98. Toronto's score in Phoenix last night. Final TOR 98, PHO 90. The Tempo won outright as a 7.5-point road dog. The half-stake call on TOR plus 7.5 cleared by 15.5 points.
The discipline framework worked.
Yesterday morning I wrote that the conviction architecture required three confirmations: model has road team favored, agent has road team winning the most-relevant late quarters, home team has structural Q4 weakness. TOR at PHO had condition (1) but not (2) or (3). The agent report had zero quarter edges, zero prop edges, zero SGP correlations. Only the model's spread edge.
The framework rule for partial architectural confirmation is half stake. We took the half stake. We won. The half stake makes 0.45 units instead of 0.91 units, but the discipline that made it a half stake instead of a full stake is the same discipline that saved us from sizing 0.91 units down on the WAS at DAL loss the night before. The framework optimizes win rate over time, not single-night outcomes.
Two ways to think about last night, same framing I have used the last two days.
The pessimist read. We left half a unit on the table by sizing down on a winner. The framework was too conservative.
The optimist read. The framework correctly sized down because the architectural confirmation was incomplete. The win happened anyway because Toronto is a better team than the line was pricing, but that is fortunate variance on top of correct sizing. If the framework had said "size up on any model edge plus 7" with no architectural check, we would also be sizing up on the next 7-point edge that has the architectural confirmation missing, and that one might lose. The win rate compounds; the unit return per win does not.
I lean optimist. The framework is producing 67 percent spread win rate on STRONG-tier picks (10-5 through May 19). That is well above the breakeven of 52.4 percent at -110 standard juice. The unit return per pick can be smaller; the win rate keeps the bankroll growing.
Tonight is small. Three games on the board.
Portland at Indiana. STRONG LEAN POR plus 12.8 per the card. Model edge plus 7.9. Same profile as TOR at PHO last night.
The architectural framework. (1) Model has road team favored at neutral — confirmed. The model has POR at minus 0.14 at neutral (essentially pick em, but POR slightly favored). The market has IND favored by 12.8. That is a massive gap and the largest spread divergence on the slate. (2) Agent has road team winning the most-relevant late quarters — NOT confirmed. The agent report on IND vs POR has zero quarter edges, zero prop edges, zero SGP ideas. (3) Home team has structural Q4 weakness — NOT confirmed. Indiana ranked 11th in clutch in 2025 but is not as structurally broken as Dallas was. They beat Seattle by 11 on Sunday.
One of three confirmations. Same as last night. Same call.
Half stake on POR plus 12.8 at IND. Per the framework.
The injury report adds a little context. Indiana has Aliyah Boston listed Day-To-Day, Makayla Timpson Day-To-Day, Justine Pissott Out, Bree Hall Out. The Fever's frontcourt depth is being tested. Portland has just Karlie Samuelson out. The injury picture favors POR more than the market is pricing. That is part of why the model edge is so large (the calibration includes the injury report).
If Boston is downgraded to Out before tipoff, the size could shift to three-quarter stake. The decision tree is clear and the data will tell us before tipoff. Worth checking the live injury feed at /transactions in the next two hours.
The other two games tonight.
Chicago at Dallas. MODEST DAL minus 1.5 spread, edge plus 3.1. Pass per tier downgrade. Dallas is hot after the blowout win over Washington Monday. CHI is decimated (Westbeld, Carrington, Stevens all Out per the live feed; Diggins and Jackson Day-To-Day). The MODEST signal is real but not worth chasing.
Connecticut at Seattle. PASS spread (model edge minus 2.35, no significant edge). MODEST UNDER 166.1 edge minus 4.76. Pass per tier.
Tonight's recommended action.
HALF STAKE: POR plus 12.8 at Indiana. Same architectural profile as TOR at PHO last night. Same call.
PASS: CHI at DAL spread (MODEST tier, not worth chasing).
PASS: CON at SEA spread (no significant edge).
PASS: CHI at DAL total (MODEST UNDER, tier downgrade).
PASS: CON at SEA total (MODEST UNDER, tier downgrade).
One bet tonight at half stake.
The framework is producing. We do not need to chase action.
Cumulative through May 19. Spread STRONG: 10-5 (67 percent). Total STRONG: 8-6 (57 percent). Combined STRONG: 18-11 (62 percent). The model is profitable on both tiers and the spread ledger is at its season high in win rate.
The conviction architecture stands at 5-for-6 (TOR at PHO last night does not count as a conviction call because it only had one of three confirmations; the architecture ledger only counts full-confirmation applications). When the architecture next has all three confirmations, we will size full stake without hesitation. Tonight is not that night.
Talk tomorrow morning. Tempo is home Friday. We are in the back half of week three of the season and the framework is working.
[ End Report ]
Share This