0.55. The units lost across Tuesday and Wednesday's editorial-sized slates. Three half-stake bets total. One winner (PHO plus 6.1 at Indiana, won outright Tuesday). Two losers (POR plus 5.65 at Chicago Tuesday by 23 points, DAL plus 8.15 at Las Vegas Wednesday by 7 points to the line). Net minus 0.55 units. The kind of two-day stretch the loosened framework is built to absorb — half-stake variance bleeds slowly, does not blow up the bankroll.
The Tuesday miss on POR was the bigger one in unit terms (a 23-point cover failure is the worst kind of loss because the bet was wrong from the first quarter on). The DAL loss Wednesday was tighter — DAL covered the first half but Las Vegas pulled away in the third quarter.
The framework reading after the two slates.
The half-stake default on bot HIGH picks went 1-2 over the two-day window. Sample of three is meaningless. The loosened framework needs 15-20 picks to evaluate. Tuesday and Wednesday together produced three observations. The framework continues per spec until the sample is large enough to recalibrate.
Cumulative per the grader through Wednesday.
Spread STRONG (all bot-labeled STRONG LEAN cards): 31-29 (51.7 percent, minus 1.36 percent ROI). Total STRONG: 30-30 (50 percent, minus 4.55 percent ROI). Combined STRONG: 61-59 (50.8 percent). Essentially break-even after 120-plus observations.
The bot's HIGH-confidence-published-picks subset (the n_picks list specifically): roughly 55-56 percent over the season per spot-checks of the grader's results. The slight edge the bot's HIGH picks have over the broader STRONG LEAN label is the entire framework's expected value contribution. The half-stake default rule is the right way to size that edge — small enough that misses do not hurt much, big enough that wins compound over a long season.
Now tonight.
Three games on the slate. The bot's picks file has one HIGH-confidence call.
Portland at Chicago. POR HIGH plus 5.30, edge plus 5.30. Model has POR favored by approximately 0.5 points at neutral. The market has CHI favored by 5.30 at home.
This is the second straight POR-at-CHI matchup the bot has elevated to HIGH. The previous one on Tuesday lost by 23 points. The bot's model is telling us tonight's matchup is essentially identical in shape — same teams, same venue, same general edge size.
Agent check. The agent has CHI Q2 plus 3.00 HIGH (Chicago answers strongly in the second quarter at home) and POR Q1 plus 2.20 MEDIUM (Portland opens with a small advantage). One HIGH signal favoring the home team, no HIGH signals favoring the picked road team. That is one HIGH conflict — not the 3-plus threshold the loosened rule from Issue 42 set for PASS. The framework default is half stake.
Half stake on POR plus 5.30 at Chicago. Yes, 48 hours after the same matchup lost by 23. The framework does not get to refuse the bet because it remembers Tuesday. The discipline is sample-size based, not memory based.
The framework discomfort. Sometimes the bot model is wrong about a matchup for two straight cards because the underlying calibration is off (Portland has been worse than the model thinks; Chicago has been better at home than the model thinks). Sometimes the bot model is right about a matchup and Tuesday's result was a 23-point variance outcome. The framework cannot distinguish between those two cases on a sample of one. The way to distinguish is to keep sizing per the rule, accumulate observations, and let the larger sample tell the story.
The other two games tonight.
Golden State at Atlanta. Picks file shows no HIGH call. Agent has four HIGH quarter signals (GSV Q1 plus 3.80, GSV Q2 plus 3.70, ATL Q3 plus 2.90, ATL Q4 plus 5.90). Two HIGH on each side with ATL Q4 the largest single-quarter edge. The architecture leans Atlanta at home but no bot HIGH pick. PASS per the framework — only bet HIGH picks at half stake by default; non-HIGH cards stay PASS unless conviction architecture clearly fires.
Connecticut at Washington. Picks file shows no HIGH call. Agent has WAS Q4 plus 2.80 HIGH only — one signal, MODEST tier. PASS per tier downgrade.
Tonight's recommended action.
HALF STAKE: POR plus 5.30 at Chicago. Default rule applies. Yes, on the same matchup that lost Tuesday.
PASS: GSV at ATL (no HIGH published pick).
PASS: CON at WAS (no HIGH published pick, MODEST signals only).
Total exposure roughly 0.5 units (one half stake). Expected value if the half-stake default holds at 55-56 percent is plus 0.05-0.08 units. A small slate by design.
The other story today is Jordan's recap of the Tempo's Wednesday blowout of Las Vegas (final 125-97). The team won by 28 at home as 4.5-point favorites despite missing Sykes and Rice. The piece is the right kind of franchise news after a stretch of bumpy editorial. Read it.
Talk tomorrow morning. The framework recalibration window keeps running.
[ End Report ]
Share This