Expected Value in Sports Betting:
The Complete Guide
Expected Value (EV) is the single most important concept in profitable sports betting. This guide explains what it means, how to calculate it, why positive-EV betting loses in the short run and wins long-term, and how to find edges the sportsbooks have mispriced.
What is Expected Value in sports betting?
Definition
Expected Value (EV) is the average amount you can expect to win or lose per bet if the same bet were placed an infinite number of times. A positive EV (+EV) means the bet returns more than it costs on average. A negative EV (−EV) means it costs more than it returns. EV is the only mathematically sound basis for evaluating whether a bet is worth placing.
Most recreational bettors evaluate bets by asking a single question: Do I think this team will win? That is the wrong question. The right question is: Are the odds being offered higher than the true probability of this outcome justifies?
A team can have a 40% chance of winning and still be the correct bet — if the sportsbook is offering odds that imply only a 30% chance. That gap between the true probability and the implied probability is the edge. Expected value is how you measure that gap in dollar terms.
The core insight
You cannot determine whether a bet is good by looking at whether it wins or loses. A bad bet can win. A good bet can lose. The only way to evaluate a bet correctly is to assess its expected value before the outcome is known.
The Expected Value formula
The EV formula is straightforward. For a binary outcome (win or lose):
Expected Value formula
EV = (Pwin × Amount won) − (Ploss × Amount lost)Where Pwin = your estimated probability of winning | Ploss = probability of losing (1 − Pwin) | Amount won = net profit if the bet wins | Amount lost = stake (typically 100 for calculation purposes)
It is also common to express EV as a percentage of stake, which makes it easier to compare bets of different sizes. An EV of +$5 on a $100 bet is a 5% edge. An EV of +$3 on a $100 bet is a 3% edge.
EV as percentage of stake
EV% = EV / Stake × 100Professional bettors typically seek edges of 2–5% or more. Anything below 1% is marginal and highly sensitive to probability estimation error.
Worked examples
Example 1: The fair coin flip (zero EV)
Scenario: Someone offers you even money (+100 American odds / 2.00 decimal) on a fair coin flip. You win $100 if heads, lose $100 if tails. The true probability of heads is exactly 50%.
EV = $50 − $50 = $0
Example 2: Positive EV — the mispriced coin
Scenario: Same fair coin, but the bookmaker offers +110 odds (decimal 2.10) on heads. You win $110 if heads, lose $100 if tails. The true probability is still 50%.
EV = $55 − $50 = +$5
Example 3: Negative EV — standard sportsbook juice
Scenario: NFL game, point spread priced at −110 on both sides. You estimate the home team covers with 52% probability and bet accordingly at −110 odds (decimal 1.909).
EV = $52 − $52.80 = −$0.80
The break-even problem
At −110 odds (the US standard for most spread and totals bets), you need to win 52.38% of bets just to break even. At −120 odds, the break-even rises to 54.55%. Most recreational bettors overestimate their win rate by 5–10 percentage points, which means they are almost certainly betting with negative EV on most wagers.
Example 4: Real sports bet — NFL moneyline
Scenario: Detroit Lions at +175 (American) moneyline to beat the Dallas Cowboys. Your statistical model estimates Detroit has a 38% chance of winning.
+175 American = (175/100) + 1 = 2.75 decimalEV = $66.50 − $62.00 = +$4.50
100 / (175 + 100) = 36.36%Your estimate: 38% > implied 36.36% → Edge exists.
The vig problem: why most bets are −EV by default
Every sportsbook builds a margin into its odds — called the vig, juice, or overround. This margin ensures the book profits regardless of outcome. Understanding vig is essential to understanding why most sports bets are negative EV even before you assess the game itself.
How vig works
Consider a two-outcome market where both sides are priced at −110. If you add the implied probabilities of both sides together, they should sum to 100% in a fair market. At −110:
Implied probability at −110
110 / (110 + 100) = 52.38%Both sides at −110: 52.38% + 52.38% = 104.76% — the extra 4.76% is the bookmaker’s margin.
