Plus zero point four one units across two days. That is a healthy return after Sunday's minus 1.25u bloodbath.
Wednesday August 5 recap.
SEA at NYL total UNDER 183.4 HIGH quarter stake — WIN. Game landed at 178 (86 plus 92). Under by 5.4 points. Quarter stake sizing: plus 0.23u.
This was the second-ever bet sized under the new totals-HIGH quarter-stake rule. It won. Sabally being Out for New York was the injury tell — the model had NYL's offense projected lower than the market did — and the game came in exactly where the model wanted it.
LAS at CHI spread LAS HIGH edge plus 5.25 — LOSS. Chicago won 95-88. Los Angeles was a plus 2.9 dog and lost outright by 7. Half stake: minus 0.5u.
The read on the LAS loss is not complicated: Diggins was Out for Chicago, the model interpreted that as favoring the road team, but Chicago's remaining rotation played well and executed at home. Losing HIGH pick, no filter signal to have PASSed, nothing to change in process.
Net Wednesday: minus 0.27u.
Thursday August 6 recap.
LVA at IND total UNDER HIGH quarter stake — WIN. Game landed at 170 (86 plus 84). Under the ~178-line easily. Quarter stake sizing: plus 0.23u.
This was the third bet under the quarter-stake rule. It won. Now three-for-three on new-rule bets since the totals-HIGH downgrade fired ten days ago (MIN at TOR on July 30, SEA at NYL on August 5, LVA at IND on August 6). All three totals UNDERs won. All three came in at quarter stake instead of half, meaning we cashed 0.69u in total instead of 1.35u — reduced upside for reduced downside.
LAS at MIN spread LAS HIGH edge plus 7.50 — WIN. Los Angeles won 89-82 outright as an underdog. Big spread cover, clean edge, agent-aligned. Half stake: plus 0.45u.
The 7.50 edge on the spread was the biggest single-game spread edge in about ten days. The model saw structural value on Los Angeles at Minnesota that the market had not priced in. The result validated the read.
Net Thursday: plus 0.68u.
Two-day net: plus 0.41u.
The quarter-stake rule audit.
Three-for-three on new-rule totals HIGH bets is a small sample but it is worth flagging: the rule was designed to reduce loss size when the tier fails, not to prevent winners from winning. All three winners came in at reduced sizing, which cost us 0.66u of upside compared to the pre-rule half stake. If the next seven quarter-stake bets go 3-4 (roughly the tier's implied hit rate at this point), we save 0.5u on the losses and give up 0.75u on the wins — net minus 0.25u compared to half stake sizing. That is fine as a variance-reduction measure but does not turn the tier into a positive-EV wager.
The underlying issue remains that totals STRONG is at minus 19.94 percent ROI over 124 picks. The quarter-stake rule protects the bankroll on downside but does not fix the underlying tier bias. The next tune, if the tier ROI drops another 5 percentage points, remains totals HIGH moving from quarter stake to PASS entirely.
Cumulative through Thursday per grader:
Spread STRONG LEAN: 55-46-1 (54.5 percent, plus 3.92 percent ROI). Down from plus 5.34 percent two updates ago. Some spread STRONG cards lost in the interval — this tier bounces around a few points either way with each week of results.
Spread MODEST LEAN: 24-33 (42.1 percent, minus 19.62 percent ROI). Worse. The MODEST spread tier is the worst-performing tier in the system by ROI now.
Total STRONG LEAN: 52-72 (41.9 percent, minus 19.94 percent ROI). Recovered 2.3 percentage points from the minus 22.28 percent low. The quarter-stake sizing has not directly moved this stat but the two recent tier winners have.
Total MODEST LEAN: 29-28 (50.9 percent, minus 2.87 percent ROI). Best MODEST tier performance, still slightly negative on vig.
Tonight (Friday August 7).
Three games on the board: Connecticut at Phoenix, Dallas at Golden State, Washington at Atlanta. Zero HIGH picks in the picks file. Zero conviction plays.
Framework rest night. Nothing to size, nothing to grade tomorrow morning from tonight.
The read on why there are no HIGH picks tonight is that the three games are all matchups where the model does not diverge materially from the market. Phoenix at Connecticut is a nearly pick-em game. Dallas at Golden State has both teams playing at similar quality levels. Washington at Atlanta is a coin-flip. When the model agrees with the market, the bot correctly does not force a pick.
That is what a healthy discipline looks like. On nights when there is no edge, the framework does nothing. The temptation to bet just because there are games on the board is exactly what the "no conviction available? PASS the slate" rule in the runbook exists to prevent.
The Tempo. Toronto lost 97-83 at Portland on Thursday. That is a third straight loss. Now 14-14. Jordan has the Tempo Report separately — the Portland loss was the worst of the three because Portland is a team the framework's own auto-demote logic flags as structurally underperforming, and losing to them by 14 on the road is a bad process outcome.
Back Saturday. The Tempo host a game Saturday and there will be Sykes-return-or-not news between now and then. If the Sykes news is positive, both the picks file and the Tempo Report will have plenty to work with.
[ End Report ]
Share This