Over/under model

Totals

An independent over/under model: a purpose-built run estimator (starters, bullpen innings, lineups, park, and weather) turned into a run distribution, anchored to the totals market. Picks are paper-only until they show positive closing-line value — graded against the sharp totals close, never the moneyline close.

Paper / simulated
Paper recordWins–losses–pushes over graded totals picks. A final total exactly on the line is a push (stake returned), never a loss.21-17-2
Graded40
Avg CLVClosing-line value in probability points vs the sharp totals close — the gate for ever taking totals live.+0.70pp
Beat the close27/41

Totals board

Tue, Jul 21, 2026

scan complete · 0 rows

No totals picks on this slate

The model evaluated the over/under on every game below and none cleared the EV threshold (or weather was unavailable). Quiet days are part of the discipline.

Recent resultsTrack record

Analyzed · below EV threshold or weather-held

15 games

Chicago White Sox @ Texas Rangers · 8:05 PM ET

Over 8.5 (total runs)

Noah Schultz vs Kumar Rocker

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Fragile edgeWeather n/aThe weather feed was unavailable for this game, so the pick is shown but held to analysis-only — we never bet a total blind to weather.-110
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.+0.4%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.52.6%
MarketThe market's fair probability for this side — the over/under price with the vig removed.50.0%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.40
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.10.4vs market 8.5runs 5.3 – 5.2

P(over): model 60.4% · market 50.0% · blended 52.6%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 99 vs away staff 4.545.18r
  • Away offensewRC+ 103 vs home staff 4.475.26r
  • Park factorin team runs abovePF 102×1.020
  • Weatherin team runs aboveunavailableweather feed unavailable×1.000
  • Home staff (SP+RP innings)SP 4.47x5.3ip + RP 4.46x3.7ip = 4.47
  • Away staff (SP+RP innings)SP 5.06x5.3ip + RP 3.80x3.7ip = 4.54

Raw model total (home + away, park & weather included) 10.43

Distribution mean: market-implied 9.16 + 1.27 bounded tilt → 10.43

Negative-binomial distribution · mean 10.4 · variance 26.1

Weather / run environment

weather feed unavailable ×1.000 · applied once, inside the per-team runs · unavailable (held to analysis-only)

Market anchor

Blend (low · data conf 40): model P(over) 60.4% × 0.25 + market 50.0% → 52.6% (P(over), used for EV)

Sharp close: betonlineag 8.5 · P(over) 50.0%

Signals

  • · weather unavailable
  • · raw-vs-market gap 1.93r near skip limit
  • · weather unavailable -> analysis only (not bet)

Detroit Tigers @ Chicago Cubs · 8:05 PM ET

Over 8.0 (total runs)

Framber Valdez vs David Peterson

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Fragile edge-106
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-0.0%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.51.4%
MarketThe market's fair probability for this side — the over/under price with the vig removed.49.2%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.68
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.9.5vs market 8.0runs 4.8 – 4.7

P(over): model 58.2% · market 49.2% · blended 51.4%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 108 vs away staff 3.974.70r
  • Away offensewRC+ 97 vs home staff 4.544.83r
  • Park factorin team runs abovePF 100×1.000
  • Weatherin team runs above30C, wind 21km/h in 9×0.971
  • Home staff (SP+RP innings)SP 4.90x5.3ip + RP 4.02x3.7ip = 4.54
  • Away staff (SP+RP innings)SP 4.04x5.3ip + RP 3.88x3.7ip = 3.97

Raw model total (home + away, park & weather included) 9.53

Distribution mean: market-implied 9.07 + 0.46 bounded tilt → 9.53

Negative-binomial distribution · mean 9.5 · variance 23.8 · push 8.7%

Weather / run environment

30C, wind 21km/h in 9 ×0.971 · applied once, inside the per-team runs

Market anchor

Blend (low · data conf 68): model P(over) 58.2% × 0.25 + market 49.2% → 51.4% (P(over), used for EV)

Sharp close: pinnacle 8.0 · P(over) 48.0%

Signals

  • · fading sharp total by 3.2pp
  • · raw-vs-market gap 1.53r near skip limit

San Diego Padres @ Atlanta Braves · 7:15 PM ET

Under 9.0 (total runs)