This means the sportsbook is operating with a ~4.8% edge on every two-sided −110 market. Over thousands of bets, the house extracts this margin reliably. Your job as a bettor is to find situations where your estimated probability of an outcome is higher than the implied probability embedded in the odds — enough to overcome the vig.
Vig comparison across bet types
| Bet type | Typical odds | Break-even win % | Approximate vig | Verdict |
|---|---|---|---|---|
| NFL / NBA spread | −110 / −110 | 52.38% | ~4.8% | Moderate |
| Moneyline favourite | −150 / +130 | Varies | ~3–5% | Moderate |
| Totals (O/U) | −110 / −110 | 52.38% | ~4.8% | Moderate |
| Same-game parlay | Varies | Varies | 10–20%+ | −EV trap |
| Futures | Varies | Varies | 15–30%+ | Very high vig |
| Live betting | Varies | Varies | Often 5–8% | High vig |
| Boosted promos | Enhanced | Below normal | Near zero or negative | Best +EV opportunity |
Same-game parlays: These are among the most profitable products for sportsbooks because the correlations between legs are extremely hard to price correctly, and the vig compounds across legs. Most same-game parlays carry effective margins of 15–25%. Avoid unless you have a specific, justified reason to believe legs are less correlated than the book assumes.
Implied probability: converting odds to probability
To calculate EV, you need to compare your probability estimate against the probability implied by the odds. Converting between odds formats and implied probability is a core skill.
American odds to implied probability
Conversion formulas
Favourite (−odds): Implied% = |odds| / (|odds| + 100) × 100
Underdog (+odds): Implied% = 100 / (odds + 100) × 100| American odds | Implied probability | You need to win more than |
|---|---|---|
| +200 | 33.3% | 33.3% to break even |
| +150 | 40.0% | 40.0% |
| +110 | 47.6% | 47.6% |
| +100 (Evens) | 50.0% | 50.0% |
| −110 | 52.4% | 52.4% (standard spread) |
| −130 | 56.5% | 56.5% |
| −150 | 60.0% | 60.0% |
| −200 | 66.7% | 66.7% |
| −300 | 75.0% | 75.0% |
Free Odds Converter
Convert between American, decimal, and fractional odds with implied probability calculated automatically.
How to find positive EV bets
Finding +EV bets requires a reliable probability estimate that is more accurate than the sportsbook’s implied probability. There are four primary approaches, each with different levels of accessibility.
1. Line shopping across sharp sportsbooks
Sharp books (Pinnacle, Circa, Bookmaker) have efficient markets shaped by professional money. When a retail sportsbook’s line differs significantly from a sharp book’s line, the retail book is usually the one that is wrong. Betting the better number at the retail book represents +EV relative to the true market.
2. Devigging the market
Remove the bookmaker’s margin from the odds to get the “true” implied probability of each outcome. Compare this against other books. Where one book’s no-vig price is significantly higher than another’s, value exists.
Simple devig example (two-outcome market)
Side A at −115 → implied = 53.49%
Side B at −105 → implied = 51.22%
Total implied = 104.71% | Vig = 4.71%
No-vig prob Side A = 53.49 / 104.71 = 51.08%
No-vig prob Side B = 51.22 / 104.71 = 48.92%If another book offers Side B at +110 (implied 47.6%), compare against the no-vig 48.92%. The no-vig estimate is higher than the implied, meaning Side B at +110 is slightly +EV.
3. Statistical and AI models
Build or use a model that generates win probabilities independently of the sportsbook’s line. Compare the model’s probability against the implied probability. Where they diverge significantly (and the model is well-calibrated), an edge may exist.
AI betting tools claim to do this automatically. The quality of their probability estimates varies enormously. Evaluating whether a model’s probability estimates are calibrated — i.e., when it says 60%, the outcome occurs about 60% of the time — is the most important criterion when assessing any AI betting system.
