Baseball · Game notebook · 2026-09-04
Chicago Cubs at Miami Marlins
01 · Forecast
What did we believe?
No current actionable forecast is substituted for this game. The recorded checkpoints below are the historical evidence.
Moneyline · 5 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Miami Marlins · 49.2%+140 at kalshi · recorded edge +7.5%
Paper recommendation
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Miami Marlins · 49.2%+140 at kalshi · recorded edge +7.5%
Paper recommendation
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
Total 8.5 · 4 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 8.5 · 48.3%+109 at polymarket · recorded edge +0.5%
No paper recommendation recorded
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 8.5 · 48.7%+101 at kalshi · recorded edge -1.1%
No paper recommendation recorded
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
Total 11.5 · 2 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 11.5 · 20.8%+313 at kalshi · recorded edge -3.4%
No paper recommendation recorded
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Under 11.5 · 79.2%-381 at kalshi · recorded edge +0.0%
No paper recommendation recorded
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
Total 4.5 · 2 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Under 4.5 · 17.5%+456 at kalshi · recorded edge -0.5%
No paper recommendation recorded
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 4.5 · 82.5%-609 at kalshi · recorded edge -3.4%
No paper recommendation recorded
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
Total 7.5 · 5 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Under 7.5 · 50.3%-112 at kalshi · recorded edge -2.5%
No paper recommendation recorded
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 7.5 · 52.0%-131 at polymarket · recorded edge -5.2%
No paper recommendation recorded
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
Total 10.5 · 2 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 10.5 · 27.9%+208 at kalshi · recorded edge -4.5%
No paper recommendation recorded
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Under 10.5 · 72.1%-263 at kalshi · recorded edge -0.4%
No paper recommendation recorded
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
Total 3.5 · 2 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 3.5 · 88.3%-1087 at kalshi · recorded edge -3.2%
No paper recommendation recorded
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Under 3.5 · 11.7%+756 at kalshi · recorded edge -0.0%
No paper recommendation recorded
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
Total 6.5 · 2 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Under 6.5 · 36.3%+188 at polymarket · recorded edge +1.2%
No paper recommendation recorded
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 6.5 · 63.7%-209 at polymarket · recorded edge -4.4%
No paper recommendation recorded
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
Total 5.5 · 2 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 5.5 · 71.8%-322 at kalshi · recorded edge -4.5%
No paper recommendation recorded
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Under 5.5 · 28.2%+252 at kalshi · recorded edge -0.2%
No paper recommendation recorded
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
Total 9.5 · 2 recorded snapshots
First look
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Over 9.5 · 34.3%+159 at kalshi · recorded edge -4.3%
No paper recommendation recorded
Quote observed .Latest pregame
· retrosheet_lightgbm_poisson_blend retrosheet-game-date-bounded-fielding-2026-v16.0.0
Under 9.5 · 65.7%-191 at kalshi · recorded edge +0.0%
No paper recommendation recorded
Quote observed .First and latest pregame are model snapshots, not necessarily the bookmaker’s opening and closing prices.
02 · Why
Connect the forecast to the baseball
Recorded first look · confirmed-lineup
Tree contributions are raw log-odds toward the home team (+) or away team (−), before calibration. They are not probability points or proof of causation.
| Baseball input | Recorded value | Contribution |
|---|---|---|
Matchup edge: the lineup's prior-year wRC+The home team's the lineup's prior-year wRC+ minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0780 |
Matchup edge: the starter's season-to-date strikeout-minus-walk rateThe home team's the starter's season-to-date strikeout-minus-walk rate minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0575 |
Matchup edge: season run differentialThe home team's season run differential minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0318 |
Matchup edge: expected winning percentage from runs scored and allowedThe home team's expected winning percentage from runs scored and allowed minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0252 |
Matchup edge: runs allowed per game this seasonThe home team's runs allowed per game this season minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | +0.0237 |
Matchup edge: batters the starter has faced this season (evidence weight)The home team's batters the starter has faced this season (evidence weight) minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0182 |
Matchup edge: the lineup's prior-year wOBAThe home team's the lineup's prior-year wOBA minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0141 |
Matchup edge: the starter's prior-year xFIP (lower is better)The home team's the starter's prior-year xFIP (lower is better) minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | +0.0132 |
Recorded input excerpt (11 values)
- Matchup edge: the lineup's prior-year wRC+
- Missing / not retained
- Matchup edge: the starter's season-to-date strikeout-minus-walk rate
- Missing / not retained
- Matchup edge: season run differential
- Missing / not retained
- Matchup edge: expected winning percentage from runs scored and allowed
- Missing / not retained
- Matchup edge: runs allowed per game this season
- Missing / not retained
- Matchup edge: batters the starter has faced this season (evidence weight)
- Missing / not retained
- Matchup edge: the lineup's prior-year wOBA
- Missing / not retained
- Matchup edge: the starter's prior-year xFIP (lower is better)
- Missing / not retained
- away lineup steamer coverage
- 0
- home lineup steamer coverage
- 0
- lineup steamer coverage difference
- 0
This bounded excerpt is not the complete feature vector. Null values are not assumed to be zero.