Walker Buehler vs Reynaldo López

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Fragile edgeWeather n/aThe weather feed was unavailable for this game, so the pick is shown but held to analysis-only — we never bet a total blind to weather.-113
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-0.3%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.52.9%
MarketThe market's fair probability for this side — the over/under price with the vig removed.50.7%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.56
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.8.6vs market 9.0runs 3.9 – 4.7

P(over): model 40.5% · market 49.3% · blended 47.1%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 103 vs away staff 4.024.70r
  • Away offensewRC+ 89 vs home staff 3.863.92r
  • Park factorin team runs abovePF 101×1.010
  • Weatherin team runs aboveunavailableweather feed unavailable×1.000
  • Home staff (SP+RP innings)SP 4.26x5.3ip + RP 3.27x3.7ip = 3.86
  • Away staff (SP+RP innings)SP 4.29x5.3ip + RP 3.65x3.7ip = 4.02

Raw model total (home + away, park & weather included) 8.63

Distribution mean: market-implied 10.08 − 1.45 bounded tilt → 8.63

Negative-binomial distribution · mean 8.6 · variance 21.6 · push 8.1%

Weather / run environment

weather feed unavailable ×1.000 · applied once, inside the per-team runs · unavailable (held to analysis-only)

Market anchor

Blend (low · data conf 56): model P(over) 40.5% × 0.25 + market 49.3% → 47.1% (P(over), used for EV)

Sharp close: pinnacle 9.0 · P(over) 48.9%

Signals

  • · weather unavailable
  • · weather unavailable -> analysis only (not bet)

New York Mets @ Milwaukee Brewers · 7:40 PM ET

Under 9.0 (total runs)

Zac Thornton vs Brandon Sproat

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Fragile edgeWeather n/aThe weather feed was unavailable for this game, so the pick is shown but held to analysis-only — we never bet a total blind to weather.-120
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-0.8%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.54.1%
MarketThe market's fair probability for this side — the over/under price with the vig removed.52.3%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.56
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.8.6vs market 9.0runs 4.0 – 4.6

P(over): model 40.5% · market 47.7% · blended 45.9%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 105 vs away staff 4.004.62r
  • Away offensewRC+ 89 vs home staff 4.074.01r
  • Park factorin team runs abovePF 98×0.980
  • Weatherin team runs aboveunavailableweather feed unavailable×1.000
  • Home staff (SP+RP innings)SP 4.49x5.3ip + RP 3.47x3.7ip = 4.07
  • Away staff (SP+RP innings)SP 3.94x5.3ip + RP 4.08x3.7ip = 4.00

Raw model total (home + away, park & weather included) 8.63

Distribution mean: market-implied 9.88 − 1.25 bounded tilt → 8.63

Negative-binomial distribution · mean 8.6 · variance 21.6 · push 8.1%

Weather / run environment

weather feed unavailable ×1.000 · applied once, inside the per-team runs · unavailable (held to analysis-only)

Market anchor

Blend (low · data conf 56): model P(over) 40.5% × 0.25 + market 47.7% → 45.9% (P(over), used for EV)

Sharp close: betonlineag 9.0 · P(over) 47.7%

Signals

  • · weather unavailable
  • · weather unavailable -> analysis only (not bet)

Tampa Bay Rays @ Toronto Blue Jays · 7:07 PM ET

Over 7.5 (total runs)

Drew Rasmussen vs Kevin Gausman

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Moderate edge-112
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-1.1%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.52.2%
MarketThe market's fair probability for this side — the over/under price with the vig removed.50.5%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.70
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.8.8vs market 7.5runs 4.8 – 4.0

P(over): model 55.5% · market 50.5% · blended 52.2%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 90 vs away staff 3.644.01r
  • Away offensewRC+ 104 vs home staff 3.744.77r
  • Park factorin team runs abovePF 101×1.010
  • Weatherin team runs above20C, wind 11km/h out 11; retractable roof assumed OPEN×1.080
  • Home staff (SP+RP innings)SP 3.74x5.3ip + RP 3.73x3.7ip = 3.74
  • Away staff (SP+RP innings)SP 3.19x5.3ip + RP 4.29x3.7ip = 3.64

