ceiling, depending on the week you have
The mean ranks the pair. Which end of the range you weigh depends on the fantasy week in front of you.
If you are ahead in your matchup and a crater would give the week away, weigh the floor. You are asking which side the model leaves less room to fall apart. If you are behind and a modest mean will not catch the gap, weigh the ceiling. You are asking which side the model leaves more room above the mean.
That is roster reading. It is not a spread, a moneyline, a win probability, or advice on a wager. Nothing on this desk is a gambling product.
Certainty is a desk grade
The card also prints a certainty score, from 0 to 96, with a band: thin, lean, clear, or strong. On the card those bands read as a thin edge, a soft lean, a clear lean, or a strong call. The score is how much this desk thinks the call can lean on the math in front of it. A wider mean gap raises it. A toss-up keeps it thin. High uncertainty, a status discount, or an availability zero cuts it.
- Thin. The math is a thin edge. Do not treat the higher mean as settled.
- Lean. A soft lean. The mean prefers a side, and the inputs are not clean enough to carry the week alone.
- Clear. The call can lean on the math. Still read the status labels beside it.
- Strong. The numbers can carry more of the call. Still not a win probability, and still not a reason to ignore an OUT tag.
It is not the probability that you win the fantasy week, and it is not the probability that the higher mean outscores the other side on Sunday. Read the band next to the status labels, not instead of them. The arithmetic behind the score is on the methodology page, labeled v0 and subject to change.
A narrow mean is soft evidence
The desk draws a toss-up line at 1.5 estimated points. Inside that gap the card says to lean neither side on mean alone, and certainty stays thin. Outside it, the higher mean is a lean. Role and listed status still matter. A tenth of a point is not a verdict. A slightly higher mean on a smaller role does not, by itself, jump the player whose job is the one you are actually starting.
What this is not
- Not affiliated with, endorsed by, or sponsored by the NFL or its member clubs. Names are identification for fantasy analysis only.
- Not an official projection, club report, or injury wire.
- Not gambling advice, odds, or a sportsbook.
- Not a backtest. The method is methodology v0, subject to change. No accuracy rate is published here, and none should be inferred.
Where the comparison cards are
Comparison tools are at the start/sit desk so the product can be used and reviewed. Those URLs use sample data. They stay out of search until licensed sports data is in place. This page does not set a date for that, and it does not promise those URLs will enter a search index.
The public description of the desk is this guide, the methodology, and the about page.
Questions
What do floor, mean, and ceiling mean here?
The mean is the ranking number: the estimate start/sit and weekly rankings sort on. Floor and ceiling are the model range around that mean, a low end and a high end. They are not a promise of points and not an official projection. A wider range means a thinner sample or a dirtier listed status. OUT, IR, and INACTIVE force the row to zero. The model does not invent a return date.
What is the certainty score not?
Certainty is a desk grade of how much the call can lean on the math. The bands are thin, lean, clear, and strong. It is not a probability of winning the fantasy week, not a betting line, and not a guarantee that the higher mean wins. Read it next to listed status, not instead of it.
Why are the sample start/sit URLs not in search?
Comparison pages on the start/sit desk are on the site so the product can be reviewed, and they use sample data. They stay out of search until licensed sports data is in place. This guide does not promise that those URLs will be indexed, and it does not give a date.
If the means are close, does the higher one win the start?
No. A narrow mean gap is soft evidence. Under 1.5 estimated points the desk treats the pair as a toss-up and keeps certainty thin. Role and listed status still matter. The higher mean is a lean, not a verdict.
When should I weigh the floor instead of the ceiling?
When you are protecting a lead in your fantasy matchup, weigh the floor: you care which side the model leaves less room to fall apart. When you need upside to catch up, weigh the ceiling. That is how to read the range for the week you have. It is not odds, a spread, or gambling advice.