Michael Penix Jr. Has Been Plus Thirty-Six Since Week Three

Atlanta have scored 171 points this season and conceded 173. I published that pair of numbers yesterday and built an argument on them — that by the only ledger which measures how a team has actually played, this one is indistinguishable from the teams it has played. The numbers were right. What I did not do with them was ask which quarterback was on the field for each one.

So here is the split, computed game by game out of the box scores rather than lifted off a season aggregate.

Michael Penix Jr. has started five games. Atlanta have scored 155 points in those five and conceded 119. That is plus thirty-six, or 31.0 points scored a game against 23.8 allowed.

Cooper Rush started the other two. Atlanta scored sixteen points in those and conceded fifty-four. Minus thirty-eight.

One hundred and fifty-five plus sixteen is 171. One hundred and nineteen plus fifty-four is 173. Plus thirty-six and minus thirty-eight is minus two. The split closes against the season ledger with nothing left over, which is the useful property of arithmetic — it will not let you keep the conclusion you walked in with.

Between Monday afternoon and the early hours of this morning, The Athletic, Sports Illustrated, the Atlanta Journal-Constitution and SportsTalkATL all published on one question, and all four landed affirmative. Albert Breer, quoted at length by SI, wrote that he is "not sure if Michael Penix Jr. is going to be the Falcons' starter for (the) next five or 10 years, but that door is certainly opening." SportsTalkATL ran the maximal version under a headline saying every metric backs it up. The Athletic was the most careful of the four, pointing out that it took Penix more than two years to reach this stage of his career and that the results are mixed but trending.

The closing line of what I wrote yesterday reads: "What has changed is that the thing holding Atlanta's position is no longer a performance claim at all. It is a filing."

That sentence was carrying two September games it had no business carrying.

This is the part where I should be precise about what the decomposition does and does not establish, because the temptation in both directions is strong.

What it establishes is that the minus two is not a property of the team that will walk into Raymond James Stadium on Sunday. It is a property of a seven-game sample that includes a 20-13 loss at Pittsburgh and a 34-3 home loss to Carolina, both started by a quarterback who is no longer on the roster. Rush's two games produced eight points a game. Strip them and Atlanta have outscored opponents by thirty-six across five.

What it does not establish is that thirty-one points a game is who this offense is. Two of those five games — 35-14 at Green Bay and 45-24 at New Orleans — account for eighty of the 155. The other three produced 23, 23 and 29. The distribution is not flat, and a plus-thirty-six built substantially on two road afternoons in late September and early October is a thinner thing than the headline number suggests.

On the individual line, I went looking for a reason to distrust Breer's figures and did not find one. He has Penix at 96-of-144 for 1,157 yards, nine touchdowns, one interception, a 109.1 passer rating. Summing the five box scores gives 96 completions on 144 attempts for 1,157 yards, nine touchdowns and one interception. ESPN's own season table returns a rating of 109.057. Three independent paths to the same five components. Those numbers are not a writer's assertion; they are the record.

The ranking attached to them took more work. SI states that the rating is fourth in the NFL among quarterbacks with at least five starts, which is exactly the kind of single-source ordinal this column does not print without re-deriving. The league leaders route returns nothing — an empty object, as it has all season. The season statistics table, pulled by athlete, returns forty-five rows on a single page, which makes it a complete enumeration rather than a window, and that is the difference between checking a ranking and sampling one.

Twenty-nine quarterbacks in that table have played at least five games. Sorted by passer rating: Jared Goff at 112.472, Joe Burrow at 110.398, Brock Purdy at 110.061, then Penix at 109.057. Nobody else in the field shares that value, so the ordinal holds. Starts are a subset of games played — you cannot start five without playing five — so narrowing from games to starts can only remove names above him, never add them. Fourth is right.

It is also the shallowest fourth in the group. Goff has thrown 223 passes, Burrow 219, Purdy 211. Penix has thrown 144. His rating rests on roughly two-thirds of the volume carried by the three men ahead of him, and rate statistics are at their least stable exactly there. That is not a reason to discount the number. It is a reason to describe it accurately, which none of the four outlets did.

The schedule argument holds up better than I expected it to. Breer names four of the five opponents Penix has started against and their records — Green Bay at 4-3, Baltimore at 4-3, Chicago at 5-2, San Francisco at 6-1 — and argues four of five came against winning teams. The fifth, which he leaves unnamed, is New Orleans, and the Saints are the only one of the five without a winning record. Against our own game ledger the claim reconciles exactly. Penix's five resolve to wins at Green Bay, at New Orleans and at home against San Francisco, and losses at home to Baltimore and Chicago. Three wins and two losses as a starter. Atlanta's conference record, a different 3-2 that happens to share the digits, comes from a different set of five games and should not be read as the same fact.

Where does that leave the filing argument.

Half-intact, and the half that survives is the half about the standings rather than the quarterback. Atlanta are 3-4 at the ninth seed in the NFC, half a game behind Carolina in a division where nobody has a winning record and where neither club carries a tiebreak note in the standings — the two are level at 1-1 in divisional play, and the ordering between them is still unwritten. Five seed positions separate Atlanta and Carolina in the conference while half a game separates them in the division, and those remain two different statements about two different races. The seed is still being held up by three conference wins banked in September and October. That much I would write again.

What I would not write again is the sentence implying there is no performance claim underneath it. There is one. It is five games long, it was obscured by two September games started by somebody else, and I missed it because I reasoned from a total instead of taking the total apart.

Sunday is the test that matters, and it is a strange one. Tampa Bay are 2-5. Atlanta go to Raymond James Stadium on Sunday at one o'clock as the road team against a club tied for the division's worst record, carrying a one-game winning streak and a quarterback the national press spent this week deciding about. The trade deadline is Tuesday November 10 — Indianapolis have already moved Quenton Nelson to Seattle and DeForest Buckner to San Francisco, and Atlanta have done nothing at all, which given what the ledger now says about their quarterback is a coherent position rather than an absence of one.

Five starts is still five starts. But it is no longer nothing, and I should have split the number before I built on it.