How do you read betting odds?

Decimal odds show your total return per unit staked, so 2.50 returns $25 from a $10 bet. Fractional odds show profit to stake, so 6/4 wins $6 for every $4. American odds show profit on $100 when positive and the stake needed to win $100 when negative.

How do I calculate implied probability?

Divide 1 by the decimal odds. Odds of 2.50 imply a 40% chance and odds of 4.00 imply 25%. For American odds, use 100 divided by the odds plus 100 for positive prices, and the odds divided by the odds plus 100 for negative prices.

How do I convert American odds to decimal?

For positive odds, divide by 100 and add 1, so +150 becomes 2.50. For negative odds, divide 100 by the number and add 1, so -125 becomes 1.80.

What is the difference between overround and hold?

Overround is how far the implied probabilities in a market exceed 100%. Theoretical hold is the share of stakes the bookmaker keeps if money is balanced, which is smaller. A market at -110 on both sides has a 4.76% overround and a 4.55% theoretical hold.

How do I remove the bookmaker’s margin?

Convert each price to an implied probability, add them up, then divide each probability by the total. One divided by each adjusted probability gives the fair odds. This proportional method is a good approximation, though bookmakers often load more margin onto long shots.

What is value betting?

Betting only when the true probability of an outcome is higher than the probability implied by the odds. Over many bets, consistently finding value is the only way to beat the margin.

What is closing line value?

The difference between the price you took and the final price before the event started. Regularly beating the closing price is strong evidence that you are finding value, even before your results show it.

Are accumulators a good bet?

They can be fun, but they multiply the margin across every leg. Four legs with a 5% overround each give an effective overround of about 21.6%, so the long-run cost is much higher than for single bets.

Are Kalshi and Polymarket legal for sports?

They operate as CFTC-regulated exchanges, but in 2026 many states argue their sports contracts are illegal betting. Federal appeals courts have split, court orders restrict access in states such as Nevada, Massachusetts and Michigan, and the issue may reach the Supreme Court. Check availability in your state before using one.

How do prediction market prices relate to odds?

A contract price in cents is an implied probability, so 40 cents means 40% and is equivalent to decimal odds of 2.50 before fees. Add trading fees and the bid-ask spread before comparing with a sportsbook price.

How Betting Odds Work: Formats, Margin & Value

Key takeaways

All formats say the same thing. Implied probability = 1 ÷ decimal odds.
Add up the implied probabilities of every outcome: the excess over 100% is the overround. Hold is the share of stakes the bookmaker expects to keep, which is smaller.
Divide each implied probability by the total to remove the margin and get a fair price.
A good bet is one priced above its true chance. Beating the closing line is the best everyday evidence that you are finding them.
Prediction markets quote prices in cents that work like probabilities, but in 2026 several US states and courts are challenging whether their sports contracts are gambling.

Three formats, one meaning

Every odds format answers the same question: how much do you get back if the bet wins? They just express it differently.

Decimal (2.50): your total return per unit staked, including the stake. A $10 bet at 2.50 returns $25, which is $15 profit plus your $10. Standard in Europe, Australia, Canada and at crypto sportsbooks, and our default.
Fractional (6/4): profit relative to stake. A $4 bet at 6/4 wins $6 profit, and you also get your $4 back. Traditional in Britain and Ireland, especially in horse racing.
American (+150 / −200): a positive number is the profit on a $100 stake; a negative number is the stake needed to win $100 profit. Standard at US sportsbooks.

A fourth format matters more every year: prediction market prices, quoted in cents between 0 and 100. A contract at 40¢ pays $1 if the event happens, so its price is directly an implied probability of 40%, equivalent to decimal 2.50 before fees.

Odds conversion table

Every row below was checked with the formulas in the next section. Decimal values are exact except 1.91, which is rounded from 1.909.

The same prices in each format. Implied probabilities are rounded to one decimal place.
Decimal	Fractional	American	Implied probability
1.20	1/5	−500	83.3%
1.25	1/4	−400	80.0%
1.50	1/2	−200	66.7%
1.80	4/5	−125	55.6%
1.91	10/11	−110	52.4%
2.00	1/1 (evens)	+100	50.0%
2.50	6/4	+150	40.0%
2.75	7/4	+175	36.4%
3.00	2/1	+200	33.3%
4.00	3/1	+300	25.0%
5.00	4/1	+400	20.0%
6.00	5/1	+500	16.7%
11.00	10/1	+1000	9.1%

Notice the asymmetry around evens. Prices shorter than 2.00 become negative American odds, and prices longer than 2.00 become positive ones. The two are never both on the same side of 2.00, which is why +100 and −100 describe the same price.

