The Rule Went 3-3. Conflict Rule Fires Twice Tonight.

3. The number of consecutive PASS calls the framework discipline produced that missed winning bets last night.

SEA at LVA. The bot's HIGH SEA spread at plus 14.5 won. Las Vegas won 101 to 91. Seattle covered the spread by 4.5 points. We PASSED per the partial-confirmation discipline. Missed approximately 0.45 units of profit.

NYL at CON. The card had STRONG LEAN on CON plus 13. The conflict rule fired (three HIGH agent signals on New York). We PASSED per the rule. Connecticut plus 13 covered (final NYL 89, CON 80, NYL won by 9, single digits). Missed approximately 0.45 units of profit.

IND at WAS. The card had MODEST LEAN UNDER 169.9. The tier-downgrade rule's PASS-by-default applied. Final 78-76, game total 154, under the line by 16. Missed approximately 0.45 units of profit.

Combined missed profit across three correct PASS calls: roughly 1.35 units at the half-stake size. The discipline framework had its first negative-units day since the framework's silent-agent filter pattern started extending across the week.

The conflict rule's record is now 3-3.

The rule has saved spread variance three times (CON at ATL by 16 points on June 2, IND at NYL by 8 points on June 6, GSV at NYL by 27 points on May 27). The rule has missed winning bets three times (NYL at CON yesterday, plus two earlier from the rule's history). Net unit-count change across the six firings is approximately flat — the saves and the misses are roughly equivalent.

That is exactly what a 3-3 record should produce. The rule is functioning as a filter that prevents the bot's HIGH on losing-side spreads while passing on the bot's HIGH on winning-side spreads when the agent reads the opposite. The structural logic of the rule (catch the roster-mismatch shape the recent-form 2026 model misses) remains correct. The unit-count flatness is the rule operating as a coin-flip on the upside as well as the downside.

The threshold review for the rule needs to happen by Sunday. The three-HIGH-on-opposite-side trigger is the structural baseline. The two-HIGH-plus-SGP variant flagged last week after the WAS at ATL blowout would have produced a different PASS call. The data argues for a tighter rule on the upside (sized half stake when the rule does not fire) and a relaxed rule on the downside (sized PASS when the rule fires at 2-HIGH-plus-SGP).

Tonight the rule fires twice.

DAL at MIN. The bot's picks file has Dallas as a HIGH spread call at plus 9.65 edge. The agent has four HIGH-confidence quarter signals on Minnesota plus a "MIN dominant game" SGP correlation. Four-on-zero is the cleanest configuration of the rule since the May 27 GSV-NYL firing. Nina has the tactical piece with the specific film on Dallas's Q3-Q4 weakness and Collier's absence affecting the agent's read.

ATL at CHI. The bot's picks file has Chicago as a HIGH spread call at edge minus 5.2 against the home dog at plus 9. The card model has the opposite read at STRONG LEAN on Atlanta minus 7.5 (favored to give 7.5). The agent has three HIGH-confidence quarter signals on Atlanta (Q1 plus 4.80, Q2 plus 4.00, Q3 plus 6.60) plus a Q4 medium and the same agent player props pointing at Atlanta's perimeter creators (Howard and Gray OVER props on the spread side).

This is the first case where the picks file and the card model disagree on the side AND the agent confirms the card model. The picks file is alone on CHI. The card and the agent agree on ATL. The conflict rule applies because the picks file's spread_pick (CHI) is on the opposite side of the agent's three HIGH signals (ATL).

PASS both per the rule.

PHO at GSV. Card has STRONG LEAN on Phoenix plus 8.2 with edge 11.0 against the home favorite. The picks file does not have this game in HIGH tier. The agent has only one medium quarter signal (PHO Q4 medium plus 1.60) and a "PHO closes" SGP. The architecture does not fire (Golden State is the home favorite by the model). This is the partial-confirmation pattern that has been the framework discipline's silent-agent PASS shape across the week.

The pattern was correct seven times and wrong once (yesterday's SEA at LVA). The rule's record at silent-agent partial confirmation is 7-1 on the spread side and 0-4 on the under side (four STRONG UNDERS we PASSED all won at the bot's grader level). The spread-side pattern still dominates. The under-side pattern needs the threshold review the rule book has been tracking.

PASS the PHO spread per the partial-confirmation discipline. PASS the total (card has it at PASS already).

Three games tonight. Three PASS calls. The discipline streak goes to nine STRONG-tier reads correctly applied at the rule level (the unit-count side will not resolve until the games play out).

The cumulative ledger.

Spread STRONG editorial-sized: 12-13 entering tonight. No bets sized in twelve days. The silent-agent filter saves on the spread side now total approximately 5 units across nine PASS calls (including one missed-win cost of 0.45 units yesterday on SEA). Net positive variance avoided is roughly 4.5 units against the partial-confirmation half-stake baseline.

Total STRONG editorial-sized: 9-9. No bets sized in two weeks. The under-side conservatism missed approximately 1.8 units of profit across four winning STRONG UNDERS this week plus the IND-WAS MODEST UNDER yesterday adding another 0.45 units. The under-tier drag now totals about 2.25 units missed.

Combined net across the silent-agent filter applied to both markets: approximately plus 2.25 units variance saved on the spread side minus 2.25 units missed on the under side equals zero net change to the ledger. The discipline neither helped nor hurt this week at the integrated level.

That is the threshold-review evidence point. A rule that nets to zero across nine applications is functioning as a coin flip. The structural argument for the rule (catch the bot's HIGH spreads that lose, avoid sized losses) still applies on the spread side. The structural argument on the under side (PASS without pace confirmation) is producing PASS calls on winning bets at a 4-of-5 rate. The under-side rule needs adjustment.

The architecture's standing remains 5-0 on non-Dallas applications, 2-3 on Dallas. No architecture firings tonight.

The plan tomorrow.

Wednesday's slate is small per the morning data pull. The new commits show CON at TOR on the Wednesday slate and a couple of Thursday games. CON at TOR is the Tempo home game Jordan will preview tomorrow morning. The conflict rule's threshold review continues across the rest of the week.

Talk tomorrow.

[ End Report ]