How to find value bets and evaluate the odds

Value bets are wagers with a positive theoretical expected value: odds of 2.00 would be favorable with a true win probability of 55%, but not at 45%. With a €10 stake, the first case gives an expected net profit of €1 per bet; the second gives an expected net loss of €1. These are calculated averages, not predictions for individual bets.

The main challenge is that you don't know the true probability of winning. You have to estimate it. Even a correctly identified value bet can lose. We'll show you how to convert odds into probabilities, account for the sportsbook's margin, challenge your assumptions, and limit your stakes.

In brief: A value bet is a wager with odds higher than the fair odds based on your probability estimate. If you estimate 55%, the fair odds are 1.82 - any offered odds above that have theoretical value.

Value bets occur when the odds are too high

A value bet exists when the offered decimal odds exceed the fair odds implied by the true probability of an outcome. The catch is that this probability is usually unknown. If you estimate a team's chance of winning at 60%, your assessment of value depends on how reliable that estimate is.

The difference between high odds and a good price

Odds of 3.00 sound attractive, but they're too low if the estimated win probability is 25%. Fair odds would be 1 ÷ 0.25 = 4.00. For a €10 stake, the expected net loss is therefore €2.50: 0.25 × €30 payout minus the €10 stake.

Conversely, odds of 1.80 with a 60% win probability would have a positive expected value. Fair odds are around 1.67; the expected net return on a €10 stake is 0.60 × €18 minus €10 = €0.80. That's a theoretical average assuming an accurate estimate, not a prediction for an individual bet.

Value bets don't guarantee profits

Value betting doesn't mean simply picking the favorite: its odds can also be too low. Unlike an arbitrage bet, a single value bet doesn't cover all possible outcomes through multiple wagers.

Even with a true win probability of 60%, the bet loses 40% of the time. A losing bet therefore doesn't disprove value, just as a winning bet doesn't prove it. What matters is the quality of your probability estimate - not the result of a single game.

Calculate value using fair odds and expected value

A bet has theoretical value when its offered odds exceed the fair odds based on your estimated win probability. You need decimal odds and a probability for the calculation: enter 55% as 0.55, not 55.

Turning a probability into fair odds

Calculate the implied probability using 1 / decimal odds. Odds of 2.00 therefore correspond to 1 / 2.00 = 0.50, or 50%. That's the minimum win probability you need to theoretically break even at those odds, excluding additional costs.

Conversely, fair odds are calculated as 1 / estimated probability. If you estimate the win probability at 55%, the calculation is 1 / 0.55 ≈ 1.82. Offered odds of 2.00 exceed that and would be a value bet under this assumption.

Calculate the expected net return using probability × decimal odds − 1. In this example, that's 0.55 × 2.00 − 1 = 0.10, or 10% of the stake. The key question remains whether your 55% estimate is reliable: the formula doesn't test your assumption.

A worked example with a €10 stake

If you bet €10 at odds of 2.00, you receive €20 if you win, including your stake. Your net profit is then €10; if the bet loses, you lose your €10 stake.

With an estimated win probability of 55%, there is a 45% probability of losing. The expected value is therefore 0.55 × €10 − 0.45 × €10 = €1 per bet.

The expected value of +€1 is a calculated average under your assumptions, not a predicted payout for your next bet. In this example, that bet ends with either a €10 net profit or a €10 loss; consecutive losses are also possible.

Account for sportsbook margins and uncertain probabilities

High odds alone don't make a value bet: your estimated probability must exceed the break-even threshold. At odds of 2.00, that threshold is 50% - but neither sportsbook odds nor your own models automatically reveal the true probability.

The margin appears in the sum of implied probabilities

Consider a two-way market with no draw, where both sides are priced at 1.90. Calculate the implied probability using 1 ÷ odds:

The 5.26 percentage points above 100% are called the overround. They are neither a guaranteed sportsbook profit nor a flat loss on every individual bet; the actual result depends partly on stakes and the event's outcome.

To remove the margin proportionally, divide each implied probability by the total: 52.63% ÷ 105.26% = 50% per side. Within this simplified model, that produces fair odds of 1 ÷ 0.50 = 2.00. This adjustment provides a market estimate, not an objectively true probability.

Your own estimate needs reliable data

Estimating probabilities requires performance data suited to the sport: expected goals in soccer, for example, or serve and return statistics in tennis. Also account for opponent strength, lineups, injuries, and venue. Record your data and estimate with a timestamp before the event so information learned later doesn't distort your evaluation.

Odds of 2.00 illustrate the impact of uncertainty: if your estimated probability ranges from 48 to 55%, the expected net return per stake ranges from 0.48 × 2.00 − 1 = −4% to 0.55 × 2.00 − 1 = +10%. On a €10 stake, that means a theoretical expected value of −€0.40 to +€1, not the outcome of an individual bet. If this uncertainty determines whether the expected value is negative or positive, skipping the bet is a sensible option.

Check value bets in six steps

1. Define the market and settlement rules. Choose a clearly defined market, such as a two-way moneyline in basketball. Before comparing odds, check that every offer treats overtime, retirement, and abandoned events the same way. Record the rules: odds that include overtime aren't directly comparable with regulation-only odds.

2. Estimate the probability independently. Record your assessment using currently available data before looking for attractive odds. If you estimate a 55% chance of winning, document your reasons, such as confirmed lineups, unavailable players, and a transparent statistical model. If crucial information is missing, postpone the decision rather than finding arguments afterward to justify high odds.

