This week
Michael is hosting BAL. We project 251.1 passing yards, 12 rushing yards. The number comes out of a drive-level simulation of the game rather than a season average — it asks how many possessions his side is likely to get, what they do with them, and how much of that flows through him.
How much he's on the field
86% — that's Michael's average snap share over 16 games. That's a full-time role. Snap share at this level is the most stable input we have — it changes slowly, and it survives a bad game in a way that yardage never does. This is the number to check first on any football page. Everything else here is downstream of it.
What kind of quarterback he is
Michael has completed 260 of 426 (61%) for 3236 yards, 14 touchdowns and 7 interceptions — 7.6 yards an attempt. Yards per attempt is the number to watch: completion percentage rewards checkdowns, and a quarterback can complete 70% of his passes without ever threatening anybody.
His last 5 games
Over that stretch Michael has 1075 total yards. A handful of games is thin evidence in a seventeen-game season. It nudges the projection; it doesn't drive it.
How the model reads him
Michael's projection starts with the game, not with him. We simulate the matchup drive by drive — how many possessions each side gets, how far they travel, how often they score — and then ask what share of that flows through him given his role. That ordering matters: a productive player on a team that runs 55 plays a game has a lower ceiling than a lesser one on a team that runs 70.
What we'd flag
Read this knowing the limits. A seventeen-game season makes every rate noisy; game script can invert a player's usage inside one afternoon; and depth-chart changes often surface too late to be in the projection.