Weekly Recap: AI Picks Went 34-58 (37% Win Rate) for Aug 3 to Aug 9, 2026

Week in Review The AI closed the week of Aug 3 through Aug 9 at 34 wins, 58 losses, and 2 pushes, a 37.0% win rate across 94 graded MLB picks. It was the roughest stretch since the middle of July, and the honest read is that the model spent five straight days fighting bad run line prices in a week where favorites kept losing. Sunday finally broke the pattern with a 7-1 day, and the way it happened points to a real fix rather than a lucky bounce. Results by League | League | Record | Win Rate | |--------|--------|----------| | MLB | 34-58-2P | 37.0% | MLB was the only board in play. Baseball is carrying the full slate until football odds open up later this month. Results by Bet Type | Type | Record | Win Rate | |------|--------|----------| | Game Lines | 33-53 | 38.4% | | Player Props | 1-5-2P | 16.7% | Game lines did most of the volume and most of the damage. Player props stayed a small sample and still came back ugly, so prop confidence needs another look before that volume grows. Digging into the game lines | Market | Record | Win Rate | |--------|--------|----------| | Moneyline | 9-13 | 40.9% | | Run Line / Spread | 18-28 | 39.1% | | Totals (Over/Under) | 7-15-2P | 31.8% | Totals were the weakest market on the board. Fifteen losses on 24 decided over/under picks is not variance, it's a pricing problem, and it lines up with a week of scoring swings that landed nowhere near the posted numbers. Day by Day | Day | Record | Win Rate | |-----|--------|----------| | Mon, Aug 3 | 3-11-1P | 21% | | Tue, Aug 4 | 5-9-1P | 36% | | Wed, Aug 5 | 4-11 | 27% | | Thu, Aug 6 | 6-6 | 50% | | Fri, Aug 7 | 4-9 | 31% | | Sat, Aug 8 | 5-11 | 31% | | Sun, Aug 9 | 7-1 | 88% | The Confidence Problem This is the most useful thing in the data, so it's worth putting front and center. Confidence scores ran backwards all week. | Confidence Band | Record | Win Rate | |-----------------|--------|----------| | 80% and up | 8-16 | 33% | | 70% to 79% | 6-14 | 30% | | 60% to 69% | 10-24-2P | 29% | | Under 60% | 10-4 | 71% | > The picks the model liked least won at 71%. The picks it liked most won at 33%. That inversion is not supposed to happen, and it tells you exactly where the leak is. High confidence in this system almost always means a heavy favorite on the run line, and heavy favorites getting laid at 1.5 runs is one of the most expensive habits in baseball betting. The low confidence bucket was full of plus 1.5 underdogs, and those cashed. Top Picks of the Week Seattle Mariners minus 1.5 vs Detroit, Aug 5. Rated at 82% and it delivered a clean 4-2 cover at plus 148. Best price on the board all week and the one high confidence run line that actually paid what it should. San Diego Padres moneyline at Arizona, Aug 4. Also 82%, cashed at plus 121 in a 9-4 win. The model had the Padres bat playing better than the market did. Total Under 9.0, Washington at Philadelphia, Aug 4. Philadelphia won 5-0 and the game finished five runs under the number. One of only seven total wins on the week, and it came with the run line win in the same game. Notable Misses Los Angeles Dodgers moneyline at Chicago, Aug 3. An 82% rating on the biggest name in the league, and the Cubs won 10-5. The model bet the Dodgers three separate ways in that game and lost all three. Toronto Blue Jays moneyline at Houston, Aug 4. Another 82% pick, another loss, 7-2. Houston and Toronto were on both sides of the model's high confidence picks in consecutive days, which is a sign the inputs were unstable rather than sharp. Philadelphia Phillies minus 1.5 vs Washington, Aug 5. The day after the Phillies delivered a 5-0 shutout, the model went back to the same run line at 82% and Washington won 10-4. Recency in the inputs is pulling too hard. What Sunday Showed Sunday's 7-1 was not a hot streak. Every one of those eight picks came in under 60% confidence, and seven of the eight were plus 1.5 underdog run lines. Atlanta, Arizona, the Athletics, Washington, Baltimore, and Toronto all covered, and three of those teams covered while losing the game outright. That's the whole point of the plus 1.5 side. You get paid on close games instead of needing blowouts. The only Sunday loss was Milwaukee at minus 1.5, a favorite, in a game they won 4-3. Same pattern as the rest of the week. Looking Ahead Two changes are on deck for the coming week. First, the confidence model gets recalibrated so a heavy favorite on the run line stops earning an automatic 80 plus rating. Second, totals volume gets cut back until the number-setting logic is retested, since 31.8% on over/unders is the single biggest hole in the ledger. The full week by week ledger stays public either way, wins and losses both. --- Related reading - Weekly Recap: AI Picks Went 51-64 for Jul 27 to Aug 2 - The Red Sox Went 21-4 in July and Just Beat the Best Team in Baseball - The White Sox Lead the AL Central, and That Breaks the Whole Trade Deadline