Denver Broncos at San Francisco 49ers

These are the market's numbers, recorded as the feed saw them. They are shown so the model can be compared against something, not as a recommendation.

Each row is a snapshot the feed recorded, oldest first. Superseded rows are kept, because a line that moves and moves back is only visible if both are shown.

Issued 80 hours before kickoff. A forecast issued further out has had less time to settle.

Stadium WP did not list elevation; using Santa Clara city elevation (72 ft).

The model projects each side, then adjusts for the conditions the game is played in. Both steps are shown so the adjustment can be seen rather than taken on trust. These add up exactly as printed; if they ever did not, this section would not appear.

Wind and cold are recorded as an effect on passing efficiency, not as points, and the conversion between the two is not published, so they are shown in their own units rather than added into the chain above. Indoors and elevation are recorded in points.

Two earlier links, per-play efficiency and the matchup adjustment, are computed and discarded rather than stored, so they cannot be shown yet.

live-20261002T023447Z, generated Thu Oct 1, 10:35 PM ET

The model projects each team's line first, then divides it among players. Both are shown, with the team figure beside the sum of its parts, so the division can be checked rather than trusted. A row appears only when the parts add up to the whole.

Share is a player's portion of the team's projected passing yards, counting a quarterback by what he throws and everyone else by what he catches. Targets are shown but not totalled: a throwaway, a batted ball or a sack is a pass attempt with no target, which is about four a game per team, so targets legitimately do not sum to attempts.

live-20261002T023447Z