How Edge of the Day works

Edge of the Day is the single highest expected-value (EV) bet found by LineGoat's automated morning scan. The scan compares Pinnacle's no-vig fair odds with the best price at nine US sportsbooks on the same game, market and line. If a US book pays enough more than the fair odds to land between 1% and 12% EV, the scan counts it as an edge. The biggest one is published automatically as Edge of the Day, and every published edge goes on the public Edge Tracker with its result and, for the top three, its closing line value (CLV).

Edge of the Day is a math estimate. A +EV bet is a long-run expectation over many bets, and any single +EV bet can lose. Nothing on this page is a promise of a win or a profit.

By The Goat, LineGoat's automated desk. No human picks or edits the edges. This page was written with AI writing tools and checked line by line against the scan's code.

Pinnacle is the sharp reference book for Edge of the Day. It takes big limits from sharp bettors, so its price is a strong estimate of the true odds. For each game, the scan takes Pinnacle's price on both sides of a moneyline or spread. It only uses markets with exactly two outcomes, so the two sides can be devigged against each other. If Pinnacle has no price on a game, the scan skips that game.

The vig is the bookmaker's margin. On a two-sided market, the two implied probabilities add up to more than 100%, and the extra is the vig. LineGoat removes it with the multiplicative method: convert each side to an implied probability, then divide each by the sum of both.

• Decimal odds from American odds: positive odds = 1 + odds / 100; negative odds = 1 + 100 / |odds|
• Implied probability = 1 / decimal odds
• Fair probability = implied probability of the side / (implied probability of side A + implied probability of side B)

The fair probability is then converted back to American odds. That's the "Fair odds (no-vig)" number on the Edge of the Day card.

Worked example (real numbers from the October 16, 2026 scan). Pinnacle priced Georgia State at +306 and James Madison at -391.

• Georgia State +306 = 4.06 decimal, implied probability 1 / 4.06 = 24.63%
• James Madison -391 = 1.2558 decimal, implied probability 79.63%
• Sum = 104.26%, so Pinnacle's overround on this market was about 4.26%
• Georgia State fair probability = 24.63% / 104.26% = 23.62%
• Fair odds = 1 / 0.2362 = 4.233 decimal = +323.3 American

The scan checks the same selection, on the same market and the same line, at nine US sportsbooks. For spreads, the US book has to offer the same number as Pinnacle (for example +3.5 against +3.5). The scan takes the highest payout among those books.

Three filters apply here. A US book's market is skipped if it was last updated more than 90 minutes before the scan. The best price has to sit between -400 and +500. The game has to start at least 10 minutes after the scan.

EV is the expected return per unit staked if the fair probability is right.

EV = fair probability × decimal odds at the best US book − 1

Continuing the example: Hard Rock Bet offered Georgia State at +350, which is 4.50 in decimal odds. EV = 0.2362 × 4.50 − 1 = 0.063, or +6.3% EV.

In plain terms, if Pinnacle's no-vig probability is accurate, a long run of bets priced like this one would return about 6.3 cents per dollar staked on average. That's an average over many bets. Georgia State at +350 still loses about three times out of four by the same math.

The scan keeps an edge only if EV is between 1% and 12%. Edges above 12% are dropped because a gap that large usually points to a stale or wrong line rather than a real price.

The scan sorts every qualifying edge from highest EV to lowest, then builds the board by taking the top edge from each game, one per game, up to six games.

• Edge of the Day is the first edge on that board: the highest EV found in the scan.
• Runners-up are the next five board edges, one per game.
• The NFL Desk and NBA Desk each list up to eight edges from their sport, one per game and market.

If nothing clears 1% EV, there is no Edge of the Day. The card says so, and the scan doesn't lower the bar to force a bet.

The scan covers at most three sports per run. Sports are checked in a fixed order (NFL, NBA, college football, NHL, MLB), and the first three with games in the next 28 hours get scanned. We pay for odds data by the request, so a budget guard cuts the scan to two sports, or one, when the month's remaining data credits run low.

Every published edge is logged to the Edge Tracker the moment it's published, with the price and the book. Edges are graded each morning at about 7:00am Arizona time, before the new scan, from final scores.

• An edge is graded once its game started more than 3.5 hours earlier and the final score is in.
• Moneylines: the side with more points wins. A tie is graded a push.
• Spreads: the selection's score plus the spread, compared with the opponent's score. Landing exactly on the number is a push.
• Units assume a flat 1-unit stake at the published price. A win pays the published odds (for example +1.25 units at +125), a loss is −1 unit, and a push is 0.
• If no final score turns up within 4 days of the scheduled start, the edge is graded void (0 units).

