What Is Value Betting: The Edge Formula, a No-Vig Example, and Why Arbs and Matched Bets Are Different Trades

Value betting means one thing. You bet when the price pays more than the outcome's true chance deserves. Picking the winner plays no part in the definition. A bet on a 45.6% underdog at +125 is a value bet, and the bet loses 54.4% of the time.

Two other trades make money from sportsbooks without picking winners. Arbitrage locks a profit by betting both sides of a game at two books. Matched betting locks a profit out of a sportsbook promotion. Each trade makes money for a different reason, needs different inputs, and ends a different way.

This guide defines the edge, works one no-vig example from the sharp line to the stake, then runs the same game through an arb ledger and a matched-bet ledger. You see where every dollar comes from, and what each certain dollar costs.

The Edge in One Line

Every price carries a break-even probability. Convert the American odds to decimal odds d, and break-even is 1 / d. Your edge per dollar is the expected value of the bet.

Positive EV is a value bet. Zero is a fair bet. Negative EV is the normal sportsbook ticket. Pinnacle's betting guide uses the same definition: a value bet is one where the true probability beats the probability the odds imply. The odds converter handles the decimal step.

The probability gap needed for a fixed edge shrinks as the price gets longer.

Read the table as a warning. At -200 a 2% edge needs a true probability 1.33 points above break-even. At +300 the same edge sits half a point above break-even. Miss your probability by half a point on a +300 price and the edge is gone. Long prices demand the most accurate probabilities.

Where the True Probability Comes From

The formula needs p, and no book prints p. You get p one of two ways. Build a model sharper than the market, or borrow the market's own numbers.

The second route has the best public evidence. Lisandro Kaunitz, Shenjun Zhong, and Javier Kreiner set each soccer result's consensus probability at 1 divided by the average odds across up to 32 bookmakers. Their rule bet the best available price only when the price beat 1 / (consensus probability - 0.05). On a result averaging odds of 2.00, the rule bets only at prices above 2.22. Across 56,435 simulated bets at closing odds, 2005 to 2015, the rule returned 3.5%. Across 672 paper and real-money bets at $50, the rule returned 6.2%.

The rule needed no forecasting model of its own. In the authors' words, finding the bets "did not require a model with higher accuracy than the bookmakers' models." Value betting needs an accurate p. In their test, the market's own average supplied one, and the profit came from the books priced away from the average.

In U.S. markets, run the same logic through a sharp book. Use a low-margin market maker's line as the consensus. Strip the vig and the fair line is your p. The line-setting guide covers why the market maker's number leads and retail books follow.

A No-Vig Worked Example

Here is the full pipeline on one moneyline. The sharp book shows the favorite at -125 and the underdog at +115. A retail book hangs the underdog at +125.

The underdog loses 54.4% of the time, and the bet is still correct. The book pays +125 for an outcome worth +119.4. The no-vig calculator runs Step 1 on any line.

Run a second de-vig method before you trust a thin edge. The basic method above splits the margin in proportion. The power method raises each implied probability to one exponent, here 1.0306, so the pair sums to 100%. The power method takes more of the margin off the longer price, which drops the fair underdog to 45.435%. The de-vig methods guide explains when the two split apart.

Step 3 is the stake. Kelly's 1956 paper in the Bell System Technical Journal defines the exponential growth rate of a gambler's capital and solves for the bet fraction with the fastest growth. For one bet at decimal odds d, the Kelly fraction equals the edge divided by the net odds.

Stake a fraction of Kelly, because p carries estimation error. The quarter-Kelly guide and the Kelly calculator cover the choice. Size from the fair probability, never from the retail book's implied probability. When the two de-vig methods disagree, take the smaller stake.

Why a Correct Bet Still Loses for Months

A 2.53% edge on +125 underdogs is a real edge. Here is how often the bettor holding the edge sits behind, by bet count, at the fair win rate of 45.57% and 1 unit a bet. The chances come from the exact binomial distribution.

After 500 correct bets, close to one bettor in three sits in the red. Kaunitz and his coauthors ran into the same problem. After their 672 paper and real-money bets, a random strategy matched or beat their profit about 1 time in 11, a p-value of 0.089, short of the usual 0.05 bar. The decade of simulated bets cleared the bar by a wide margin. The months of paper and real-money betting did not.

Results lag the edge. Track your price against the close alongside your record. The CLV guide shows how, and the sample size guide covers how many bets a record needs.

Arbitrage: A Value Bet Plus a Hedge You Pay For

Now add a third book. Book X hangs the favorite at -120, while the retail book still shows the underdog at +125.

No probability estimate entered the arb. You need two prices and the stake split. Now score each leg against the sharp book's fair line.

The leg values change with the de-vig method. The total does not. Expected value adds across bets, and a locked result has an expected value equal to the result. The two legs sum to $10.20 at any true probability you plug in. Two consequences follow.

• Every arb holds at least one value bet. If both legs carried zero or negative EV at some true probability p, then p ≤ 1 / d1 and 1 - p ≤ 1 / d2. Add the two and the booksum reaches at least 100%. A booksum under 100% rules out the case.
• The hedge dilutes the edge. At the basic fair line, put the whole $1,000 on the +125 underdog and your expected profit is $25.316, with the swings from the last section. The arb locks $10.204, so the arb gives up $15.11 of expected profit for certainty. At the power fair line the gap is $22.283 against $10.204.

