133 games graded, 133 settled. No unresolved rows, no disputed calls, no scoreboard ambiguity to sort through on Sunday morning. That's a clean sheet administratively, which is the least interesting thing about a college football Saturday but also the most useful — a full slate means a full grade, and the record reflects it honestly.

The shape of the week was fairly conventional: no tracked upsets surfaced in the data, and the one matchup that drew our attention was a conference game with a real edge reading behind it. We'll get to that. The card stayed empty for the second time this fall, which is worth naming plainly before moving on.

No card plays were journaled for Week 3's Saturday slate. No edge cleared both floors required to commit a staked position, so the budget stayed put. An empty card is a decision, not an oversight — and it goes into the record the same way a staked loss would.

There is nothing to grade here, and nothing to hide in that fact.

The tracked leans went 2-0 on the week. Both entries point to the same game: Ole Miss Rebels over the visiting LSU Tigers, Saturday, September 19.

The blend came in at 62% for Ole Miss against a market sitting at 54% — an 8.1 percentage-point edge. The spread had the Rebels as 2.5-point home favorites with a total of 57.5. Final score: Ole Miss 32, LSU 24. The Rebels covered and won outright, so both lean entries resolve as hits.

A few things worth saying plainly: the data brief carries this game twice under the leans list, and we're reporting it as such rather than quietly collapsing it into one line. The underlying read was the same both times — blend at 62%, market at 54%, 8.1-point edge. One result, two graded rows, both marked hit. That's the record as it exists.

The final also came in under the 57.5 total (32 + 24 = 56), which wasn't a tracked lean but is worth noting for context. The edge the model saw was on the side, and the side landed. Leans are tracked and graded in public; no money was on either entry.

566 games graded on the season across all tracked weeks. The blend model's calibrated record covers 125 of those games where we have the full probability set: log-loss of 0.3662, Brier score of 0.1139, accuracy of 84.8%.

For comparison, the market baseline over the same 125-game window: log-loss 0.3488, Brier 0.1093, accuracy 88.0%. The market is outperforming the blend on every logged metric at this point in the season. That's the honest read of 125 games, and 125 games — while a real sample — is still early enough that the gap could reflect noise, calibration lag, or genuine market efficiency the model hasn't closed. We're not going to dress that up.

The blend leans on the market at a documented weight, and that's visible in how the two track together. What the blend adds, when it adds anything, is the divergence — places where the model's read and the market's read separate meaningfully. This week's lean was one of those. The record will tell us, over time, whether those divergences are signal or coincidence.

The Ole Miss-LSU lean hit cleanly, and it's worth understanding why that matters and why it doesn't. One result confirms the model read the situation right this Saturday. It doesn't confirm the model will read the next one right, or that 8-point blend-versus-market edges are reliably exploitable. Small samples are noise — the brief carries that note for a reason, and we're still well within the range where a clean weekend and a rough one can look interchangeable in the long-run log.

The more durable takeaway from Week 3 is process-level: the card stayed empty because nothing cleared both floors, the leans were tracked and graded transparently, and the market is still ahead of the blend on calibration. Those three facts together describe where we actually are, which is more useful than celebrating a 2-0 lean week or worrying about an empty card.

The record is public, it's forming in real time, and we'll keep building it the same way next Saturday. Leans are tracked information, not instructions — the card is the staked record, and right now the card has nothing on it. That's the position going into Week 4.