Edges are graded at the price we published. Prices move after the scan, so the price you can get later may be different.

Closing line value measures whether the published price beat the sharp market's last price before the game. We record CLV for The Goat's top three edges each day: Edge of the Day and the first two runners-up.

1. Every 15 minutes, a check looks for top-three edges whose game starts within the next 20 minutes.
2. For each one, it pulls Pinnacle's current two-sided price on that market. That's the "closing line" for our purposes: the last price captured shortly before the scheduled start, usually 5 to 20 minutes before.
3. The closing price is devigged the same way as in Step 2.
4. CLV = closing fair probability × published decimal odds − 1.

If Pinnacle moved a spread to a different number by then, the tracker notes the new line (for example "line moved to +4") and doesn't compute CLV, because the two prices are no longer on the same bet. If Pinnacle has no price at that point, the tracker shows "no closing line".

Worked CLV example (made-up close, for illustration only). Suppose Georgia State closed at Pinnacle +320 / James Madison -410. Implied probabilities are 23.81% and 80.39%, which sum to 104.20%. Georgia State's closing fair probability is 23.81% / 104.20% = 22.85% (fair odds +338). CLV = 0.2285 × 4.50 − 1 = +2.8%. The published +350 would still beat the closing fair price.

Why track CLV: final scores are noisy over small samples, while every bet gets a CLV number whether it wins or loses. An analysis of 87,960 Pinnacle soccer odds pairs found that the ratio of pre-closing to closing odds tracked actual returns almost 1:1 (football-data.co.uk, 2016). That study covers soccer, not US sports, and CLV is still an estimate. Positive CLV over a small number of bets says little. What CLV is, with more examples →

The Edge Tracker lists every edge the scan has published, from the Edge of the Day board and the desk lists. For each edge it shows:

• Date, desk and the bet (sport, game, market, selection)
• The published price and the book, and EV at publish
• Pinnacle's no-vig closing odds and CLV (top three edges only; others show "not tracked")
• Result (win, loss, push, void or pending), the final score and units at a flat 1-unit stake

The summary adds totals: edges published, record, average CLV, average EV at publish and pending count. Units and ROI appear once at least one edge has been graded. Results are not deleted.

• Whether you could get the price. The tracker records the price at scan time. Lines move, and books may limit or close accounts.
• Every book or every state. The scan covers nine US books. A book may not operate in your state.
• Totals, player props or live betting. The scan covers pregame moneylines and spreads only.
• CLV for every edge. Only the daily top three get a closing line.
• Every sport every day. At most three sports are scanned per day, so some days skip NHL or MLB even when they have games.
• A promise. EV and CLV are estimates built on Pinnacle's prices.

• Pinnacle can be wrong too. Its price is treated as the best available estimate of the true odds, not the truth.
• The devig method favors underdogs. The multiplicative method splits the vig in proportion to each side's implied probability. Other common methods (additive, power, Shin) take more of the vig off the underdog, so they give it a lower fair probability. The gap is biggest when Pinnacle's margin is wide. In the Georgia State example, with a 4.26% margin, the additive method gives a fair probability of 22.5% instead of 23.6%, and the EV at +350 drops from 6.3% to about 1.2%. That means our scan can rank longshot underdogs higher than other methods would.
• Longshot EV is more sensitive. A one-point error in fair probability moves EV by 4.5 points at +350, but under 2 points at -150. Picking the highest EV each day leans toward bigger prices for that reason alone.
• The scan runs once a day. An edge can disappear minutes after the scan.

What is Edge of the Day?

Edge of the Day is the highest-EV bet from LineGoat's daily automated scan, found by comparing Pinnacle's no-vig fair odds with the best price at nine US sportsbooks. It's published automatically each morning around 7:00am Arizona time.

How is EV calculated?

EV is fair probability × decimal odds − 1. The fair probability comes from removing the vig from Pinnacle's two-sided price, and the odds are the best of nine US books. Edges from 1% to 12% EV qualify.

Does Edge of the Day always win?

No. Edge of the Day is a long-run math estimate. A single +EV bet often loses, and short-term results swing a lot.

Why is there sometimes no Edge of the Day?

Edge of the Day needs at least 1% EV at a price between -400 and +500. If nothing clears that bar, the card says so and the scan doesn't force a bet.

Is Edge of the Day a human pick?

No. The Goat, the NFL Desk and the NBA Desk are names for our automated scan. No tipster, handicapper or editor picks or reviews the edges before they go up. Meet the desks.