Converting between formats

You rarely need to convert by hand, because most sportsbooks let you switch format in settings. But knowing the formulas makes it easy to check a price or compare a US book with a European one.

From	To decimal	To implied probability
Decimal d	Already decimal	1 ÷ d
Fractional a/b	1 + a ÷ b	b ÷ (a + b)
American +A	1 + A ÷ 100	100 ÷ (A + 100)
American −A	1 + 100 ÷ A	A ÷ (A + 100)
Prediction market price c (in cents)	100 ÷ c	c ÷ 100

To go from decimal to American: if the decimal price is 2.00 or more, American odds are +(d − 1) × 100; if it is below 2.00, they are −100 ÷ (d − 1). So 2.75 becomes +175, and 1.80 becomes −100 ÷ 0.80 = −125.

Worked example: comparing a US price with a decimal price

A US book offers −150 and a crypto book offers 1.70 on the same team. Convert −150 to decimal: 1 + 100 ÷ 150 = 1.667. The crypto book’s 1.70 is the better price. In probability terms, −150 implies 150 ÷ 250 = 60.0%, and 1.70 implies 58.8%: you need the team to win less often to break even at 1.70.

From odds to probability

Every price encodes a probability. For decimal odds, divide 1 by the odds: 2.50 implies 40%. For American odds, +150 implies 100 ÷ 250 = 40%, and −200 implies 200 ÷ 300 = 66.7%.

This number is your break-even point. At 2.50 you need the bet to win more than 40% of the time to profit over many bets. At −110, the standard price on US point spreads and totals, you need 52.4%.

Thinking in probabilities is the most useful habit in betting. It turns “is this a good bet?” into a concrete question: do I think this outcome happens more often than the odds suggest? It also explains why favourites are not automatically safer bets: a short price simply demands a higher win rate.

Overround: the margin in the prices

Add up the implied probabilities of every outcome in a market and you will get more than 100%. The excess is the overround, also called the margin, vig or juice: the bookmaker’s built-in edge.

Worked example: working out an overround

A tennis match priced at 1.80 and 2.00 implies 55.6% + 50.0% = 105.6%. The overround is about 5.6%. At 1.90 and 1.95, it is 52.6% + 51.3% = 103.9%: a noticeably cheaper market.

Margins vary by bookmaker, sport and market. Top football leagues and major US sports main lines are usually cheapest; niche leagues, player props, same-game parlays and long-term outrights are usually most expensive. Comparing one market across several books before betting is the quickest way to save money systematically, and it is why our sportsbook rankings weight odds quality.

Overround versus hold

People often use “margin” and “hold” interchangeably, but they measure different things.

Overround is a property of the prices: how far the implied probabilities exceed 100%.
Theoretical hold is the share of total stakes the bookmaker keeps if money is balanced in proportion to the prices: 1 − 1 ÷ (sum of implied probabilities).
Actual hold is what the bookmaker really kept over a period: win ÷ handle. It depends on results and on how much customers bet on high-margin products.

Worked example: −110 on both sides

Both sides of an NFL spread are priced at −110 (1.909). Each implies 52.38%, so the book sums to 104.76% and the overround is 4.76%. Theoretical hold is 1 − 1 ÷ 1.0476 = 4.55%. Check it with money: a customer stakes $110 on each side, $220 in total. One side wins and is paid $210 back. The book keeps $10, and $10 ÷ $220 = 4.55%.

Actual hold is usually much higher than the theoretical hold on a single market. Nevada’s sportsbooks took $8.07 billion in bets in calendar 2025 and kept $601.4 million, a hold of about 7.5%, with mobile books near 7% and retail books at 8.6%, according to regulator figures reported by InGame. That gap comes from parlays, props and outrights with bigger margins, and from ordinary variance in results.

Removing the margin to find a fair price

To judge whether a price is good, you need an estimate of the true probability. A good starting point is the market itself with the margin taken out. The simplest method, proportional normalisation, divides each implied probability by the total.

Worked example: margin-free prices in a three-way market

A football match is priced home 2.10, draw 3.40, away 3.60. Implied probabilities are 47.62%, 29.41% and 27.78%, which sum to 104.81%: an overround of 4.81%. Divide each by 1.0481: home 45.43%, draw 28.06%, away 26.50%. Fair decimal odds are 1 ÷ 0.4543 = 2.20, 3.56 and 3.77. If another book offers home at 2.30, your expected return at the fair probability is 0.4543 × 2.30 − 1 ≈ +4.5% of stake.

