Major League Baseball predictions for Friday, August 21, 2026
Model win probabilities and confidence tiers for every Major League Baseball game on Friday, August 21, 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
- winPittsburgh Pirates at Los Angeles DodgersModel call: Los Angeles Dodgers (70% confidence, Strong) · Final 4-5
- winSan Francisco Giants at Boston Red SoxModel call: Boston Red Sox (70% confidence, Strong) · Final 4-6
- winSt. Louis Cardinals at Philadelphia PhilliesModel call: Philadelphia Phillies (69% confidence, Moderate) · Final 6-7
- winAtlanta Braves at Milwaukee BrewersModel call: Milwaukee Brewers (69% confidence, Moderate) · Final 1-2
- winLos Angeles Angels at Texas RangersModel call: Texas Rangers (69% confidence, Moderate) · Final 1-2
- winMinnesota Twins at San Diego PadresModel call: San Diego Padres (69% confidence, Moderate) · Final 2-6
- winWashington Nationals at Miami MarlinsModel call: Miami Marlins (69% confidence, Moderate) · Final 2-3
- winAthletics at Houston AstrosModel call: Houston Astros (69% confidence, Moderate) · Final 0-4
- lossCleveland Guardians at Colorado RockiesNo published call, model passed on this game · Final 9-1
- winTampa Bay Rays at Baltimore OriolesNo published call, model passed on this game · Final 3-5
- winChicago Cubs at Seattle MarinersNo published call, model passed on this game · Final 5-6
- winDetroit Tigers at Kansas City RoyalsNo published call, model passed on this game · Final 2-5
- winNew York Mets at Chicago White SoxNo published call, model passed on this game · Final 4-6
- winCincinnati Reds at Arizona DiamondbacksNo published call, model passed on this game · Final 0-9
- winToronto Blue Jays at New York YankeesNo published call, model passed on this game · Final 1-3
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