Raw model total (home + away, park & weather included) 8.78

Distribution mean: market-implied 8.21 + 0.57 bounded tilt → 8.78

Negative-binomial distribution · mean 8.8 · variance 21.9

Weather / run environment

20C, wind 11km/h out 11; retractable roof assumed OPEN ×1.080 · applied once, inside the per-team runs

Market anchor

Blend (mid · data conf 70): model P(over) 55.5% × 0.35 + market 50.5% → 52.2% (P(over), used for EV)

Sharp close: pinnacle 7.5 · P(over) 50.2%

Baltimore Orioles @ Boston Red Sox · 7:10 PM ET

Over 8.5 (total runs)

Kyle Bradish vs Eduardo Rivera

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Moderate edge-115
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-2.1%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.52.3%
MarketThe market's fair probability for this side — the over/under price with the vig removed.51.2%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.71
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.9.7vs market 8.5runs 4.8 – 4.9

P(over): model 54.5% · market 51.2% · blended 52.3%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 97 vs away staff 3.904.94r
  • Away offensewRC+ 99 vs home staff 3.654.76r
  • Park factorin team runs abovePF 108×1.080
  • Weatherin team runs above27C, wind 21km/h out 11×1.080
  • Home staff (SP+RP innings)SP 3.74x5.3ip + RP 3.52x3.7ip = 3.65
  • Away staff (SP+RP innings)SP 3.70x5.3ip + RP 4.19x3.7ip = 3.90

Raw model total (home + away, park & weather included) 9.70

Distribution mean: market-implied 9.30 + 0.40 bounded tilt → 9.70

Negative-binomial distribution · mean 9.7 · variance 24.3

Weather / run environment

27C, wind 21km/h out 11 ×1.080 · applied once, inside the per-team runs

Market anchor

Blend (mid · data conf 71): model P(over) 54.5% × 0.35 + market 51.2% → 52.3% (P(over), used for EV)

Sharp close: pinnacle 8.5 · P(over) 51.7%

Pittsburgh Pirates @ New York Yankees · 7:05 PM ET

Under 9.5 (total runs)

Bubba Chandler vs Will Warren

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Moderate edge-114
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-2.4%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.52.0%
MarketThe market's fair probability for this side — the over/under price with the vig removed.51.0%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.83
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.9.7vs market 9.5runs 4.8 – 4.9

P(over): model 46.3% · market 48.9% · blended 48.0%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 104 vs away staff 4.124.87r
  • Away offensewRC+ 113 vs home staff 3.784.84r
  • Park factorin team runs abovePF 99×0.990
  • Weatherin team runs above27C, wind 24km/h in 2×1.019
  • Home staff (SP+RP innings)SP 4.29x5.3ip + RP 3.04x3.7ip = 3.78
  • Away staff (SP+RP innings)SP 4.01x5.3ip + RP 4.28x3.7ip = 4.12

Raw model total (home + away, park & weather included) 9.71

Distribution mean: market-implied 10.04 − 0.33 bounded tilt → 9.71

Negative-binomial distribution · mean 9.7 · variance 24.3

Weather / run environment

27C, wind 24km/h in 2 ×1.019 · applied once, inside the per-team runs

Market anchor

Blend (mid · data conf 83): model P(over) 46.3% × 0.35 + market 48.9% → 48.0% (P(over), used for EV)

Sharp close: pinnacle 9.5 · P(over) 49.0%

Athletics @ Arizona Diamondbacks · 9:40 PM ET

Over 9.5 (total runs)

Jack Perkins vs Kohl Drake

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Fragile edge-103
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-2.7%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.49.4%
MarketThe market's fair probability for this side — the over/under price with the vig removed.48.4%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.68
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.10.4vs market 9.5runs 5.0 – 5.4