Proportional normalisation assumes the margin is spread evenly. In practice bookmakers usually load more margin onto long shots, a pattern called the favourite–longshot bias. More refined methods, such as the power or Shin methods, shift more of the margin onto the outsiders. For everyday use, proportional is close enough on short-priced markets; on markets with big outsiders, treat its fair prices for long shots as slightly generous.

The fairest reference prices come from the sharpest books and exchanges, which accept large bets from professionals and move quickly. Removing the margin from those prices gives a much better estimate of the true probability than doing the same with a recreational book’s prices.

Value beats winners

A profitable bettor is not someone who picks winners. It is someone who bets when the true probability of an outcome is higher than the probability implied by the odds. Expected value per unit staked is probability × decimal odds − 1.

Backing a true 50% chance at 2.20 loses half the time but returns 0.50 × 2.20 − 1 = +10% of stake on average.
Backing a true 80% favourite at 1.10 wins most of the time but returns 0.80 × 1.10 − 1 = −12% on average, because 1.10 implies 90.9%.

Judge every bet by the price you took, not the result. A good bet can lose and a bad bet can win; over hundreds of bets, price is what decides the outcome. That is also why short-term results tell you very little: even a genuine 3% edge can sit behind a losing record for months.

Closing line value

The closing line is the last price before an event starts. By then the market has absorbed team news, injuries and sharp money, so the closing price, with the margin removed, is widely treated as the best available estimate of the true probability.

Closing line value (CLV) compares the price you took with that closing price. If you consistently get better prices than the close, you are probably finding value, even before results confirm it. If you consistently get worse prices, you are probably losing to the margin, even during a lucky winning run.

Worked example: measuring CLV

You back a team at 2.20 on Tuesday. By kick-off the price has shortened to 2.10, and the margin-free closing price is 2.05, implying 48.8%. Your estimated edge is 2.20 ÷ 2.05 − 1 ≈ 7.3%, or equivalently 0.488 × 2.20 − 1 ≈ +7.3%. Whether this bet wins or loses, you took a price the market later judged too generous.

CLV is a tool for evaluating your process, not a guarantee. It is most reliable in liquid markets such as major football leagues, the NFL and the NBA, and least reliable in thin markets where prices move on small bets. Be aware too that some sportsbooks limit stakes for customers who regularly beat the close.

Accumulators, bet builders and boosts

Combining selections multiplies the odds, and it multiplies the margin too. If each of four legs carries a 5% overround, the combined price is divided by 1.05⁴ = 1.2155 compared with fair odds, so the accumulator’s effective overround is about 21.6% and you get back around 82p in expected value for every pound staked.

Bet builders and same-game parlays on a single match often carry higher margins still, because the outcomes are correlated and the book prices that correlation conservatively. That is a large part of why US actual hold is far higher than the 4.55% theoretical hold on a −110 line.

Odds boosts can genuinely reduce the margin on a specific bet, so compare the boosted price with the margin-free price at other books before assuming it is good value. Our bonuses hub and bonus guide explain how betting offers and their wagering terms work.

Betting exchanges

On a betting exchange, customers bet against each other instead of against a bookmaker. You can back an outcome (bet that it happens) or lay it (bet that it does not), and the exchange charges a commission on net winnings in each market instead of building a margin into the prices.

Because the back and lay prices sit close together on liquid markets, exchange prices are often near fair value, which makes them a useful reference even if you bet elsewhere.

Worked example: backing and laying

You back an outcome at 3.00 for £10 and it wins. Your profit before commission is £20. With a 2% commission on net winnings, you pay £0.40 and keep £19.60. Your effective decimal price is 1 + 2 × 0.98 = 2.96. If you instead lay at 3.00 for a £10 backer’s stake, you win £10 if the outcome fails but your liability is £20 if it happens. Commission rates differ by exchange and market; check the current rate before betting.

Exchanges are available in Britain, Ireland, Australia and parts of Europe. In most US states, sports betting means a licensed sportsbook, which is why prediction markets have become the closest US equivalent.

Prediction markets: Kalshi, Polymarket and the 2026 state fight

Prediction markets list event contracts: yes-or-no positions on a real-world outcome that typically pay $1 if it happens and nothing if it does not. In the US they operate as federally regulated exchanges, designated contract markets overseen by the Commodity Futures Trading Commission (CFTC), rather than as state-licensed sportsbooks.

