Edge Scout · Probability and value-bet research engine
Admin entry →Results →Model lab →

Find markets where your model beats the book.

Edge Scout compares model probability against no-vig market probability, then ranks the biggest edges. The goal is calibrated decision support — not guaranteed betting picks.

Core formula

Edge = Model Probability − No‑Vig Market Probability

Positive edge means the model estimates a higher chance than the market after bookmaker margin is removed.

Tracked metrics

CLV · Brier · Log Loss

MVP model

Poisson xG → 1X2

Next model layer

Dixon‑Coles adjustment

Top value

Ranked model edges

Sample slate

Premier League · Today · 20:00

Liverpool

Aston Villa vs Liverpool

+39.0%

Model

84.9%

No-vig

45.9%

Odds

2.05

Chinese Super League · Today · 13:35

Beijing Guoan

Beijing Guoan vs Qingdao Hainiu

+35.6%

Model

99.1%

No-vig

63.5%

Odds

1.48

La Liga · Today · 18:30

Valencia

Valencia vs Rayo Vallecano

+17.2%

Model

60.8%

No-vig

43.6%

Odds

2.18

Market board

1X2 probability comparison

Premier League · Today · 20:00

Aston Villa vs Liverpool

Historical strength → Poisson xG 0.47–2.77 · no-vig adjusted

Aston Villa

-25.3%
Model
4.1%
Market
29.4%
Current odds
3.20

Draw

-13.8%
Model
11.0%
Market
24.7%
Current odds
3.80

Liverpool

+39.0%
Model
84.9%
Market
45.9%
Current odds
2.05

Chinese Super League · Today · 13:35

Beijing Guoan vs Qingdao Hainiu

Historical strength → Poisson xG 4.74–0.00 · no-vig adjusted

Beijing Guoan

+35.6%
Model
99.1%
Market
63.5%
Current odds
1.48

Draw

-20.5%
Model
0.9%
Market
21.4%
Current odds
4.40

Qingdao Hainiu

-15.2%
Model
0.0%
Market
15.2%
Current odds
6.20

La Liga · Today · 18:30

Valencia vs Rayo Vallecano

Historical strength → Poisson xG 1.58–0.62 · no-vig adjusted

Valencia

+17.2%
Model
60.8%
Market
43.6%
Current odds
2.18

Draw

-4.8%
Model
24.9%
Market
29.7%
Current odds
3.20

Rayo Vallecano

-12.5%
Model
14.3%
Market
26.8%
Current odds
3.55

Risk note

Edge Scout is a research and probability-calibration tool. It does not guarantee wins, does not remove variance, and should not be treated as financial advice. Future versions should track closing line value, Brier Score, Log Loss, and calibration buckets before any staking model is trusted.