EVScout Methodology

Every sportsbook price implies a probability. American odds convert directly: -110 implies 52.4%, +150 implies 40%, +300 implies 25%. This is the win rate at which the bet would break even at that price.

Example: a team priced at -150 implies a 60% chance. If you believe — or the market suggests — the real chance is higher, the price is attractive. If lower, it isn't. Implied probability is just the price restated; it says nothing by itself about what will happen.

Sportsbooks build a margin into every market. Add up the implied probabilities of every outcome in a two-way market and the total is almost always above 100%. A market priced -110 / -110 totals 104.8%. That surplus — 4.8% here — is the estimated sportsbook margin, often called the vig or hold.

Because of the vig, raw implied probabilities overstate each outcome's likelihood. Using them directly as "chances" would make nearly every bet look worse than it is.

EVScout removes the estimated margin to produce an Estimated Fair Market Probability — what the implied probability would be if the market had no built-in margin. We scale the implied probabilities proportionally back to 100% (the multiplicative method).

Example: in a -110 / -110 market, each side's 52.4% scales down to 50%. That 50% is the estimated fair market probability — an estimate derived from market prices, not a "true probability." The market can be wrong, and so can this estimate.

The estimated fair market probability converts back into an Estimated Fair Odds price — the odds that would correspond to a break-even bet at that probability. A fair probability of 50% is fair odds of +100; 60% is fair odds of -150. Comparing the sportsbook's price to the estimated fair odds shows how much of the offered price is margin.

Estimated EV compares the price you can actually get against the estimated fair price: (estimated fair probability × offered decimal odds) − 1, expressed as a percentage of your stake. An estimated EV of +5% means that, if the fair-price estimate is right, a bettor making this bet repeatedly would average a 5% return on stake over the long run.

Positive estimated EV does not mean a bet will win — most individual +EV bets still lose. Negative estimated EV does not mean a bet cannot win. EV is a statement about the price, not a prediction about the outcome.

In a parlay, leg prices multiply: combined decimal odds are the product of each leg's decimal odds, and the combined probability is the product of each leg's probability. That multiplication also compounds the vig, which is why parlays usually carry a much larger effective margin than single bets. EVScout computes the parlay's combined price, combined estimated fair probability, and overall estimated EV, and shows each leg's individual price quality alongside it.

Multiplying leg probabilities assumes the legs are independent — that one outcome tells you nothing about another. Legs from the same game often violate this: a team winning and its star scorer going over a points total tend to happen together. When legs are correlated, the true combined probability differs from the independent estimate.

EVScout flags same-game legs and correlation warnings when detected. When we cannot reliably quantify the adjustment, we say so explicitly rather than inventing a precise number. Flagged correlation means treat the combined figures as estimates, not that the parlay is good or bad.

Prices come from a licensed odds data provider covering major US sportsbooks. Where a sharp, low-margin book (such as Pinnacle or Circa) quotes the same market, we anchor the fair estimate to it, because those markets tend to be the closest available reference for a fair price. When no sharp price is available, we de-vig the consensus of the books we can see.

Every evaluation shows how fresh each underlying price was when computed and how many sportsbooks contributed to the market consensus. Stale prices are flagged, and results computed on thin market coverage are labeled with lower confidence — High, Moderate, or Limited — so you can see when a number rests on fewer sources.

The letter grade is a plain label attached to the computed estimated EV — nothing more. There is no separate model, adjustment, or discretion behind it: the same estimated EV always produces the same grade. The thresholds below are read directly from the evaluation engine.

EVScout Grade Model v1.0

AI does not determine your grade. Every number on the results screen — odds, probabilities, estimated EV, fair prices, the grade itself — comes from the deterministic calculations described above.

The AI layer's only job is to explain: it summarizes what's weakening the parlay, describes possible improvements, and translates the quantitative results into plain English. It receives only the figures already computed and is instructed never to introduce odds, probabilities, statistics, injuries, or predictions of its own — and it cannot modify any calculated value. If the plain-English explanation is unavailable, you still get the complete calculated analysis.

• Odds change rapidly; a price shown minutes ago may no longer be available.
• Market consensus can be wrong — the crowd's price is not a guarantee of fairness.
• Estimated fair probability remains an estimate, not the actual chance of the outcome.
• Correlation between legs can be difficult or impossible to quantify reliably.
• Betting outcomes contain randomness; short-run results will deviate from any estimate.
• Manually entered legs have no market to de-vig against.
• EVScout does not guarantee profitability. No method can.