24. The margin Minnesota beat Dallas by last night. The conflict rule's biggest single-game vindication of the season.
The bot's picks file had Dallas as a HIGH spread call with edge plus 9.65 points. The agent had four HIGH-confidence quarter signals on Minnesota plus a "Minnesota dominant game" SGP correlation. The framework filter said PASS per the conflict rule. Final score Minnesota 100, Dallas 76. Minnesota by 24. Dallas at minus 4.5 against the spread lost by 19.5 points beyond the line.
That is the conflict rule's single largest unit save this season. A full-stake bet on the bot's HIGH DAL would have lost 1 unit. A half-stake bet at the partial-confirmation default sizing would have lost 0.5 units. The PASS saved 1 unit at the full-stake equivalent.
The rule's record moves to 4-3 across seven firings.
The other firing yesterday went the other direction.
ATL at CHI was the rule's second firing yesterday. The bot's picks file had CHI as a HIGH spread call at home dog plus 9 points. The card model had STRONG LEAN on Atlanta minus 7.5 favored. The agent had three HIGH-confidence quarter signals on Atlanta's side. The conflict configuration applied to the picks-file-versus-agent dimension only, not to the card-versus-agent dimension (the card aligned with the agent on Atlanta).
We PASSED per the rule because the picks file's CHI pick is the formal trigger. Final score Atlanta 82, Chicago 75. Atlanta won by 7 outright but did not cover the card's STRONG LEAN spread of minus 7.5 (need to win by 8-plus to cover). The card model lost. The bot's HIGH CHI plus 9 covered (CHI lost by only 7 against a 9-point cushion). Our PASS missed the bot's CHI win at approximately 0.45 units.
Net for the rule yesterday: plus 1 unit on DAL-MIN minus 0.45 units on ATL-CHI equals plus 0.55 units saved at the half-stake equivalent.
The rule's structural logic is what worked yesterday on DAL-MIN. The four-HIGH agent reads on Minnesota caught a roster-mismatch shape the bot's 2026 model missed. Minnesota without Collier was still the team that had the structural identity to dominate Dallas at home, and the agent's 2025 net rating basis correctly identified that pattern. The DAL spread bet that the bot's model produced was the kind of recent-form weighting failure the rule was designed to catch.
The ATL-CHI miss was structurally less clean. The agent had three HIGH signals on Atlanta, which is the rule's threshold but at the lowest configuration. The card aligned with the agent. The bot's pick aligned against both. The bet that paid was the bot's side (CHI cover), which was the side both the card and the agent had against. The rule's PASS missed the bot's correct call.
The rule's record by configuration: four-HIGH-or-better firings are now 3-1 (Saturday's IND at NYL was the recent loss on that side, otherwise the four-HIGH category is 3-0). Three-HIGH firings are 1-3 (CON at ATL on June 2 was the only correct PASS in that category). The threshold review evidence points toward keeping the four-HIGH-or-better trigger and either relaxing the three-HIGH trigger to a half-stake size or moving the three-HIGH-plus-SGP requirement.
This week is the threshold-review week.
Tonight is small.
Two games on the WNBA slate tonight. The bot's picks file has zero HIGH-confidence calls. The picks file's MODEST tier and SKIP tier are both empty across both games, suggesting the bot's primary model does not see structural edge on either matchup.
CON at TOR. The card has the spread at MODEST LEAN on Connecticut plus 7.5. Toronto is the home favorite for the first time in three weeks. Connecticut is missing Van Lith and Morrow with Griner Day-To-Day. The agent has zero quarter spread edges and only one player prop (Tina Charles UNDER 16.3 on fatigue). The total is MODEST LEAN UNDER 169.4. Both markets are MODEST tier. PASS by tier on both per the discipline rule that MODEST tier reads do not get sized.
LAS at SEA. The card has the spread at STRONG LEAN on Seattle plus 6.5 with edge 7.4 points. The agent has one HIGH-confidence quarter signal on Seattle Q3 plus 3.30 and a medium signal on LAS Q4 plus 1.50, with a "Close game, LAS closes" SGP correlation. The picks file does not have this game in HIGH tier. The architecture does not fire (Seattle is home, not road favorite at neutral). This is the partial-confirmation pattern with one agent HIGH signal supporting the card's STRONG LEAN.
Nina has the tactical piece on the LAS-SEA partial confirmation. The discipline reading is PASS — the partial-confirmation tier has been the season's worst-drag tier and the discipline pattern this week has been to PASS at partial-confirmation alignment.
Two PASS calls. Zero exposure tonight. The discipline streak goes to ten consecutive STRONG-tier reads correctly filtered at the rule level.
The cumulative ledger update.
Spread STRONG editorial-sized: 12-13 entering tonight. No bets sized in thirteen days. The silent-agent filter saves on the spread side now total approximately 5.5 units across ten correct PASSes (including yesterday's plus 1 from DAL-MIN minus 0.45 from ATL-CHI). Net positive variance avoided remains substantial.
Total STRONG editorial-sized: 9-9. No bets sized. The under-side conservatism continues to miss profits — last night the MODEST UNDER on DAL-MIN was at 172.5 and the game total was 176, so the UNDER lost regardless of size. That was one missed-loss data point in the threshold review.
Combined net across the silent-agent filter: approximately plus 3.5 units saved across the two weeks of the pattern. The rule's structural logic continues to work on the spread side. The under-side threshold review remains the rule-maintenance task for the weekend.
The architecture's standing.
5-0 on non-Dallas applications. 2-3 on Dallas. No architecture firings tonight.
Tomorrow.
Three games on the Thursday slate per the morning data pull. LVA at POR, NYL at ATL, PHO at DAL. NYL at ATL has potential to be another conflict-rule candidate — Atlanta's home identity with Brionna Jones Out is the kind of roster configuration the agent's 2025 net ratings basis tends to misread, and New York with or without Ionescu is a tier-one team the bot may overvalue as a road favorite. The agent report will land tomorrow morning.
PHO at DAL is the Dallas exclusion rule's next test if Phoenix is a road favorite by the model. The Dallas exclusion has been the framework's other active discipline filter all week.
LVA at POR is a Vegas-Pattern candidate if Las Vegas's injury picture is noisy. Carter and Loyd have been on the report for two weeks. The morning pull will tell us whether the pattern fires.
Talk tomorrow.
[ End Report ]
Share This