Every streak ends as a question. Crown climbed for 7 straight shows — 82.1 in early July to 93.075 in Atlanta, the steepest sustained rise in the top tier — and then gave back -0.65 in Winston-Salem. Third place both nights; nothing about the placement changed. What changed is that the trend line, the entire load-bearing wall of the Crown-is-coming argument, finally printed a point going the wrong way.

The honest answer to “does it matter” is: it depends on which autopsy you trust. The caption sheet says no. The model says maybe a lot. They can’t both be right, and there’s a date on the calendar — Saturday, August 1 — when one of them stops being able to hide.

Total drop: -0.65. GE Visual alone: -0.20. Strip that one caption out and Crown’s Sunday sheet beats Saturday’s — GE Music up, both visual proficiency captions steady or up, and percussion — the caption that has dragged below the rest of their sheet all month — jumping -0.10 to a season high. A show that regressed doesn’t post its best drum number of the summer on the same night. A show that ran into one chair with a different opinion does.

Saturday, Crown’s 18.9 tied Bluecoats for the best GE Visual in the building. Twenty-four hours later a different judge had them 0.8 below their own Saturday number and 0.8 behind Bluecoats. Note the pattern: Sunday’s chair held the leader flat and marked both chasers down — Blue Devils lost 0.35 on the same sheet. Atlanta’s judge saw a three-way race in this caption; Winston-Salem’s saw a hierarchy. Same shows, one day apart. That re-spread is a room’s taste, not a design collapsing — the standard cross-panel error bar this desk puts on every weekend split. The noise case, in full: one caption, one chair, one night, against a sheet that otherwise improved.

The model’s logic is brutal and consistent: it fits trajectories, and a first-ever down point at the exact moment rivals keep climbing doesn’t just lower the current score — it flattens the slope every projection is built on. Hence medal odds falling from 36.2% to 27.2% on a night Crown finished third. The strangest artifact is the Boston row: Crown sits 0.925 ahead of Boston Crusaders on the season board and beat them head-to-head Saturday — their first win in 7 meetings — yet the model gives Boston 51.0% to Crown’s 27.2%. It is betting, in plain terms, that Saturday was Crown’s ceiling and Sunday was the correction. That is a falsifiable bet.

Six meetings, six Boston wins, each one narrower — then Atlanta, where the line finally crossed: Crown by 0.7. One night later Crown dipped with Boston nowhere on the property, and the model quietly moved Boston back ahead in every projection that matters. Which means Saturday’s result — the one actual head-to-head data point — is currently being outvoted by a curve fit. Either the sheet was right in Atlanta or the model is right now. Allentown adjudicates.

The two shows before it — Myrtle Beach and Salem, against fields Crown outscores by ten — will produce big pretty numbers that answer nothing; treat them as scoreboard hygiene. Allentown is the real instrument. The desk’s thresholds, stated in advance so we can be graded: if Crown beats Boston again on the shared sheet, the dip was one chair’s taste, the streak resumes, and the model’s 27.2% is too pessimistic by half. If Boston flips Atlanta’s 0.7 back the other way, then Sunday wasn’t noise — it was the first public frame of a slowdown the model detected before any human wanted to say it out loud. Either way, this desk prints the answer the morning after.