EVERY MODEL,
ONE BOARD
One page for every forecasting model I run β whichever sport is in season sits on top, the rest wait their turn. TRACKER β bookmaker odds and experimental model picks. Not betting advice.
Record Β· accuracy Β· last graded day, game-weighted across every graded day. A model only reports once its picks have been graded. ROI is an em dash β nothing here stakes money. Log-loss baseline: 0.693 = coin flip, ~0.691 = always pick home.
βΎ MLB DAILY MODEL
15 GAMES Β· RUN 2026-08-28Betting-blind Elo (K=3, +24 home, margin-of-victory weighted). Backtest 2009β2025: 56.7% straight up, log-loss 0.680 vs 0.691 for always-pick-home.
| Matchup (home second) | SP (away vs. home) | SP adj (Elo, away/home) | Model pick | Win prob | Sim score |
|---|---|---|---|---|---|
| Orioles @ Athletics | Brandon Young vs. Jacob Lopez | +1.1 / +17.4 | Orioles | 52% | 5β4.5 |
| Rockies @ Braves | Tomoyuki Sugano vs. Grant Holmes | +13.9 / -7.0 | Braves | 67% | 5.3β3.8 |
| Reds @ Cubs | Rhett Lowder vs. David Peterson | -0.5 / -12.3 | Cubs | 63% | 5.1β4 |
| Royals @ Guardians | Michael Wacha vs. Tanner Bibee | +7.9 / +0.3 | Guardians | 56% | 4.6β4.2 |
| Dodgers @ Tigers | Tarik Skubal vs. Drew Anderson | +20.5 / -13.7 | Dodgers | 57% | 4.8β3.8 |
| Phillies @ Angels | Andrew Painter vs. Reid Detmers | -30.0 / +1.7 | Phillies | 53% | 4.6β4 |
| Rangers @ Brewers | Cody Bradford vs. Logan Henderson | +3.1 / +19.8 | Brewers | 67% | 5.1β3.5 |
| White Sox @ Twins | Luis Castillo vs. Dean Kremer | +7.5 / +5.5 | White Sox | 51% | 4.8β4.4 |
| Astros @ Mets | Hunter Brown vs. Christian Scott | +25.7 / +6.3 | Astros | 50% | 4.6β4.3 |
| Red Sox @ Yankees | Patrick Sandoval vs. Cam Schlittler | -13.0 / +16.7 | Yankees | 57% | 4.5β4 |
| Diamondbacks @ Giants | Kohl Drake vs. Blade Tidwell | +0.9 / -13.5 | Diamondbacks | 53% | 4.7β4 |
| Pirates @ Cardinals | Jared Jones vs. Quinn Mathews | +2.5 / +1.2 | Cardinals | 53% | 4.5β4.4 |
| Padres @ Rays | Casey Mize vs. Shane McClanahan | -0.6 / +2.3 | Rays | 54% | 4.3β4.2 |
| Mariners @ Blue Jays | Emerson Hancock vs. Dylan Cease | +5.6 / +22.7 | Blue Jays | 57% | 4.6β4.1 |
| Marlins @ Nationals | Eury PΓ©rez vs. Jackson Kent | +6.4 / +1.6 | Marlins | 51% | 4.6β4.2 |
Score is the mean of 10,000 sims, conditioned on the picked side winning (winnerβs runs first), shown at one decimal β integer rounding collapses nearly every game onto 6β3. Starter identity IS a model input: each probableβs rolling game score (exponentially weighted, 20-start half-life) is compared to his teamβs staff average, and the difference enters the teamβs Elo at 3.0 points per game-score point β the SP adj column, so the effect is auditable per game. Rookies/thin history fall back to a shrunk staff rating; TBD uses the staff rating. Rest and travel add up to a few Elo points each.
UP NEXT SEASON
NFL
Off-seasonKickoff September 2026. The odds feed is already live β last pull 2026-08-28, no games inside the 48-hour window.
LightGBM straight-up and against-the-spread models are trained but not yet wired into the slate, so there is no graded record to show.
NBA
Off-seasonTip-off October 2026. The odds feed is already live β last pull 2026-08-28, no games inside the 48-hour window.
Logistic-regression winner model plus a ridge score model on rolling team stats. Picks post to the slate; grading starts with the first played game.
MLB data generated 2026-08-28 21:03:21 Β· odds feed last pulled 2026-08-28 01:58:23 ET (NFL) and 2026-08-28 01:58:23 ET (NBA).