MLB Picks Today: Predictions & Best Bets for September 25, 2026 | SharpSide

Friday’s 17-game slate calls for separating disagreement with the market from confidence in the winner. Our three featured matchups are Texas Rangers @ Minnesota Twins, with a 1.6 runs SIDE edge toward Minnesota Twins, tied for the day’s largest; Baltimore Orioles @ New York Yankees; and New York Mets @ Washington Nationals. This brief refreshes at 11:30 am Eastern based on new analytics, weather factors, pitching changes, and line movement, so featured matchups may change between the morning publish and the refresh. Today’s Best Bet is highlighted immediately below.

Today's Best Bet: Toronto Blue Jays ML

Today’s Best Bet is Toronto Blue Jays ML (-137) against the Cincinnati Reds, the single game and line SharpSide has the highest confidence in today. The model projects Cincinnati–Toronto at 3.1–4.4, a Toronto victory with a 1.3-run predicted margin. That is a clear projected advantage, and it is roughly in line with the market’s implied margin: both model and market favor Toronto. This is a conviction play, not necessarily our biggest pricing edge, and agreement with consensus does not disqualify the game the model is most sure who will win. Best Bet: Toronto Blue Jays ML (-137)

Texas Rangers @ Minnesota Twins

Texas starter Jacob deGrom brings a 3.11 FIP and .284 xwOBA-against that suppress Minnesota’s scoring outlook; Minnesota starter Joe Ryan’s 3.75 FIP and .297 xwOBA-against likewise indicate better-than-average defense-independent pitching and contact suppression against Texas. SharpSide projects Texas–Minnesota at 3.3–4.0, a narrow Twins win versus consensus at 4.2–3.3 for Texas, making the model 1.6 runs more favorable to Minnesota Twins while seeing a tighter game. Minnesota’s 4.76 relief ERA is worse than the 4.10 league benchmark and leaves late-inning run prevention as a meaningful risk. Even so, Ryan’s 3.75 FIP helps contain Texas, and Minnesota’s unlucky .284 BABIP earns a 0.09-run upward offensive correction. At -103, Minnesota’s price implies a probability just above even money, while our calibrated outlook describes a close game leaning slightly toward Minnesota—not a comfortably resolved advantage. The park/weather score is 0 with no notable conditions, and the 7.3 model total offers a +0.3-run betting-line edge against the posted 7.0 total, not against consensus. Friday opens a three-game series between eliminated clubs, making possible call-up auditions and lineup changes more relevant than postseason urgency and worth checking before first pitch. **How the model decided: The biggest factor favoring Minnesota Twins is Texas Rangers's vulnerable 4.45 bullpen ERA, worse than the roughly 4.10 league average, which improves Minnesota Twins's scoring outlook late. Secondary support comes from Minnesota Twins's unlucky 0.284 BABIP, worth a positive 0.09-run offensive regression. At -103 the market prices Minnesota Twins near 51% while the model has them around 55%, so a meaningful share of this lean is price value rather than fundamentals. Lean: Minnesota Twins ML (-103)

Baltimore Orioles @ New York Yankees

Baltimore starter Brandon Young carries a 4.53 FIP, while Yankees starter Brendan Beck’s 10.53 FIP and .435 xwOBA-against indicate poor results and hard contact allowed—but Beck’s 7.7 innings this season are far too small a sample to establish his talent level. SharpSide projects Baltimore–New York at 3.8–3.1, a slim Orioles win against consensus at 2.8–3.7 for the Yankees, making the model 1.6 runs more favorable to Baltimore Orioles in a tighter game. New York’s strong 3.11 relief ERA works against Baltimore; the model subtracts 0.45 runs from the Orioles’ projection. That said, Young’s 3.62 ERA and Baltimore’s better-than-average 3.79 relief ERA provide two run-prevention supports for the Orioles. At +105, Baltimore is priced below an even-money win probability, while our outlook is essentially a coin flip with the faintest edge to Baltimore, making this a modest price-based lean. Conditions subtract 0.36 runs from the projected total, yet the model’s 6.9 runs still produce a +0.4-run betting-line edge against the posted 6.5 total. This three-game series opens with Baltimore eliminated and New York holding a clinched wild card; without a confirmed lineup here, Yankees rest days are likely, so recheck personnel and workload plans before first pitch. **How the model decided: The biggest factor favoring Baltimore Orioles is the opposing starter's 10.53 FIP, a worse-than-average mark that raises Baltimore Orioles's scoring outlook. Supporting it: the opposing starter's 0.435 xwOBA-against, a worse-than-average contact-quality mark that benefits Baltimore Orioles's offense. At +105 the market prices Baltimore Orioles near 49% while the model has them around 52%, so a meaningful share of this lean is price value rather than fundamentals. Lean: Baltimore Orioles ML (+105)

