The Los Angeles Dodgers host the San Francisco Giants at Dodger Stadium on Friday, September 18, 2026, with first pitch at 10:15 p.m. ET. The standings frame the night bluntly: the Dodgers, at 93-60, have clinched the NL West and are playing for a top-two seed and a first-round bye; the Giants, at 64-89, are running out a difficult season under first-year manager Tony Vitello. Yet the head-to-head record complicates the tidy story — San Francisco holds a 4-3 season edge, having taken three of four in the May series at this same ballpark, and the Giants have won 2 straight coming in. The Dodgers have won 1 straight.
Our engine names Tarik Skubal as the Los Angeles starter, with San Francisco's Game 1 arm still TBD. That single fact does a great deal of the work below, but it is not the whole case. The number is.
The number
The pick is Los Angeles moneyline. Our model assigns the Dodgers an 85% probability to win. The market-implied probability is 72%. That is an edge of 13 points.
For context on the market side, the Dodgers have been listed around -300 and the Giants around +240. Those prices are consistent with the roughly 72% figure our model is measuring against. The 13-point gap is the entire reason this game appears on today's card; without it, there is no pick.
Why the model leans this way
Three inputs drive the read.
First, offense. Our ratings separate these lineups sharply — Los Angeles at 57.9, San Francisco at 25.1. That is not a small gap, and it is the single largest structural factor in the projection.
Second, the starter and the run environment. With Skubal on the mound for Los Angeles, the model projects a lower-scoring game than the market: our total is 7.5 against a market total of 8.5, a modest Under lean. A suppressed run environment generally favors the stronger, more consistent offense to grind out enough, and the park factor here of 0.97 — slightly pitcher-friendly relative to a 1.0 neutral baseline — nudges the same direction. The projected margin is 3.13 runs in favor of Los Angeles.
Third, the shape of the projection. The model expects Los Angeles to win 6-2. That score aligns with everything above: a controlled total, a decisive offensive edge, and a home starter the model rates highly. Taken together, those elements produce the 85% figure. None of this is a claim of certainty; it is a probability, and a 15% chance of a Los Angeles loss is real and built into the number.
The case against
Honesty requires stating what beats this pick, and it beats it more than one time in seven.
Start with the head-to-head. San Francisco is 4-3 against Los Angeles this season and took three of four here in May. Whatever the model measures in aggregate, this specific matchup has not gone one way in 2026. The Giants also arrive on a 2-game winning streak, while Los Angeles has won only 1 straight — a small edge in current form that sits with the underdog, not the favorite.
Motivation cuts the wrong way, too. Los Angeles has already clinched the division; the incentive to fully extend regulars, or to manage a pitcher's workload with the postseason in view, is a genuine variable a model does not always price. San Francisco, meanwhile, has a clear spoiler role, with six games against Los Angeles across the season's final 10 days.
The pitching picture carries its own risk. The Giants' Game 1 starter is listed as TBD; an unknown arm is a wider distribution of outcomes than a known one, and that uncertainty can break in the underdog's favor. And baseline variance in a single nine-inning game is large. A -300 favorite is expected to lose roughly one in four times at the market's own price; even at the model's more confident 85%, roughly one in seven of these games ends with the underdog winning. Bullpen sequencing, one swing in a suppressed-scoring game, or a short outing from a rested-managed starter can all deliver that outcome.
Net: the model's 13-point edge is the reason to be interested, but the underdog's matchup history, current form, and ordinary variance are the reasons this is a probability and not a promise.
21+. Predictions carry risk; only stake what you can afford to lose, and reach out for help if it stops being fun.