back to terminal

betmind / docs

Methodology

Last updated: 2026-05-26 · v1.0.0

1. Data sources

Pinnacle

Primary sharp book reference. No-vig lines used for true probability.

BetCRIS

Latin American sharp market. Cross-validated against Pinnacle.

Circa Sports

US sharp book. Used for NFL, NBA lines.

Betfair Exchange

Peer-to-peer prices for liquidity confirmation.

ESPN / Stats Perform

Injury reports, lineup data, historical stats.

The Action Network

Sharp money percentages, public betting splits.

SportRadar

Live event data, odds feeds (7 sources in status bar).

2. EV calculation

Formula:

EV% = (p_true × decimal_odds − 1) × 100

p_true = 1 / (line_home_no_vig)
line_no_vig = devigged from Pinnacle opening

Devigging uses the multiplicative method. We take Pinnacle opening lines (within 2min of market open) before sharp action moves them. This gives the cleanest signal.

3. CLV (Closing Line Value)

CLV% = (bet_odds / close_odds − 1) × 100

CLV measures whether you beat the market at close. A bettor with consistent positive CLV has a genuine edge, regardless of short-term results. Target: CLV > +1.5% over 500+ bets.

4. Sharp % signal

"Sharp %" is derived from: (line movement direction matching sharp book consensus) × (reverse line movement flag) × (steam move detection within 90s). Normalized to 0–100 scale.

5. Threshold for publication

A pick is surfaced when: EV > +2.0% AND sharp% > 55 AND odds available at 3+ books. Parlays are not recommended (edge degrades multiplicatively).

6. Known limitations

  • Closing line data has ~30s latency from live market.
  • Injury data from public sources. Late scratches may not be reflected.
  • Odds on display may be 15min delayed for free accounts.
  • Model does not account for correlated parlays.
  • Small sample warning: EV is meaningless over < 200 bets.

Questions or corrections → /transparency