Recorded latest pregame · confirmed-lineup
Tree contributions are raw log-odds toward the home team (+) or away team (−), before calibration. They are not probability points or proof of causation.
| Baseball input | Recorded value | Contribution |
|---|---|---|
Matchup edge: the lineup's prior-year wRC+The home team's the lineup's prior-year wRC+ minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0780 |
Matchup edge: the starter's season-to-date strikeout-minus-walk rateThe home team's the starter's season-to-date strikeout-minus-walk rate minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0575 |
Matchup edge: season run differentialThe home team's season run differential minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0318 |
Matchup edge: expected winning percentage from runs scored and allowedThe home team's expected winning percentage from runs scored and allowed minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0252 |
Matchup edge: runs allowed per game this seasonThe home team's runs allowed per game this season minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | +0.0237 |
Matchup edge: batters the starter has faced this season (evidence weight)The home team's batters the starter has faced this season (evidence weight) minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0182 |
Matchup edge: the lineup's prior-year wOBAThe home team's the lineup's prior-year wOBA minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | -0.0141 |
Matchup edge: the starter's prior-year xFIP (lower is better)The home team's the starter's prior-year xFIP (lower is better) minus the away team's. A positive value means the home team has more of this statistic. Whether that is favorable depends on the statistic and the model; higher ERA or xFIP is generally worse. | Not retained | +0.0132 |
Recorded input excerpt (11 values)
- Matchup edge: the lineup's prior-year wRC+
- Missing / not retained
- Matchup edge: the starter's season-to-date strikeout-minus-walk rate
- Missing / not retained
- Matchup edge: season run differential
- Missing / not retained
- Matchup edge: expected winning percentage from runs scored and allowed
- Missing / not retained
- Matchup edge: runs allowed per game this season
- Missing / not retained
- Matchup edge: batters the starter has faced this season (evidence weight)
- Missing / not retained
- Matchup edge: the lineup's prior-year wOBA
- Missing / not retained
- Matchup edge: the starter's prior-year xFIP (lower is better)
- Missing / not retained
- away lineup steamer coverage
- 0
- home lineup steamer coverage
- 0
- lineup steamer coverage difference
- 0
This bounded excerpt is not the complete feature vector. Null values are not assumed to be zero.
Tabular model explanation
shapiq-imputation-baseline-sv-v1 · class-1 logit scale · 512 evaluation budget. This explains the tabular opinion, not the recommendation policy.
- Matchup edge: the lineup's prior-year wRC+: -0.0981
- Matchup edge: start-of-day margin above or below the wild-card cutoff: -0.0443
- Matchup edge: proximity to the start-of-day playoff qualification cutoff: -0.0216
- Matchup edge: the lineup's season-to-date defensive error prevention: -0.0214
- Matchup edge: the starter's season-to-date strikeout-minus-walk rate: -0.0202
- Matchup edge: batters the starter has faced this season (evidence weight): -0.0181
- Matchup edge: the starter's prior-year xFIP (lower is better): 0.0175
- Matchup edge: expected winning percentage from runs scored and allowed: -0.0160
Eye-test prompt: which recorded input fits what you saw, and which would you question? A single win or loss cannot establish that the reasoning was right.
03 · Movement
What changed?
Recorded moneyline prices
Prices move at a fixed selection; a change in a totals line is a different contract. Collection runs independently of opening this notebook.
Latest available books and main / alternate totals
Current board coverage may no longer include completed games. An empty board is not proof that no historical quotes were collected.
Model snapshots
No retained movement series is available in the published summary. First/latest checkpoints remain above.
04 · Review
Price quality and result are separate questions
No settled score recorded yet.
| Market | Latest pregame selection | Forecast | Outcome | Recorded CLV |
|---|---|---|---|---|
| Total 8.5 | Over 8.5 | 48.7% | Pending | +1.9% |
| Moneyline | Miami Marlins | 49.2% | Lost | +0.0% |
| Total 11.5 | Under 11.5 | 79.2% | Won | Unavailable / incomparable |
| Total 4.5 | Over 4.5 | 82.5% | Won | Unavailable / incomparable |
| Total 7.5 | Over 7.5 | 52.0% | Lost | Unavailable / incomparable |
| Total 10.5 | Under 10.5 | 72.1% | Won | Unavailable / incomparable |
| Total 3.5 | Under 3.5 | 11.7% | Lost | Unavailable / incomparable |
| Total 6.5 | Over 6.5 | 63.7% | Won | Unavailable / incomparable |
| Total 5.5 | Under 5.5 | 28.2% | Lost | Unavailable / incomparable |
| Total 9.5 | Under 9.5 | 65.7% | Won | Unavailable / incomparable |
Recorded CLV compares the first and latest eligible pregame implied probabilities under the existing history policy. It is not a verified sportsbook closing tick. Missing or changed-selection comparisons stay unavailable.
A good price can lose; a poor forecast can win. Use forecast quality and matched model cohorts to judge repeated performance.
Technical details
- Forecast captured
- Unavailable
- Moneyline captured
- Unavailable
- Total market captured
- Unavailable
- Score last fetched
- Unavailable
The score source does not supply a per-game update timestamp; fetch time records when this score feed was retrieved, not when the score changed.
Totals difference withheld: comparison requires known pregame capture times, with the market captured no more than 30 minutes before the forecast.
Model input coverage
This record does not contain per-game input coverage. Availability cannot be inferred from the displayed probabilities.