P(over): model 52.3% · market 48.4% · blended 49.4%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 93 vs away staff 4.685.43r
  • Away offensewRC+ 102 vs home staff 3.925.03r
  • Park factorin team runs abovePF 103×1.030
  • Weatherin team runs above36C, wind 5km/h out 4; retractable roof assumed OPEN×1.080
  • Home staff (SP+RP innings)SP 4.00x5.3ip + RP 3.81x3.7ip = 3.92 (SP rate missing->league)
  • Away staff (SP+RP innings)SP 4.07x5.3ip + RP 5.56x3.7ip = 4.68

Raw model total (home + away, park & weather included) 10.45

Distribution mean: market-implied 9.97 + 0.48 bounded tilt → 10.45

Negative-binomial distribution · mean 10.4 · variance 26.1

Weather / run environment

36C, wind 5km/h out 4; retractable roof assumed OPEN ×1.080 · applied once, inside the per-team runs

Market anchor

Blend (low · data conf 68): model P(over) 52.3% × 0.25 + market 48.4% → 49.4% (P(over), used for EV)

Sharp close: pinnacle 9.5 · P(over) 48.6%

Signals

  • · home starter TBD/unrated

St. Louis Cardinals @ Los Angeles Angels · 9:38 PM ET

Under 9.0 (total runs)

Matthew Liberatore vs Walbert Ureña

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Moderate edge+100
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-3.3%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.48.4%
MarketThe market's fair probability for this side — the over/under price with the vig removed.47.7%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.77
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.9.7vs market 9.0runs 4.3 – 5.3

P(over): model 50.4% · market 52.3% · blended 51.6%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 95 vs away staff 4.585.35r
  • Away offensewRC+ 96 vs home staff 3.704.35r
  • Park factorin team runs abovePF 101×1.010
  • Weatherin team runs above30C, wind 14km/h out 12×1.080
  • Home staff (SP+RP innings)SP 3.25x5.3ip + RP 4.35x3.7ip = 3.70
  • Away staff (SP+RP innings)SP 4.87x5.3ip + RP 4.18x3.7ip = 4.58

Raw model total (home + away, park & weather included) 9.70

Distribution mean: market-implied 10.45 − 0.75 bounded tilt → 9.70

Negative-binomial distribution · mean 9.7 · variance 24.3 · push 8.4%

Weather / run environment

30C, wind 14km/h out 12 ×1.080 · applied once, inside the per-team runs

Market anchor

Blend (mid · data conf 77): model P(over) 50.4% × 0.35 + market 52.3% → 51.6% (P(over), used for EV)

Sharp close: pinnacle 9.0 · P(over) 52.5%

Miami Marlins @ Houston Astros · 8:10 PM ET

Over 9.0 (total runs)

Tyler Phillips vs Tatsuya Imai

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Fragile edgeWeather n/aThe weather feed was unavailable for this game, so the pick is shown but held to analysis-only — we never bet a total blind to weather.-120
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-3.4%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.52.7%
MarketThe market's fair probability for this side — the over/under price with the vig removed.52.3%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.48
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.10.1vs market 9.0runs 5.2 – 4.9

P(over): model 53.8% · market 52.3% · blended 52.7%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 101 vs away staff 4.314.89r
  • Away offensewRC+ 104 vs home staff 4.435.19r
  • Park factorin team runs abovePF 100×1.000
  • Weatherin team runs aboveunavailableweather feed unavailable×1.000
  • Home staff (SP+RP innings)SP 4.58x5.3ip + RP 4.22x3.7ip = 4.43
  • Away staff (SP+RP innings)SP 4.55x5.3ip + RP 3.96x3.7ip = 4.31

Raw model total (home + away, park & weather included) 10.08

Distribution mean: market-implied 10.45 − 0.37 bounded tilt → 10.08

Negative-binomial distribution · mean 10.1 · variance 25.2 · push 8.3%

Weather / run environment

weather feed unavailable ×1.000 · applied once, inside the per-team runs · unavailable (held to analysis-only)

Market anchor

Blend (low · data conf 48): model P(over) 53.8% × 0.25 + market 52.3% → 52.7% (P(over), used for EV)

Sharp close: pinnacle 9.5 · P(over) 47.3%

Signals

  • · weather unavailable
  • · weather unavailable -> analysis only (not bet)

