BAL vs HOU prediction for July 17, 2026: Our Monte Carlo simulation ran 10,000 game iterations and projects HOU 6.2 - BAL 5.0. HOU is favored with a 57.7% win probability. The run line is 1.5 and the total is 8.5. Model projects 11.3 total runs.
HOU
6.2
Projected Score
VS
O/U 8.5
BAL
5.0
Projected Score
Win Probability
HOUBAL
+1.5
Run Line (HOU)
8.5
Total Line
10,000
Simulations
Calibrated accuracy at this confidence: 56.0% (2,907 games)
Projected Runs Range 10th – 90th percentile
BAL
357
HOU
468
Projected
HOU 6.2 — BAL 5.0
Actual
HOU 2 — BAL 3
Starting Pitcher Matchup
Dean Kremer R
BAL
FS35%83 mph45% whiff
FF23%93 mph12% whiff
SI16%92 mph15% whiff
Peter Lambert R
HOU
FF31%94 mph17% whiff
CH22%87 mph35% whiff
SL16%86 mph26% whiff
Weather Impact
Minute Maid Park
85°F9 mph windRoof: retractable
HR: 1.036 Total: 1.018
thin air
Bullpen Comparison
BAL
4.34ERA
4.01FIP
8.76K/9
3.44BB/9
1.30WHIP
HOU
4.13ERA
4.29FIP
8.50K/9
4.68BB/9
1.32WHIP
Betting Edges
TOTAL UNDER 8.5
-31.7% EV
+100
RUN_LINE AWAY -1.5
-26.4% EV
+142
TOTAL OVER 8.5
+19.9% EV
-122
F5_ML AWAY
-17.5% EV
-106
RUN_LINE HOME +1.5
-17.4% EV
-172
ML AWAY
-12.0% EV
-104
First 5 Innings & NRFI
BAL F5
2.8 runs
35.7% win
HOU F5
3.6 runs
52.5% win
F5 Total
6.4
NRFI
45.9%
YRFI
54.1%
Avg 1st Inn Runs
1.29
HR Spotlight
Avg HRs
2.8
Over 0.5 HR
93%
Over 1.5 HR
76%
No HR
7%
Pete Alonso BAL30.0%
ISO: 0.250 | Barrel: 18.9% | vs Peter Lambert | Park: 0.99x
Samuel Basallo BAL30.0%
ISO: 0.260 | Barrel: 8.9% | vs Peter Lambert | Park: 0.99x Platoon: 1.12x
Yordan Alvarez HOU30.0%
ISO: 0.323 | Barrel: 13.8% | vs Dean Kremer | Park: 0.99x Platoon: 1.12x
Pitcher Strikeout Projections
Dean Kremer
0.0 K projected
BAL | K/9: 0.0
Peter Lambert
0.0 K projected
HOU | K/9: 0.0
Injury Report
BAL8 injured
Blaze Alexander 3B10-DAY-IL
Keegan Akin RP60-DAY-IL
Ryan Helsley RP15-DAY-IL
Felix Bautista RP60-DAY-IL
Colin Selby RP60-DAY-IL
Chris Bassitt SP15-DAY-IL
+2 more
HOU7 injured
Brice Matthews CF10-DAY-IL
Mike Burrows SP15-DAY-IL
Ronel Blanco SP60-DAY-IL
Bennett Sousa RP60-DAY-IL
Hayden Wesneski SP60-DAY-IL
Brandon Walter SP60-DAY-IL
+1 more
AI Intelligence Analysis
LEAN
Model projects 11.28 total runs with 19.9% edge vs market 8.5 on PITCHER MISMATCH favoring HOU: Peter Lambert (B- 0.44, 3.39 ERA) vs Dean Kremer (B 0.632, 4.42 ERA). Kremer is BETTER pitcher (0.632 vs 0.44 grade), but model favors HOME team anyway. Edge is PARK-DRIVEN (1.0 neutral park, but BAL is road team undervaluing). 19.9% edge historically mixed = break-even recommendation. LEAN at 0.5.
Key Factors
- PITCHER MISMATCH AGAINST MODEL PICK: Dean Kremer (B 0.632, 27.6% K%, elite stuff 0.582, command 0.693) is objectively BETTER than Peter Lambert (B- 0.44, 22.6% K%, moderate stuff 0.347, command 0.554). Kremer advantage is ~0.192 grade. Yet model favors HOU (home team) with 65.9% over prob.
- Model Logic: Home park + batter advantage + bullpen depth = HOU over value despite pitcher disadvantage. This is PLAUSIBLE but contradicts PITCHER ATTRIBUTION RULE. Model must state: 'HOU favors despite pitcher disadvantage because park/lineup/bullpen swing to home team.'
- Park Factor: Minute Maid 1.0 (neutral). 85.4°F with 9.2 mph wind (1.0 mph tail) = neutral weather baseline. No park advantage.
- Bullpen Context: BAL missing Ryan Helsley (15-day IL elbow, closer prospect). HOU relatively fresh. Lever bullpen edge to HOU.
- 19.9% Edge: Falls historically mixed. Edge likely disappears in execution.
Risk Factors
- PITCHER QUALITY REVERSED: Kremer (B, 0.632) > Lambert (B-, 0.44). Model is playing against pitcher advantage. This only works if park/bullpen/home field swing is >0.192 grade.
- historically mixed: a supportive track record means edge is break-even in real execution. Model has no edge here.
- Neutral Park: 1.0 PF means no home field advantage from park. HOU edge must come entirely from bullpen/lineup/home-field psychology.
PITCHER DISADVANTAGE
Edge Analysis
Moneyline
HOU 57.7%
-17.4 pts
Run Line
+1.5
-17.4 pts
Total
8.5
+19.9 pts
How this prediction was generated: This page shows output from the Olympus Bets MLB Baseball Monte Carlo engine. Each game is simulated 10,000 times using real-time team data, injury reports, and current odds. Probabilities are calibrated using Bayesian methods and sized via the Kelly Criterion. Probabilities are calibrated using Bayesian methods and sized via the Kelly Criterion. Full methodology →
Permanent record: this game page is retired 90 days after the event. The full graded MLB 2026 record — every pick, win rate, units and ROI — lives on the MLB 2026 season report card.