The process in one sentence
Estimate a fair price, compare available market prices, and track whether qualified alerts beat the close.
Implied probability
American odds can be converted into the probability represented by the sportsbook price.
implied probability = 100 / (odds + 100)implied probability = abs(odds) / (abs(odds) + 100)Sportsbook vig
Sportsbooks build margin into a market. When every outcome is converted to raw implied probability, the total usually exceeds 100%.
No-vig fair probability
LineRaptor removes the reference market's margin by normalizing all raw outcome probabilities so they sum to 100%.
no-vig fair probability = outcome raw implied probability / sum of market probabilitiesQualified MVP paper bets require a complete, valid Pinnacle market. If the reference is incomplete or invalid, the scanner does not force an alert.
Reference-First Hybrid Model
LineRaptor uses a reference-first hybrid model. Pinnacle's no-vig probability remains the primary fair-price estimate, while independently devigged prices from other complete sportsbook markets form a separate median consensus.
When the reference and consensus agree that an offered price is positive EV, the alert receives a higher confidence label. If consensus disagrees, LineRaptor stores the signal for internal analysis but withholds it from normal customer alerts. Consensus-only alerts remain experimental and disabled by default.
Supported Sportsbook Feed
The default beta feed focuses on supported U.S.-facing sportsbook prices for alertable opportunities. Pinnacle remains a reference source for fair-price estimation, but it is not treated as a normal customer-actionable sportsbook.
Offshore or unsupported books are excluded from default customer alerts and default consensus calculations unless a future advanced/internal setting explicitly enables them.
Edge
Edge is the gap between the estimated fair probability and the probability implied by the offered sportsbook price.
edge = fair probability - offered implied probabilityExpected value per 1u
EV estimates the theoretical long-term value of risking 1 betting unit if the fair probability estimate is accurate. It is not a one-bet profit forecast.
EV = fair probability × win units - (1 - fair probability) × 1uPositive odds: profit = odds / 100 units
Negative odds: profit = 100 / abs(odds) unitsCurrent qualification rules
Not every positive estimate becomes a qualified alert. These are current beta settings, not permanent promises.
- Minimum edge: 1.5%
- Minimum EV per 1u: +0.03u
- At least three comparison books
- Pinnacle reference market required
- Complete reference market required
- Exact point matching for spreads and totals
- Positive odds capped at +1200
- Negative odds capped at -1000
- Acceptable reference hold
- At least five minutes before event start
- Best available sportsbook price only
Closing-line value
CLV compares the price at alert time with the price near market close or event start.
CLV probability difference = closing implied probability - opening implied probabilityPositive CLV generally means the alert beat the closing price. A bet can lose and still have positive CLV, or win while having poor CLV. It is a validation signal, not proof of guaranteed profitability.
Paper tracking
LineRaptor records each alert as if 1 unit were risked, without placing a real bet. It tracks opening odds, closing odds, CLV, result, paper units won or lost, EV at detection, closing EV when available, and whether the alert beat the closing line.
Variance and limitations
Mathematically promising opportunities can still lose. The objective is not to win every alert; it is to identify prices that appear better than fair value over a meaningful sample.
- Odds can move quickly or disappear.
- Data may be delayed or unavailable.
- Fair probability is an estimate and reference books are imperfect.
- Small samples are noisy.
- Positive EV does not guarantee profit.
- Users remain responsible for their decisions.
- LineRaptor cannot account for every real-world factor.
Estimate fair price. Compare the market. Track the close.
That is the entire LineRaptor loop. The value is in applying it consistently and measuring the evidence honestly.
Continue to the FAQ