Cincinnati Reds @ Seattle Mariners · 9:40 PM ET

Over 7.5 (total runs)

Chase Burns vs Luis Castillo

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Moderate edge-114
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-3.6%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.51.4%
MarketThe market's fair probability for this side — the over/under price with the vig removed.50.9%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.75
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.8.4vs market 7.5runs 4.4 – 4.0

P(over): model 52.2% · market 50.9% · blended 51.4%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 91 vs away staff 3.814.01r
  • Away offensewRC+ 96 vs home staff 3.954.39r
  • Park factorin team runs abovePF 95×0.950
  • Weatherin team runs above25C, wind 9km/h out 6; retractable roof assumed OPEN×1.080
  • Home staff (SP+RP innings)SP 4.35x5.3ip + RP 3.39x3.7ip = 3.95
  • Away staff (SP+RP innings)SP 3.10x5.3ip + RP 4.84x3.7ip = 3.81

Raw model total (home + away, park & weather included) 8.40

Distribution mean: market-implied 8.26 + 0.14 bounded tilt → 8.40

Negative-binomial distribution · mean 8.4 · variance 21.0

Weather / run environment

25C, wind 9km/h out 6; retractable roof assumed OPEN ×1.080 · applied once, inside the per-team runs

Market anchor

Blend (mid · data conf 75): model P(over) 52.2% × 0.35 + market 50.9% → 51.4% (P(over), used for EV)

Sharp close: pinnacle 7.5 · P(over) 51.0%

Los Angeles Dodgers @ Philadelphia Phillies · 6:40 PM ET

Under 9.0 (total runs)

Justin Wrobleski vs Zack Wheeler

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Moderate edge-117
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-3.8%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.51.8%
MarketThe market's fair probability for this side — the over/under price with the vig removed.51.6%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.80
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.9.4vs market 9.0runs 4.8 – 4.7

P(over): model 47.8% · market 48.4% · blended 48.2%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 95 vs away staff 3.954.66r
  • Away offensewRC+ 114 vs home staff 3.354.75r
  • Park factorin team runs abovePF 102×1.020
  • Weatherin team runs above29C, wind 20km/h out 20×1.080
  • Home staff (SP+RP innings)SP 2.56x5.3ip + RP 4.49x3.7ip = 3.35
  • Away staff (SP+RP innings)SP 3.98x5.3ip + RP 3.90x3.7ip = 3.95

Raw model total (home + away, park & weather included) 9.41

Distribution mean: market-implied 9.97 − 0.56 bounded tilt → 9.41

Negative-binomial distribution · mean 9.4 · variance 23.5 · push 8.3%

Weather / run environment

29C, wind 20km/h out 20 ×1.080 · applied once, inside the per-team runs

Market anchor

Blend (mid · data conf 80): model P(over) 47.8% × 0.35 + market 48.4% → 48.2% (P(over), used for EV)

Sharp close: pinnacle 9.0 · P(over) 48.0%

San Francisco Giants @ Kansas City Royals · 7:40 PM ET

Under 9.0 (total runs)

Tyler Mahle vs Luinder Avila

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Moderate edge+102
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-3.9%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.47.6%
MarketThe market's fair probability for this side — the over/under price with the vig removed.47.2%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.75
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.9.8vs market 9.0runs 5.3 – 4.5

P(over): model 51.7% · market 52.8% · blended 52.4%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 98 vs away staff 4.274.51r
  • Away offensewRC+ 100 vs home staff 4.945.34r
  • Park factorin team runs abovePF 104×1.040
  • Weatherin team runs above35C, wind 20km/h in 19×0.920
  • Home staff (SP+RP innings)SP 4.45x5.3ip + RP 5.65x3.7ip = 4.94
  • Away staff (SP+RP innings)SP 4.04x5.3ip + RP 4.59x3.7ip = 4.27

