Skip to main content

Fantasy tools

Start / Sit Assistant

A start/sit decision is a binary comparison, not a ranked list. Choose two players and get a recommendation, a confidence read, and every factor that produced it — including the ones pointing the other way.

Window
Preseason
Players available
186
Factors weighed
6, summing to 100

It is the preseason, so this comparison runs on draft-season projection and positional rank. The matchup, recent-form, availability and schedule factors have no live games to read yet and are held neutral until the season starts — treat the verdict as a projection call, not a weekly matchup call.

Scoring
Try

Player A

De'Von Achane

Sit

RBRB7at BUF

Proj
13.6
Match
C+ matchup grade
L3
16.2
Trend
10Up 10 spots since last week

17.0 touch/g · 7.2 tgt/g

Player B

James Cook

Start

RBRB5vs MIA

Proj
14.3
Match
B matchup grade
L3
14.4
Trend
No change since last week

16.0 touch/g · 3.0 tgt/g

Close call

Close call — lean James Cook

The model separates these two by 6 points out of 100, which is inside its own noise band. James Cook gets the nod on matchup, but this is a decision you can make on roster construction or game-day news without giving much up.

Confidence
60Low
Margin
6 /100

Start

James Cook

14.3

Sit

De'Von Achane

13.6

Why

Favours CookFavours Achane

Six factors, weighted to sum to 100. Every one is shown, including the 2 that point the other way.

  1. Weekly projection

    32% of the model
    13.6 pts14.3 pts

    James Cook projects 0.7 points higher on the PPR board (14.3 to 13.6).

  2. Matchup

    18% of the model
    C+ · def #29B · def #12

    James Cook draws the better matchup: grade B against MIA, a defense ranked 12 of 32 in fantasy points allowed to the position, against grade C+ and a rank of 29.

  3. Opportunity

    20% of the model
    89th pct58th pct

    De'Von Achane commands more of his offense: 17.0 touch/g · 7.2 tgt/g, which is the 89th percentile at his position, against 16.0 touch/g · 3.0 tgt/g and the 58th percentile. Volume is the most repeatable input on this list.

  4. Recent form

    14% of the model
    16.2 ppg14.4 ppg

    De'Von Achane has been the better recent producer at 16.2 points per game over the last three and is trending down against his season average, while James Cook sits at 14.4 and is holding level.

  5. Availability

    10% of the model
    No designationNo designation

    Neither player carries a designation on the practice report.

  6. Schedule strength

    6% of the model
    6/104/10

    James Cook holds the friendlier rest-of-season schedule at 4 of 10 against 6. This matters more for a hold-or-drop call than for one week.

Before you set this lineup

60Low confidence, 60 out of 100
  • Projections here are illustrative local values, not the output of a trained or market-derived model.
  • Nothing in this build reads your league's scoring settings, roster construction or playoff picture.
  • The two projections are within a point of each other, which is well inside the error band of any weekly projection.

About this result: the recommendation is produced by a weighted comparison of illustrative local values — a projection board, a usage sample and a matchup grade that all ship with this build. It is not a trained model, not a market consensus, and it does not know your league.

Method

How The Recommendation Is Built

The weights are fixed and stated. You should be able to disagree with one input without discarding the whole answer — which is why every factor is shown separately.

How it reached that answer

  1. Both players are looked up on the same weekly board the rankings page reads, in the scoring format you selected.
  2. Six factors are scored independently: projection (32% of the model), opportunity (20%), matchup (18%), recent form (14%), availability (10%) and rest-of-season schedule (6%).
  3. Volume is converted to a percentile inside each player's own position group, which is what lets a running back and a receiver be compared fairly in a flex decision.
  4. Each factor produces an edge between −1 and 1. Multiplied by its weight and summed, that gives a margin out of 100 — the number shown beside the verdict.
  5. Confidence blends three separate things: the size of the margin, how much the six factors agree with each other, and how settled the underlying board already considered each player's role.

What it does not do

  • Read your league's scoring settings, roster construction, or who you are playing this week.
  • Account for weather, in-game script, Vegas totals, or anything that happens after the last practice report.
  • Use snap counts, route participation or injury news more recent than the sample data shipped with this build.
  • Claim an accuracy rate. There is no backtest behind these weights; they are a stated, inspectable set of assumptions rather than a fitted model.

This initial build ships with demonstration data. Projections, rostered percentages and movement are illustrative and will be replaced when a live data provider and editorial backend are connected.

Compare up to four

When the decision is not binary, put two to four players side by side across the same inputs.

Optimise the whole lineup

Set every slot at once instead of resolving one pair at a time, with an explanation per slot.

See the full board

The weekly rankings these projections come from, by position and scoring format.