New York Mets @ Washington Nationals

The Mets’ starter is TBA, so the model uses a league-average baseline instead of pitcher-specific inputs; Washington’s Andrew Alvarez brings a 3.13 FIP and 0.42 HR/9, both better than league benchmarks and supportive of suppressing Mets scoring. SharpSide projects New York–Washington at 3.6–4.1, a narrow Nationals win versus consensus at 4.5–3.6 for the Mets, making the model 1.4 runs more favorable to Washington Nationals while projecting a tighter game. Washington’s poor 5.01 relief ERA threatens the lean, while the Mets’ better-than-average 3.75 relief ERA makes scoring harder for the Nationals. Despite those factors, Alvarez’s 3.13 FIP and the 0.05-run upward correction to Washington’s unlucky .287 BABIP provide concrete support for Washington. At -108, the price requires a win probability slightly above even money, while our outlook is essentially a coin flip with the faintest edge to Washington, leaving little room for a worse number. Conditions subtract 0.27 runs from the projected total, with the 7.7 model total producing a −0.3-run betting-line edge against the posted 8.0 total. Both eliminated teams begin a three-game series Friday, with potential call-up auditions adding lineup uncertainty and Washington’s modest offensive correction insufficient to make the non-starter evidence independently convincing. **How the model decided: The biggest factor favoring Washington Nationals is Washington Nationals's starter carrying a 3.13 FIP, better than league average. Supporting it: Washington Nationals's starter running a 0.42 HR/9, better-than-average home-run suppression. Lean: Washington Nationals ML (-108)

First Five Innings: Chicago Cubs @ Boston Red Sox (F5 OVER)

Cubs starter David Peterson’s supplied ERA-side blend combines a 5.16 F5-only ERA over 120.3 innings with his 4.96 season ERA over 143.3 innings, using the specified 60/40 blend to reach 5.02. Boston starter Brayan Bello’s corresponding blend combines a 5.31 F5-only ERA over 79.7 innings with his 4.49 season ERA over 112.3 innings, producing the supplied 4.96 input. The Bayesian starter model then uses a 65/35 ERA/Statcast xERA blend, scales to five innings, and applies 0.92 early-game damping and a 0.930 park multiplier; separate pitcher xERA values are not supplied here. The result is 4.75 model F5 runs versus 3.38 market-implied runs, a +1.37-run edge toward F5 OVER, with the implied benchmark calculated from full-game consensus rather than treated as a quoted betting line. Chicago’s BABIP luck is neutral, Boston’s lucky BABIP receives a 0.06-run downward correction, and neither offense gets a baserunning adjustment. The over case therefore comes from the starter scoring environment despite the park multiplier and Boston’s offensive regression pulling the other way. Lean: F5 OVER 3.5 total

Batter Prop: Leody Taveras Over 0.5 Total Bases

Leody Taveras Total Bases Over 0.5 in Baltimore Orioles @ New York Yankees is a market-based lean, not a claim built on an unprovided hitter projection. The cross-book consensus price moved from -101 to -135, a supplied shift of 7.2 probability-points toward the OVER from opening. Consensus movement this aggressive historically flags smart money, although it does not prove who placed the wagers or guarantee the outcome. The current -135 price is less attractive than the opener, so the movement is evidence to weigh—not permission to chase any price. Lean: Leody Taveras Over 0.5 Total Bases (-135)

How the model decided

In Texas Rangers @ Minnesota Twins, deGrom’s strong peripherals lower Minnesota’s scoring expectation, while Texas’ weak relief pitching and Minnesota’s upward BABIP correction restore runs to the home side. In Baltimore Orioles @ New York Yankees, Beck’s poor contact results raise Baltimore’s scoring outlook with substantial sample uncertainty, while two effective bullpens and conditions subtracting 0.36 runs restrain the total. For New York Mets @ Washington Nationals, Alvarez’s pitching suppresses Mets runs, Washington’s bullpen pushes them back up, and the unannounced opposing starter keeps the side conditional. Our win-language calibration is fitted across every settled projection we have published, and these close featured games do not support meaningful fine-grained probability estimates. The model also considered advanced Sabermetric data and other analytics.

How to read this

This brief presents the model’s view against market consensus and available prices, not financial advice or a promise of results. Confirm starters, lineups, and current odds before acting, especially in a final-week slate where roster decisions can change the projection.

Projections update after publish. This brief is locked in at the 11:30 am ET refresh, but the model keeps recalculating all day as lineups are posted, weather shifts, pitching changes are announced, and lines move — a side or total highlighted here can flip before first pitch. Always confirm the current number on the live projections page (which locks 30 minutes before each game's first pitch) before placing a bet.