MLB matchup analysis
Astros vs Mariners Prediction, Odds and Model Pick
Wednesday, September 23, 2026 at T-Mobile Park · 10:10 PM ET
Leo decision: Pass on Astros vs Mariners
The case for Houston Astros
The case for Seattle Mariners
Why it is not official
Why the setup score is what it is
Why the price is what it is
Read the full model output
The combined Lab Rating did not clear the official threshold.
Final result
Astros 5, Mariners 6
How the analysis held up
Seattle Mariners won. Leo's model favored the other side.
Leo's two internal reads disagreed pregame by 31.4 points — the team-strength model favored Houston Astros, the run-projection model favored Seattle Mariners. The run-projection model had the winner here.
George Kirby (No clear pitching-plan edge), the starter Leo's pitcher score favored, actually outpitched Ethan Pecko: 5.0 IP, 4 ER (7.20 ERA) against 2.2 IP, 3 ER (10.13 ERA). The edge held up. Neither starter actually pitched well tonight, though — this is a smaller-of-two-bads comparison, not a starter who dealt.
The Houston Astros bullpen was flagged as elevated risk pregame (High risk). In relief it actually allowed 2 earned runs over 6.2 innings (2.70 ERA) — the risk read did not show up on the scoreboard tonight. Note: 1 of the runs charged to the Houston Astros pitching staff overall was unearned — not something the bullpen's stuff should be blamed for.
The Seattle Mariners bullpen was not flagged pregame, but it had a night worth noting: 0 earned runs allowed over 5.0 relief innings (0.00 ERA) — quietly better than the pregame read implied. Note: 1 of the runs charged to the Seattle Mariners pitching staff overall was unearned — not something the bullpen's stuff should be blamed for.
Houston Astros scored 5 in a game that went 10 innings — not compared against the 3.2 pregame projection, which assumed nine.
Seattle Mariners scored 6 in a game that went 10 innings — not compared against the 4.8 pregame projection, which assumed nine.
Likely driver: Ethan Pecko, the pick's own starter, was rocked for a 10.13 ERA, and the Seattle Mariners bullpen quietly shut the game down (0.00 ERA over 5.0 innings), unflagged pregame.
Game information
| Matchup | Houston Astros at Seattle Mariners |
|---|---|
| Date | Wednesday, September 23, 2026 |
| First pitch | 10:10 PM ET |
| Venue | T-Mobile Park |
| Starting pitchers | Ethan Pecko vs George Kirby |
| Weather | Game-time forecast: 58°F, 0% precipitation chance, 5 mph wind from N. This venue has a roof, but the roof status is not confirmed. |
Starting pitcher matchup
Full Pitcher Matchup Tool →| Metric | Astros | Mariners |
|---|---|---|
| Pitching plan | Limited starter | Traditional starter |
| Expected innings | 4⅓ | 6 |
| Throws | R | R |
| Leo pitcher score (higher is better) | 63 | 61 |
| ERA (lower is better) | 3.77 | 4.24 |
| WHIP (lower is better) | 1.12 | 1.39 |
| K/9 (higher is better) | 6.0 | 7.6 |
| K-BB% (higher is better) | 6.0% | 13.8% |
| K% (higher is better) | 16.4% | 19.9% |
| BB% (lower is better) | 10.3% | 6.1% |
| BB/9 (lower is better) | 3.8 | 2.3 |
| HR/9 (lower is better) | 1.6 | 1.1 |
| Ground-ball rate | 39.7% | 48.1% |
| Fly-ball rate | 60.3% | 51.9% |
Pitcher edge: Leo gives Ethan Pecko a 2-point edge over George Kirby, driven mainly by WHIP (1.12 vs 1.39) and ERA (3.77 vs 4.24).
How to read this: the scorecards above name the starters; every number lives in this table.
Strikeout Projections Full strikeout projections →
Same data source as the Pitcher Matchup Tool, where every starter on the slate is compared side by side.
Recent form
Full Stats page →Astros are 3-7 in their last 10 and averaging 3.7 runs per game over the last 15 days; Mariners are 4-6 in their last 10 and averaging 5.0 runs per game over the last 15 days. Season run differential separates them by almost nothing (-0.34 against -0.39).
| Metric | Astros | Mariners |
|---|---|---|
| Last 10 | 3-7 | 4-6 |
| Last 10 by venue (away team on the road, home team at home) | 4-6 on the road | 4-6 at home |
| OPS, last 15 days | 0.760 | 0.757 |
| wOBA, last 15 days (hot/cold read) | 0.330 | 0.331 |
| Runs per game, last 15 days | 3.7 | 5.0 |
| K% last 15 days (lower is better) | 23.7% | 20.1% |
Last 15 days only. Hot and cold streaks for all 30 teams live on the Stats page.
30-day offense
Full offense tool →| Last 30 days | Astros | Mariners |
|---|---|---|
| OPS | 0.718 (-0.008 vs szn) | 0.721 (+0.031 vs szn) |
| Runs per game | 3.85 (-0.579 vs szn) | 4.70 (+0.653 vs szn) |
| wOBA (model offense input) | 0.314 | 0.320 (+0.013 vs szn) |
| K% (lower is better) | 23.1% (+1.2% vs szn) | 23.2% |
Over the last 30 days the Mariners have been the better offense, 0.320 wOBA to 0.314. They are running hot against their own season line (+0.013 wOBA).
Season profile
Full Stats page →| Metric | Astros | Mariners |
|---|---|---|
| Record | 78-79 | 73-84 |
| Run differential / game (team quality) | -0.34 | -0.39 |
| Runs scored / game (offense, season) | 4.43 | 4.05 |
| Runs allowed / game (defense, season, lower better) | 4.77 | 4.44 |
| Season OPS | 0.725 | 0.691 |
| K% season (lower is better) | 21.9% | 23.1% |
| OPS vs opposing hand (season) | 0.729 | 0.709 |
Season-long team quality, not recent form. Run differential and run environment come from the season standings; splits for all 30 teams live on the Stats page, including the run environment table.
The Map
Full interactive map →Every team on today's slate, on any two axes you pick. Astros and Mariners are highlighted.
Bullpen matchup
Full Bullpen Fatigue Index →Risk is what the model actually uses: fatigue blended with how well the pen has pitched. High fatigue with high efficiency is a tired pen that is still getting outs.
| Metric | Astros | Mariners |
|---|---|---|
| Fatigue | 8.7/10, High risk | 6.1/10, Normal |
| Efficiency | 4.9/10, Average | 4.6/10, Average |
| Combined risk | 8.8/10, High risk | 6.3/10, Tired |
| Relief innings, last 7 days | 33.0 | 18.0 |
| Back-to-back arms | 4 | 4 |
| 7-day ERA | 4.64 | 4.50 |
| 7-day WHIP | 1.21 | 1.44 |
Fatigue measures workload. Efficiency measures recent run prevention. Combined risk is what the moneyline and totals systems use.
Run total projection
Full Totals Projections →The model and market are close. Team totals remain research-only until their graded sample is established.
More games today
All matchups →Every Leo pick, graded in public
Free daily model card by email, or open today's full slate. Membership adds delivery before first pitch.
Model outputs are not guarantees. Leo provides analysis and education only. Every official pick remains visible on the Results page. 21+. If you or someone you know has a gambling problem, call 1-800-GAMBLER.