3. Compare odds available at the same time. Compare only the same outcome under identical settlement rules, and record the time. At an estimated 55%, the difference between 1.80 and 1.95 turns a negative theoretical expected value into a positive one:

4. Calculate value and the stake you can actually place. At odds of 1.95, the implied probability is about 51.28%, calculated as 1 ÷ 1.95. Your estimated 55% corresponds to fair odds of around 1.82; on a €10 stake, that produces an expected net profit of €0.725 before fees, not a predicted profit on this individual bet. Check how much you can actually stake at the displayed odds, and deduct attributable costs. There is no universal minimum value threshold: uncertainty in your estimate can erase the calculated edge.

5. Check again immediately before placing the bet. Review current information, settlement rules, and the odds displayed on your bet slip; update your probability if new information warrants it. If the odds fall to 1.75 while your estimate remains 55%, the expected value becomes 0.55 × 1.75 − 1 = −3.75%. The earlier odds of 1.95 are no longer a basis for your decision.

6. Stick to your budget and record your decision. Bet only within your predefined entertainment budget; once you reach your stake limit, skip the bet regardless of its calculated value. Record the time, market, odds, estimated probability, stake, and reasoning, then add the result later without rewriting your original assessment. If you find no well-supported gap between your estimate and the offered odds, don't bet: this is a review process, not a requirement to place bets every day.

Avoid common mistakes and misleading signs of success

Three consecutive wins don't automatically make a team stronger. Weak opponents, red cards, or an unusually high chance-conversion rate may explain the results. Betting as a fan or giving recent results too much weight can quickly inflate your probability estimate - and make you see value bets where none exist.

Small samples and hindsight can mislead

Ten or twenty winning or losing bets aren't enough to evaluate a method reliably. Longer losing streaks are more likely at high odds than at low odds; conversely, a short winning streak can simply be luck. There is no fixed number of bets that proves success.

If you develop a model using data from a past season, test it on separate data from a later period. Adjusting rules after every result can teach the model old coincidences rather than relationships that carry over: this overfitting makes historical results look better than actual predictive performance. Don't retrospectively select only the leagues or markets that performed well, either.

Closing odds are a benchmark, not proof

If you took odds of 2.10 and the closing odds shortly before the event are 1.95, that can suggest you got a favorable price. This comparison is called closing line value and is meaningful only for the same market with identical settlement rules. Closing odds remain a market price, however: they prove neither your estimated probability nor future profits.

Tip: You're more likely to find value at sportsbooks with low margins. In our analyses, Duelbits offers soccer odds with a margin below 4%, and Roobet below 3% - the break-even threshold there is significantly lower than at sportsbooks with margins of 6-8%.

Parlays don't automatically create value by combining several picks, either. Two independent outcomes, each with a 60% probability, have only a 36% probability of both occurring. If the outcomes are dependent, you can't use this multiplication - the joint probability must be estimated separately.

Limit your stakes and plan for losing streaks

A fixed entertainment budget comes before any staking formula

Only use disposable entertainment money you can afford to lose in full for value bets. Money for rent, savings, or ongoing obligations is off-limits. Gambling isn't income - even when your calculation shows a positive expected value.

With a fixed budget of €200, flat stakes of €2 per bet equal exactly 1% of the initial budget. This is a calculation example, not a safe or universally suitable stake. If your remaining budget falls to €100, the same €2 already represents 2% of it.

The Kelly criterion calculates a stake fraction from the odds and estimated win probability. It isn't a necessary tool, and it stands or falls on your estimate: overestimating the chance of winning can make the formula suggest excessive stakes.

Higher stakes don't make up for losses

Even with independent bets that each have a 55% win probability, consecutive losses can occur. The probability that the next five bets all lose is 0.45⁵ = about 1.85%. Across many bets, the probability of experiencing such a streak at some point is higher.

Positive expected value therefore neither prevents losing streaks nor protects your budget. Even a theoretically favorable bet can lose, and the assumed win probability can be wrong. Don't double your stake after losses: €2 can quickly become €4, €8, €16, and €32.

Warning: Even genuine value bets often lose several times in a row. Never increase your stake after a losing streak - that's the fastest way to lose a budget despite positive expected value. Our bankroll management guide explains how to plan your stakes.

Set stake, loss, and time limits before placing your first bet, and don't raise them to recover losses. Take a break if you lose control, such as exceeding your loss limit or adding more money. For more guidance, see setting limits and recognizing warning signs.

Include costs and exchange-rate risks in crypto betting

Depositing Bitcoin doesn't change a team's probability of winning. The calculation for value bets stays the same: your estimated probability must exceed the probability implied by the odds. Crypto payments do, however, affect costs and the euro value of your balance.

First subtract the stake from the bet's gross expected value, then account for payment fees and conversion costs. Example: With a €20 stake and an expected net return of 5% from the bet before payment costs, your theoretical betting edge is €1. If the deposit and subsequent withdrawal cost €3 combined, the expected overall result for just this one bet is −€2. That erases the theoretical edge; the actual betting result can differ substantially.

However, you shouldn't allocate the full fixed transaction cost to every bet. If you use the same deposit for multiple bets, assess the costs of the entire payment process and allocate them reasonably across that usage. We explain the individual cost sources in crypto payment fees and networks.

Track betting results and coin prices separately, recording the stake, payout, and respective euro exchange rate, for example. A profit in BTC can still be worth fewer euros than your original stake if the BTC-to-euro exchange rate falls. Stablecoins also carry risks, including loss of their peg or problems with the issuer.

Only a well-supported probability estimate, suitable odds, and manageable total costs justify classifying a wager as a potential value bet. If you can't reliably assess the probability or costs, skip the bet.