Plus 0.63 units across three days. That is the return-from-break performance for the framework.
The three-day breakdown.
Thursday July 30: three sized picks, net plus 0.18u. - CON at CHI CON HIGH half stake — loss (CHI won 94-88), minus 0.5u. - MIN at TOR UNDER 188.1 HIGH quarter stake (first bet under new totals rule) — win (final 176), plus 0.23u. - NYL at LVA NYL HIGH half stake — win (NYL covered plus 5.6 as LVA won 104-99), plus 0.45u.
Friday July 31: one sized pick, plus 0.45u. - SEA at ATL SEA HIGH half stake — win (SEA won 98-89), plus 0.45u.
Saturday August 1: no HIGH picks. Rest.
Three-day framework net: plus 0.63u.
The read.
The MIN-TOR UNDER pick is the important one to call out. That was the first pick sized under the new totals-HIGH-to-quarter-stake rule. It won. The game landed at 176 total against a 188.1 line — well UNDER, no half-point drama, clean result.
The old sizing would have been half stake and produced plus 0.45u instead of plus 0.23u. The new sizing produced plus 0.23u. That is a meaningful reduction in the upside on winning picks, but it is the trade-off for the downside protection when the tier keeps producing losing HIGHs. On a losing bet, the difference is minus 0.25u instead of minus 0.5u — the same 0.22u swing.
The framework rule is not designed to maximize upside on winning bets. It is designed to reduce exposure to a tier that has demonstrated structural underperformance across a 100-plus pick sample. Thursday's winner does not change the tier logic; the tier remains at quarter stake until the ROI recovers above minus 15 percent.
Cumulative check per grader (through Friday):
Spread STRONG LEAN: 53-43-1 (55.2 percent, plus 5.34 percent ROI). Up from plus 4.64 percent pre-break. Continues to be the tier that holds its edge.
Spread MODEST LEAN: 23-29 (44.2 percent, minus 15.56 percent ROI). Worse. The MODEST tier does not survive the vig at coin-flip win rates.
Total STRONG LEAN: 46-66 (41.1 percent, minus 21.59 percent ROI). Another 2.9 percentage points worse. If this pattern continues, the next rule tune is to move totals HIGH from quarter stake to PASS entirely. I would set that trigger at either an ROI below minus 25 percent or 10 consecutive losing totals HIGH picks, whichever comes first.
Total MODEST LEAN: 28-28 (50.0 percent, minus 4.55 percent ROI). Only tier better than break-even by wins but drags negative on vig.
Tonight (Sunday August 2). Four games on the board, two HIGH picks.
Pick one: IND at MIN. DOUBLE HIGH.
Spread IND HIGH edge plus 5.80. Model has this near pick-em, market has MIN as a 5.8-point favorite at home. No star injuries either side.
Total UNDER 193.2 HIGH edge minus 13.35. Model total 179.85, agent projected total 187.6 — both under the market's 193.2. Agent leans UNDER less confidently than the model, but same direction.
Agent quarter markets: one HIGH edge on MIN Q2 (plus 3.40), the rest MEDIUM. Not enough to trigger the 3-HIGH conflict PASS rule on the spread.
Sizing: - Spread IND HIGH: HALF stake per Issue 42. 0.5u. - Total UNDER HIGH: QUARTER stake per new rule. 0.25u. - Total on this game: 0.75u
The double HIGH means the framework has real conviction on this being the wrong direction from the market — MIN is being overvalued at home while the total is being over-projected. Two independent market signals from the same model. If both hit, this is the game that anchors the night.
Pick two: CON at DAL. CON HIGH edge plus 6.75. Griner Out for Connecticut.
Model has Dallas favored by 4.65; market has Dallas favored by 11.4. That is a 6.75-point edge favoring Connecticut on the spread.
Agent has TWO HIGH-tier quarter edges on the Dallas side: DAL Q3 plus 5.00 and DAL Q4 plus 4.70. Both late-quarter edges. This is the exact pattern that has fired the model-versus-agent PASS filter in the past — the picks file goes one way and the agent shows structural dominance in the closing quarters.
Two HIGH agent edges is one short of the formal 3-HIGH PASS trigger. And there is no SGP idea on this game to trigger the "2 HIGH plus SGP equals PASS" rule I proposed in the mid-season report. So by the formal rule, this stays at HALF stake.
But I want to flag: the last time the framework had this exact pattern (2 HIGH late-quarter agent edges against a picks-file HIGH pick) was CON at IND on July 22, where I formally sized the pick before the framework filter fired via three-HIGH-conflict on the third quarter edge that appeared later in review. Indiana won by 35.
My read tonight: **CON at DAL sizes at HALF stake by formal rule, but I am flagging this as the same pattern.** If the framework loses this bet in the way I expect it might (Dallas dominates the second half), that is data suggesting the 2-HIGH-late-quarter trigger should become a formal PASS rule in the next Maya piece.
Sizing: HALF stake CON. 0.5u.
Total exposure tonight: 1.25u across three bets on two games.
The Tempo. Toronto at Golden State tonight after Thursday's 32-point home loss to Minnesota. Not on the picks file — model SKIP. Jordan has the Tempo Report with the full read on what changed between Tuesday's road win and Thursday's home blowout, and what tonight has to look like.
Back tomorrow with the grades. The IND-MIN double HIGH and the CON-DAL late-quarter-agent-conflict flag are the two things I will be watching most closely.
[ End Report ]
Share This