NPB · SportsGap
Slate
6 games
NPB · research only
Moneyline leans
182-162-5
-9.1% flat 1u · 349 games since 07/16
Totals leans
195-151-16
+5.1% flat 1u · 362 games since 07/16
Team totals leans
379-360
-5.5% flat 1u · 739 picks since 07/16
Rakuten@ORIX
MLRakuten+13055.1%+12.9✗
TTRakuten U 3.5≈-14251.4%-4.5✓
TTORIX U 3.5≈+10460.9%+14.1✓
Seibu@Lotte
MLLotte+10855.8%+9.1✓
TotalO 7-14164.9%+8.6✓
TTLotte O 3.5+10754.8%+8.6✓
Hanshin@DeNA
MLHanshin-11957.8%+4.9✗
TotalO 6.5-10559.3%+10.0✓
TTHanshin O 3.5+10355.6%+8.6✓
TTDeNA O 2.5-12562.2%+9.2✓
Giants@Hiroshima
MLGiants-12550.2%-3.8✓
TotalO 6.5+10154.5%+6.6✗
TTHiroshima O 2.5-11360.6%+9.9✗
Yakult@Chunichi
TotalU 7≈-12262.5%+9.5✓
TTYakult U 2.5≈+10755.8%+9.7✓
TTChunichi U 3.5≈-13157.1%+3.0✗
Nippon-Ham@SoftBank
TotalO 7+10352.7%+5.2✓
TTNippon-Ham U 3.5≈-14451.0%-5.3✗
How to read this page

Each row is one game. Model is our model's probability that its pick wins. Book is the probability the sportsbook's price implies once its built-in margin (the vig) is removed, so the two sides add up to 100%. Gap is model minus book, in percentage points: a positive gap means the model likes the pick more than the market does, and the bigger the gap, the more they disagree.

Totals are the combined runs both teams score, over or under the posted number; team totals count one team's runs only. Units is the suggested stake, where 1 unit is whatever amount you normally bet; Pass means no bet. Once a game is final, a green row means the model's pick won and red means it lost. Every result is graded at the price posted before first pitch, and no pick is ever removed afterwards.

A gap is a disagreement, not a guarantee: the market is often right, and good picks still lose often. Never bet more than you can afford to lose.

Research only: no units. Since mid-July the book's prices have been more accurate than the NPB model in every market. The totals model was updated on Oct 5, 2026; its leans are shown and graded here, but nothing is staked until it has a record of its own.

NPB Moneylines

Model win probability for its pick vs. the book's price with the vig removed. Gap is in percentage points.

2-2 · model leans
GameAway / HomeModel pickModelBookGapResult
Rakuten@ORIX
Rakuten at ORIX
+130 / -147Rakuten
55.1%
42.2%+12.9Lost ✗
Seibu@Lotte
Seibu at Lotte
-122 / +108Lotte
55.8%
46.7%+9.1Won ✓
Hanshin@DeNA
Hanshin at DeNA
-119 / +106Hanshin
57.8%
52.8%+4.9Lost ✗
Giants@Hiroshima
Giants at Hiroshima
-125 / +111Giants
50.2%
54.0%-3.8Won ✓

NPB Totals

Model probability on its side of the posted total vs. the book's no-vig price. ≈ marks an under price derived from the over price.

4-1 · model leans
GameTotalOver / UnderLeanModelBookGapResult
Hanshin@DeNA
Hanshin at DeNA
6.5-105 / ≈-112Over
59.3%
49.2%+10.0Won ✓
Yakult@Chunichi
Yakult at Chunichi
7+105 / ≈-122Under
62.5%
53.0%+9.5Won ✓
Seibu@Lotte
Seibu at Lotte
7-141 / ≈+120Over
64.9%
56.3%+8.6Won ✓
Giants@Hiroshima
Giants at Hiroshima
6.5+101 / ≈-118Over
54.5%
47.9%+6.6Lost ✗
Nippon-Ham@SoftBank
Nippon-Ham at SoftBank
7+103 / ≈-120Over
52.7%
47.4%+5.2Won ✓

NPB Team Totals

Each team's runs over or under its posted team total: model vs. the book's no-vig price.

6-3 · model leans
GameTeamTotalOver / UnderLeanModelBookGapResult
Giants@Hiroshima
Giants at Hiroshima
Hiroshima2.5-113 / ≈-107Over
60.6%
50.6%+9.9Lost ✗
Hanshin@DeNA
Hanshin at DeNA
Hanshin3.5+103 / ≈-125Over
55.6%
47.0%+8.6Won ✓
Hanshin@DeNA
Hanshin at DeNA
DeNA2.5-125 / ≈+103Over
62.2%
53.0%+9.2Won ✓
Nippon-Ham@SoftBank
Nippon-Ham at SoftBank
Nippon-Ham3.5+118 / ≈-144Under
51.0%
56.3%-5.3Lost ✗
Rakuten@ORIX
Rakuten at ORIX
Rakuten3.5+116 / ≈-142Under
51.4%
55.9%-4.5Won ✓
Rakuten@ORIX
Rakuten at ORIX
ORIX3.5-126 / ≈+104Under
60.9%
46.8%+14.1Won ✓
Seibu@Lotte
Seibu at Lotte
Lotte3.5+107 / ≈-129Over
54.8%
46.2%+8.6Won ✓
Yakult@Chunichi
Yakult at Chunichi
Yakult2.5-129 / ≈+107Under
55.8%
46.2%+9.7Won ✓
Yakult@Chunichi
Yakult at Chunichi
Chunichi3.5+108 / ≈-131Under
57.1%
54.1%+3.0Lost ✗
Model's pick wonLost