Streamers

One half is the only judgement anywhere in the arithmetic, and it is here rather than buried: it means half a point of matchup is worth half a point of projection. That is enough to reorder two candidates you already thought were interchangeable, which is the decision a streamer actually is, and not enough to lift a replacement-level player above a genuine starter because he drew a soft afternoon. If you disagree with the weight, the two numbers behind it are printed on every row so you can weigh them yourself.

Where the numbers come from

The points-allowed figure is the defense-versus-position board: every completed stat line from every game a defense has played, kept for the position you are looking at, scored in your format, divided by the games that defense has played. There is no model in it and no adjustment. The projection is the consensus board, which is the median of several independent forecasters with our own model deliberately left out of the vote. The week and the fixtures come from the scoreboard, which knows which clubs meet and when. Change the scoring chip and both halves are recomputed, so you are reading your own league's format rather than someone else's relabelled, with the one exception named below: a defense's projected total arrives already scored and cannot be recomputed.

The weekly kicker and defense projections come from ESPN's published fantasy projections, read once a day on our own schedule and stored, never fetched while you are on the page. Every row carries how many independent forecasters stand behind it and when it was last published, and the board prints the age of the oldest row it is showing rather than the freshest, so "updated an hour ago" is true of everything on screen and not only of the best line on it.

Kicker and defense are projected by the week, not by the season

Quarterback and tight end get the full formula, because what exists for them is a season forecast: it says how good the player is and nothing at all about who he plays, which is precisely the gap the matchup edge fills. Kicker and defense are different. Both now carry a projection made for the coming week, one fixture at a time, taken from ESPN's published fantasy projections. A forecast built for one game already has the opponent inside it, so adding the matchup edge on top would count that opponent twice. It is not added. The matchup figure is still worked out and still printed beside the number, because it is useful to see, but on those two boards it carries no weight in the order and the row says which number moved it.

Kicker figures remain the noisiest on the site by a distance, projection or not, because how often a kicker scores depends on how often his own offense stalls inside range rather than on the defense opposite him. Treat the kicker board as a tiebreak between two names you already like.

The defense number is the source's own projected total in the source's own default scoring, and that is stated rather than smoothed over. This site scores passing, rushing, receiving and kicking from raw components, so it can recompute any of those into your format; it has no rule for turning a sack or a shutout into a point, so it has nothing to recompute a defense with. Inventing one to make the number look like ours would be worse than naming whose it is. That figure therefore does not move when you change the scoring chip, and the kicker figure does. Both are labelled on the row.

When no defense projection has been published for the coming week, the board falls back to what it did before, and says so. A defense is then rated by the offense opposite it: the consensus per-game projections of that club's quarterback, backs, receivers and tight ends, added up. Less offense opposite is a better week, so the edge is inverted for that board alone. It is a computed number with a unit and a source rather than an opinion, and it is printed on the row, but it is a proxy: it knows nothing about how often an offense turns the ball over, gives up sacks or goes three and out, which is where a large share of real defensive fantasy points come from. It also counts only the players the consensus covers, so the number of players behind each figure is shown beside it, and the count line names it as the ranking for as long as it is one.

Early in a season, half this board is last season

Three games is not a rate. Through the opening weeks one outlier afternoon moves a defense fifteen places, so any defense with fewer than four games this season shows last season's figure instead, tagged on the row, with the game count travelling beside it. That means that for a few weeks in September and October a single ranking is genuinely built on two different measurements, and the rows that use the older one say so. The alternative, ranking only on the current season, would put every defense in an essentially arbitrary order for a month, which is worse and quieter about being worse.

When we do not know who plays whom

If the coming week's fixtures are not published, this page does not guess. Every row loses its opponent and its matchup figure, the ranking falls back to the projection alone, the band at the top says so in plain words, and a second panel appears listing the defenses ranked by points allowed across the season so you can pair them with your own roster by hand. A schedule reconstructed by counting games played is a guess dressed as a fact, and this page would rather show you less.

The league filter

Paste your Sleeper league id and the board drops everyone already rostered in that league, which is the difference between a list of good streamers and a list of streamers you can actually add. The league is read once while the tab is open and nothing about it is stored anywhere. Team defenses are matched by club, players by identity, and if any roster entry cannot be identified with confidence the page says how many rather than quietly leaving them on the board. No account is involved and no permission is asked for: a Sleeper league id is public, and reading it needs nothing else.

What this page will not tell you

Points allowed is not defensive quality. A defense whose own offense scores quickly faces more passing attempts and will allow more receiving points for reasons that have nothing to do with its secondary. A defense that has already played three of the best offenses in the league sits near the top of this board and may be excellent. It is also a team-level number, and the matchup you care about is a player-level one: a defense can be generous to tight ends as a group while erasing the specific tight end you were about to start. Nothing here knows who is healthy, who is suspended, whether a starting quarterback left in the first quarter or whether the fourth quarter was backups against backups. Every one of those sits in these averages looking like a normal afternoon.

Use it at the edges. The two or three names at the top of a position are worth a start you were unsure about; the middle of any of these boards is inside the noise of a fantasy week and should not move a lineup on its own. Pair it with the start/sit comparator when the decision is between two named players, with the defense versus position board when you want all thirty-two defenses rather than the ones on this week's slate, and with the injury report before you act on any of it.