Major League Baseball predictions for Saturday, August 22, 2026
Model win probabilities and confidence tiers for every Major League Baseball game on Saturday, August 22, 2026, published before first pitch, tip or kick and graded automatically afterwards. Research and analytics only, not betting advice.
Major League Baseball matchups and model calls
- winSan Francisco Giants at Boston Red SoxModel call: Boston Red Sox (70% confidence, Strong) · Final 2-3
- lossToronto Blue Jays at New York YankeesModel call: New York Yankees (69% confidence, Moderate) · Final 4-3
- lossCincinnati Reds at Arizona DiamondbacksModel call: Arizona Diamondbacks (69% confidence, Moderate) · Final 11-5
- winWashington Nationals at Miami MarlinsModel call: Miami Marlins (69% confidence, Moderate) · Final 2-4
- lossLos Angeles Angels at Texas RangersModel call: Texas Rangers (69% confidence, Moderate) · Final 3-0
- winMinnesota Twins at San Diego PadresModel call: San Diego Padres (69% confidence, Moderate) · Final 5-7
- winSt. Louis Cardinals at Philadelphia PhilliesModel call: Philadelphia Phillies (69% confidence, Moderate) · Final 3-12
- lossAthletics at Houston AstrosModel call: Houston Astros (69% confidence, Moderate) · Final 4-3
- lossNew York Mets at Chicago White SoxModel call: Chicago White Sox (69% confidence, Moderate) · Final 10-5
- winPittsburgh Pirates at Los Angeles DodgersNo published call, model passed on this game · Final 3-4
- winTampa Bay Rays at Baltimore OriolesNo published call, model passed on this game · Final 2-3
- winChicago Cubs at Seattle MarinersNo published call, model passed on this game · Final 4-5
- lossCleveland Guardians at Colorado RockiesNo published call, model passed on this game · Final 4-3
- winDetroit Tigers at Kansas City RoyalsNo published call, model passed on this game · Final 1-3
- winAtlanta Braves at Milwaukee BrewersNo published call, model passed on this game · Final 1-4
How these Major League Baseball projections are built
Each matchup runs through an ensemble of opponent-adjusted efficiency, recent form, availability of key players, rest and travel, and sport-specific factors. The blended output is calibrated against past results before a confidence tier is assigned, and games below the league publish floor are recorded as a pass instead of a call.
What calibration meansReading the confidence tiers
Lean is a slight model preference, Moderate a clear one, and Strong is reserved for calls clearing both the confidence and edge floors for Major League Baseball. Every tier is graded separately in the public ledger, so you can check whether Strong calls actually outperform Lean calls over a real sample.
Confidence tiers explained