Two half-point beats and one blowout PASS. That is the shape of Tuesday.
Start with the PASS because it validated the rule.
CON at IND was a HIGH-tier pick on Connecticut plus 11 per the picks file, edge plus 6.40. The framework filter flagged the model-versus-agent conflict: the 2025-net-ratings agent had three HIGH-tier quarter edges on the Indiana side (Q1, Q3, Q4) plus a "IND dominant game" same-game-parlay call. Historical read on those conflicts has been that the agent wins the disagreement. So the framework PASSed.
Result: Indiana 123, Connecticut 88. A 35-point Fever blowout at Gainbridge Fieldhouse. The bot's spread HIGH pick on Connecticut plus 11 would have lost by a country mile. If we had sized it at half stake per the default rule, that is negative 0.5u we did not need to spend.
Score one for the filter. This is the second time in about three weeks the model-versus-agent conflict rule has fired on a HIGH-tier pick that would have blown up spectacularly. The pattern that trips it — big HIGH-tier value on an underdog spread while the agent shows structural dominance across multiple quarters on the favorite side — keeps producing the same kind of ugly outcome. The rule stays in place with more confidence.
The two sized bets.
LVA at Washington. Spread LVA HIGH at half stake. Model had Vegas at minus 6.7 on the road; market had Washington slightly favored at home. Final: Washington 100, Las Vegas 99. Vegas lost by one point outright. That is a spread loss any way you slice it. Negative 0.5u.
The read that made this a HIGH pick was reasonable — Vegas coming off a 26-point win over Toronto with Wilson at full minutes, Washington bottom-quartile defensively, no injury complications flagged. The agent aligned with two HIGH-tier quarter edges on the LVA side. Nothing about the framework said PASS. Vegas simply did not execute on the road against a Mystics team that clearly played their best game in a stretch. Sometimes the HIGH pick is a bad beat and there is not a deeper lesson to learn from it.
CHI at NYL. Total UNDER 188.7 HIGH at half stake. Both stars out (Diggins for Chicago, Sabally for New York). Model total 173.68, agent projected total 174.0. Market at 188.7. Edge negative 15.02.
Final: 94-95, total 189. Missed the UNDER by half a point. Negative 0.5u.
Half-point beats on totals are the runningest theme of the season. The framework has been correctly on the right side of the number more often than not, but the WNBA's shortened pace environment produces games that hit right around the line all the time. Nothing to adjust in the process; the number was correctly modeled, the injuries were correctly flagged, the agent aligned. It just came in one point too high.
Net Tuesday: negative one unit on the two sized bets.
The compounding read.
The framework filter saved 0.5u on the CON PASS. The two sized bets combined to lose 1.0u. Net after filter benefit: negative 0.5u. The pattern that keeps showing up in July is that the framework rules are working — we are avoiding the blowouts, we are sizing the clean HIGHs at half stake per Issue 42 — but the variance on the sized bets has been rough.
Six of the last ten HIGH-tier picks we have sized have lost. Four have won. That is roughly the historical HIGH-tier hit rate (55-60 percent implied) with a bad six-game run baked in. Nothing to change in the process; the discipline is what protects the bankroll during stretches like this.
Cumulative check per the grader.
Spread STRONG LEAN (card tier, not picks file HIGH): 49-41-1 (54.4 percent, plus 3.90 percent ROI). Jumped from plus 0.93 percent yesterday. Three STRONG spread covers from underdogs Tuesday night — Portland plus 5, Seattle plus 10.5, Phoenix plus 2.3 all covered outright or clean. The spread tier continues to be the healthiest read in the system.
Total STRONG LEAN: 45-58 (43.7 percent, minus 16.59 percent ROI). Down from minus 16.23 percent. Two more total STRONG losses last night (CHI-NYL OVER, CON-IND OVER, DAL-POR OVER) offset by two wins (MIN-SEA UNDER, PHO-LAS UNDER). The trend line has not moved enough to trigger the quarter-stake downgrade rule I flagged (that needs another two percentage points of ROI decline).
Total MODEST LEAN: 27-28 (49.1 percent, minus 6.28 percent ROI). Consistent.
Spread MODEST LEAN: 22-24 (47.8 percent, minus 8.70 percent ROI). Slight drift down from yesterday.
Tonight. Zero games on the board. Second rest day in three days. Nothing to size, nothing to grade tomorrow morning from tonight.
The Tempo. Off tonight. Next game Friday at home against a TBD opponent (schedule verification pending). Jordan will pick it up when there is a game to write about.
Thursday's slate will populate overnight. Back tomorrow with whatever the bot serves up.
[ End Report ]
Share This