Behind the numbers

Our Birdland Metrics model weights each series differently. As the season starts to move towards the Fall, these series weights become more dramatic for teams on the playoff bubble. The Orioles fall into that category this year. This model scores every remaining Orioles series by how much it can actually move the playoff needle and then layers in per-game probabilities adjusted for who's pitching.

Series leverage

Each series receives a leverage score that combines four factors:

• Counterfactual swing — the raw difference in playoff odds between winning a sweep and getting swept. The wider the gap, the higher the leverage.
• Temporal multiplier — late-season series carry more weight, up to 3× in September, reflecting the shrinking margin for error.
• Competitor weight — matchups against division rivals (1.5×) and wild card contenders (1.25×) are amplified, since those games directly affect the teams Baltimore is racing.
• Race-tightness factor — leverage peaks when playoff odds are near 50%. The model cares most about series that can genuinely tip the balance.

Series that clear the top 10% of leverage scores are marked High Importance. The next tier, division matchups and other high-stakes opponents, are flagged as Key Series. Everything else is standard.

Game probability

Individual game win probabilities start with an Elo-based estimate adjusted for home field advantage. When a probable starting pitcher has been announced, that baseline is further refined using each starter's FIP (Fielding Independent Pitching) relative to the league average. A starter pitching well below average FIP pushes the odds in Baltimore's favor; an ace on the other side does the opposite. Games without an announced starter, shown as plain dashed circles, use the Birdland Metrics Elo baseline alone.

A dot filled with diagonal stripes means the model has a starter-adjusted probability for that game. Green indicates the model favors Baltimore (above 55%), red indicates the opponent is favored (below 45%), and yellow marks a genuine toss-up.

These probabilities carry real uncertainty. FIP is most reliable over a large sample of innings — a starter with a 3.20 FIP through 40 innings may look very different by September. Early-season or injury-returning pitchers are especially prone to noisy figures. Trade deadline acquisitions won't appear until their FanGraphs stats update, and newly announced starters may take a day to propagate. Treat the game-level probabilities as informed estimates, not precise predictions.

Data sources

Schedule and game results come from the MLB Stats API, updated daily after each game concludes. Probable pitchers are sourced from the MLB Stats API when available (typically 4–5 days out) and supplemented by ESPN's scoreboard API as a fallback for games up to 10 days ahead. Pitcher FIP figures are pulled from FanGraphs' 2026 season statistics. Playoff odds and projected win totals are derived from 10,000 Monte Carlo season simulations run daily using our Elo rating model, with current ratings lightly regressed toward each team's preseason projection.