Reading a prediction market price

A “Yes” price of 62¢ implies a 62% chance and is equivalent to decimal 1 ÷ 0.62 = 1.61 before fees. The margin appears in two places. First, the spread: if Yes costs 62¢ and No costs 40¢ to buy, together they cost 102¢ for a guaranteed $1 payout, an overround of 2%. Second, trading fees, which platforms charge per contract and publish in their fee schedules.

Worked example: converting a contract price

You buy 100 Yes contracts at 62¢: $62. Suppose the platform’s fee on this trade is $1.65, so your total cost is $63.65. If Yes wins, you receive $100, a profit of $36.35, and your effective decimal price is 100 ÷ 63.65 = 1.57. A sportsbook offering 1.65 on the same outcome would be the better price. Always include fees when comparing.

Where the legal fight stands in October 2026

Kalshi and Polymarket’s US exchange argue that federal commodities law gives the CFTC exclusive authority over their contracts. A growing number of states say their sports contracts are simply unlicensed sports betting. The federal appeals courts have split:

Third Circuit (New Jersey): ruled 2–1 for Kalshi in April 2026.
Ninth Circuit (Nevada): ruled 3–0 against Kalshi on 28 August 2026; Kalshi has asked for a rehearing.
Sixth Circuit (Ohio and Tennessee): ruled unanimously on 25 September 2026 that the states can regulate these contracts under their gambling laws; Kalshi has asked for a rehearing.
Fourth Circuit (Maryland): argued in May 2026, no ruling reported.
Supreme Court: petitions are pending, including one from New Jersey; none had been granted by mid-October, and Kalshi’s response in the New Jersey case is due on 9 November.

State actions multiplied through the year. Rhode Island’s attorney general sued Kalshi and Polymarket in state court on 21 May 2026, and Kalshi countersued in federal court the same day. On 24 September New York sued QCX LLC, which operates Polymarket US, alleging unlicensed gambling and that 18- to 20-year-olds could trade sports contracts although the state’s minimum age for mobile sports betting is 21; Polymarket moved the case to federal court. A Michigan state-court injunction against Kalshi took effect on 1 September, and court orders in Nevada and Massachusetts restrict sports contracts there. The CFTC has sided with the platforms in several cases.

Before you use a prediction market

Access to prediction markets changes state by state and sometimes week by week. Check the platform’s own availability page before funding an account, keep only what you need on the platform, and remember that state problem-gambling protections may not apply to a federally regulated exchange.

For context on where licensed sports betting is available, see our US market page. US-licensed books we review include DraftKings, FanDuel and BetMGM.

Comparing prices across sportsbooks

Line shopping is the simplest edge available to any bettor. The same match can be priced differently across books, especially in props and smaller leagues.

General patterns only. Compare live prices before betting.
Book type	Typical margin profile	Good for
Sharp or low-margin books and exchanges	Low on main markets	Reference prices and larger stakes
Large licensed recreational books	Low on major main lines, higher on props and parlays	Promotions, live betting, cash-out
Crypto sportsbooks	Varies widely by market	Fast crypto deposits and withdrawals, wide market lists

Licensed books such as bet365 are known for deep live-betting menus, while crypto books such as Stake and Sportsbet.io offer coin deposits and fast withdrawals. Check each against the margin-free price before you bet. Our crypto sports betting guide covers the payment side, and our deposits and withdrawals guide explains which methods are allowed where you live. Remember that crypto sportsbooks are not licensed in every country; check our UK and other market pages for what applies to you.

Bankroll rules that survive contact

Set aside a dedicated bankroll you can afford to lose.
Flat-stake 1–2% of it per bet rather than varying stakes with confidence.
Never increase stakes to chase a loss or “make back” a bad day.
Record every bet: date, market, odds, stake, closing price and result. The spreadsheet is more honest than memory.
Set deposit limits in your account before you start.

Worked example: why stake size matters

With a $1,000 bankroll and 1% flat stakes, each bet is $10. Ten losses in a row, which happens to any bettor eventually, cost $100 and leave $900. Staking 10% per bet would turn the same run into a $1,000 loss if you kept the stake fixed, or leave only about $349 if you staked 10% of whatever remained (0.9 to the power of 10 is 0.349). Small, fixed stakes let a genuine edge survive bad runs.

Betting should be paid entertainment with an analytical hobby attached, never income you rely on. Our responsible gambling guide covers limits, blocking tools and warning signs, and our responsible gambling hub lists where to get support.

Sources

18+ only. Gambling involves risk and should be treated as paid entertainment, never as income. Check that online gambling is legal where you live before you play. Get help and set limits →

How Betting Odds Work: Formats, Margins and Value