Raw model total (home + away, park & weather included) 9.85

Distribution mean: market-implied 10.51 − 0.66 bounded tilt → 9.85

Negative-binomial distribution · mean 9.8 · variance 24.6 · push 8.4%

Weather / run environment

35C, wind 20km/h in 19 ×0.920 · applied once, inside the per-team runs

Market anchor

Blend (mid · data conf 75): model P(over) 51.7% × 0.35 + market 52.8% → 52.4% (P(over), used for EV)

Sharp close: pinnacle 9.0 · P(over) 52.3%

Minnesota Twins @ Cleveland Guardians · 6:40 PM ET

Under 8.0 (total runs)

Kendry Rojas vs Parker Messick

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Moderate edge-103
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-4.3%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.48.6%
MarketThe market's fair probability for this side — the over/under price with the vig removed.48.4%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.76
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.8.8vs market 8.0runs 4.2 – 4.6

P(over): model 51.1% · market 51.6% · blended 51.4%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 91 vs away staff 4.284.62r
  • Away offensewRC+ 104 vs home staff 3.404.15r
  • Park factorin team runs abovePF 97×0.970
  • Weatherin team runs above26C, wind 18km/h out 11×1.080
  • Home staff (SP+RP innings)SP 3.13x5.3ip + RP 3.80x3.7ip = 3.40
  • Away staff (SP+RP innings)SP 3.87x5.3ip + RP 4.88x3.7ip = 4.28

Raw model total (home + away, park & weather included) 8.77

Distribution mean: market-implied 9.35 − 0.58 bounded tilt → 8.77

Negative-binomial distribution · mean 8.8 · variance 21.9 · push 8.9%

Weather / run environment

26C, wind 18km/h out 11 ×1.080 · applied once, inside the per-team runs

Market anchor

Blend (mid · data conf 76): model P(over) 51.1% × 0.35 + market 51.6% → 51.4% (P(over), used for EV)

Sharp close: pinnacle 8.0 · P(over) 51.5%

Washington Nationals @ Colorado Rockies · 8:40 PM ET

Under 13.0 (total runs)

Carson Palmquist vs Michael Lorenzen

PaperSimulated pick. Totals stay paper-only until they show positive CLV vs the totals close — never sized with real money on this page.Moderate edge-110
EVExpected value: the model's edge from its blended over/under probability vs the price. +5% ≈ you'd net 5% of stake per bet long-run if the model is right.-4.5%
ModelThe model's probability for THIS side (over or under), after blending toward the de-vigged totals market. Used to compute EV.50.0%
MarketThe market's fair probability for this side — the over/under price with the vig removed.50.0%
KellyRecommended stake as a fraction of bankroll (quarter-Kelly, capped by tier). PAPER while totals are in simulation.0.0%
ConfData confidence (0–100): both starters rated, offense/bullpen data, weather feed up, lineups confirmed, and our build close to the market.78
CLVClosing-line value vs the SHARP totals close, in probability points. + = the market moved toward our side after we bet (we beat the close) — the gate to take totals live.—
Projected totalThe purpose-built run estimator's expected total (a bounded tilt off the market total), with the over/under split it implies.13.7vs market 13.0runs 7.4 – 6.3

P(over): model 49.9% · market 50.0% · blended 50.0%

Show run breakdown ▾

Run environment

  • Home offensewRC+ 106 vs away staff 4.576.27r
  • Away offensewRC+ 112 vs home staff 5.117.37r
  • Park factorin team runs abovePF 112×1.120
  • Weatherin team runs above32C, wind 10km/h in 4×1.025
  • Home staff (SP+RP innings)SP 5.13x5.3ip + RP 5.08x3.7ip = 5.11
  • Away staff (SP+RP innings)SP 3.86x5.3ip + RP 5.60x3.7ip = 4.57

Raw model total (home + away, park & weather included) 13.65

Distribution mean: market-implied 14.16 − 0.51 bounded tilt → 13.65

Negative-binomial distribution · mean 13.7 · variance 34.1 · push 7.0%

Weather / run environment

32C, wind 10km/h in 4 ×1.025 · applied once, inside the per-team runs

Market anchor

Blend (mid · data conf 78): model P(over) 49.9% × 0.35 + market 50.0% → 50.0% (P(over), used for EV)

Sharp close: pinnacle 13.0 · P(over) 50.0%