Arbs are rare when you use sportsbook prices alone. Egon Franck, Erwin Verbeek, and Stephan Nüesch studied seven seasons of the top five European soccer leagues, 2004/05 through 2010/11, with ten bookmakers and the Betfair exchange. Bookmaker prices alone produced 102 arbs, about one per 125 matches, averaging a 0.9% return. Pairing a bookmaker price with a lay bet on the exchange, a bet against the outcome, produced an arb in 19.2% of matches, averaging 1.4%. The bookmakers lost money on average on the prices behind those arbs. The arb guide covers how U.S. books detect the pattern.

Matched Betting: The Edge Is the Promo

Matched betting uses the same hedge, aimed at a promotion instead of a price error. The terms below come from a 2025 FanDuel new-customer offer: bet $5 or more and receive $300 in bonus bets only if the first bet wins. FanDuel's terms say a winning bonus bet returns the winnings, not the stake. The bonus bets expire 7 days after receipt, and the offer was one per person.

Step 1 prices the bonus. A bonus bet pays profit only, so you place the bonus at a long price and hedge the other side with cash. The example assumes +400 for the bonus and -450 for the hedge at a second book.

Step 2 hedges the $5 qualifier. The bonus arrives only if the qualifier wins, so the hedge carries the bonus value inside the ledger. Both sides sit at -110.

The $103.44 comes from the book's promo budget. No price error and no probability estimate went into the trade. Each hedge cost edge. On a fair coin flip, the unhedged qualifier was worth $108.86, so the hedge cost $5.42, the $119.29 hedge stake times the 4.545% expected loss at -110. Against the fair line of the +400 and -450 pair, the unhedged bonus was worth $235.71, so the conversion hedge cost another $17.53.

Then the trade ends. One offer per person means one ledger per book. The promo categories guide prices other bonus bet sizes and odds.

Three Trades Side by Side

The last row comes from one document. Penn Sports Interactive's Massachusetts house rules for ESPN BET, posted by the Massachusetts Gaming Commission and effective 5/23/2025, cover all three trades. Rule 2.12 lets the book set minimum and maximum wager limits patron by patron. Rule 2.2(d) lists engaging in arbitrage as a reason to refuse a wager. Rule 2.18 lets the book restrict, suspend, or limit a promotion patron by patron.

Kaunitz and his coauthors ran into stake limits in practice. A few months into real-money betting, bookmakers capped their stakes below the $50 bet size or refused the bets, and the experiment ended. The limits guide covers what books watch.

Calculator: Value and Arb Checker

Enter the sharp book's two prices, then the best price you see on each side at any book. The checker de-vigs the sharp line both ways, scores each price, sizes any bet positive under both methods at both fair probabilities, and flags an arb when the best prices sum under 100%. The defaults reproduce the examples above.

How to Use the Three Trades

• De-vig the sharpest line you see, then compare every retail price to the fair odds. Bet a price only when the edge stays positive under both de-vig methods.
• Size each value bet with a fraction of Kelly from the fair probability. Expect months of red even with a real edge.
• Take an arb when the prices line up, and remember the hedge leg trades expected profit for certainty. The value leg alone pays more on average.
• Work each promo once, hedged, and count the cash the hedges tie up before you start.
• Keep a record of the price you took and the closing price on every bet. The record tells you which trade you ran.

Want to see how the calls get made? Look around the free Discord room first. No card needed.

Sources and Method

Kaunitz, Zhong, and Kreiner (2017, arXiv 1710.02824) supply the consensus-odds strategy, the 0.05 margin rule, the 3.5% and 6.2% returns, the 0.089 p-value, and the account limits. Franck, Verbeek, and Nüesch (Economica, 2013) supply the arbitrage frequencies and average returns. Kelly (Bell System Technical Journal, 1956) supplies the growth-maximizing stake. FanDuel's 2025 promotion page supplies the bonus bet terms: $5 qualifier, $300 in bonus bets on a win, stake not returned, 7-day expiry, one per person. Penn Sports Interactive's Massachusetts house rules supply rules 2.2(d), 2.12, and 2.18. Pinnacle's betting guide supplies the value bet definition.

Every other number comes from the formulas shown, and you recompute each one in a spreadsheet. Decimal odds come from 1 + 100 / |odds| for favorites and 1 + odds / 100 for underdogs. The basic de-vig divides each implied probability by the booksum. The power method solves for the exponent k with both implied probabilities raised to k summing to 1. The chance of being behind uses the exact binomial with BINOM.DIST. Arb stakes split the bankroll by each leg's implied probability over the booksum. The matched-bet hedge solves for the stake equalizing both outcomes.

Limits: the sharp line, the retail prices, and the +400 and -450 hedge pair are illustrations, not quotes from a live market. The FanDuel offer ended in 2025, and current offers carry their own terms. Real arbs and hedges also carry fill risk: a book refuses or cuts one leg after you place the other.