4. Promoted odds and boosts
Enhanced odds promotions at retail sportsbooks are sometimes the clearest +EV opportunities available. A bet boosted from −110 to +100 on a 50/50 market is a direct +EV gift. These are time-limited and have stake limits, but they represent genuine mathematical edges when the underlying market is efficient.
The probability estimation problem
Every method of finding +EV bets depends on your probability estimates being more accurate than the sportsbook’s. If your estimate is wrong, the “edge” is illusory. This is why tracking your EV and calibration over hundreds of bets is essential — it tells you whether your estimation process is genuinely accurate or systematically biased.
EV vs. variance: why correct bets lose
This is the most psychologically difficult aspect of expected value betting, and where most recreational bettors abandon a sound strategy prematurely.
A +EV bet does not win more often than it loses in the short run. A bet with a 52% win rate still loses 48% of the time — which means it loses in nearly half of all trials. Over a 50-bet sample, even a 5% edge is frequently invisible due to variance.
How many bets to evaluate EV reliably?
| Your true edge | Bets needed to be 95% confident | Bets needed to be 99% confident |
|---|---|---|
| 10% edge | ~400 bets | ~700 bets |
| 5% edge | ~1,500 bets | ~2,600 bets |
| 3% edge | ~4,200 bets | ~7,300 bets |
| 1% edge | ~38,000 bets | ~66,000 bets |
Most bettors make 200–500 bets per season. At a 5% edge, that sample is too small to distinguish skill from variance at the 95% confidence level. This is why professional bettors track Closing Line Value (CLV) rather than relying solely on results — CLV is a leading indicator of edge that becomes statistically meaningful much faster than win rate.
Read next: Closing Line Value explained
CLV is the most reliable proxy for long-term edge. Learn why beating the closing line is more important than your win rate.
Win rate vs. EV: why a 60% win rate can still lose money
Win rate is not a reliable indicator of profitability. The relationship between win rate and profit depends entirely on the odds at which those wins occur. Consider two bettors over 100 bets at $100 stake each:
| Bettor | Win rate | Average odds | Net result | Verdict |
|---|---|---|---|---|
| Bettor A | 60% | −150 ($67 per win) | 60 × $67 − 40 × $100 = $4,020 − $4,000 = +$20 | Marginal profit |
| Bettor B | 45% | +160 ($160 per win) | 45 × $160 − 55 × $100 = $7,200 − $5,500 = +$1,700 | Strong profit |
| Bettor C | 55% | −130 ($77 per win) | 55 × $77 − 45 × $100 = $4,235 − $4,500 = −$265 | Loss despite 55% win rate |
Bettor C has a 55% win rate and is losing money. Bettor B wins fewer than half their bets and is making significant profit. The difference is expected value at the odds taken — not win rate.
EV and the Kelly Criterion: sizing bets by edge
Once you have identified a +EV bet, the next question is: how much to bet? This is where the Kelly Criterion connects directly to EV.
Kelly sizing is based on the same inputs as EV: your probability estimate (p) and the net odds (b). The formula determines what fraction of your bankroll maximises long-term growth:
Kelly Criterion
f* = (bp − q) / bf* = fraction of bankroll to bet | b = net decimal odds (odds − 1) | p = win probability | q = 1 − p
A negative Kelly result means negative EV — the two are mathematically equivalent ways of saying the same thing: do not place this bet.
Most systematic bettors use half-Kelly or quarter-Kelly to reduce variance while preserving most of the growth benefit. Full Kelly is theoretically optimal but assumes perfectly calibrated probability estimates — a condition rarely met in practice.
Free Kelly Criterion Calculator
Enter your win probability and odds to calculate the optimal stake — with full, half, and quarter Kelly outputs.
Tracking your EV: the betting log
The only way to know whether you are genuinely finding +EV bets — rather than getting lucky — is to maintain a detailed betting log and measure your results against expectations over time.
What to record for each bet
| Field | What to record | Why it matters |
|---|---|---|
| Date and time | Exact timestamp of bet placement | Tracks line movement; compares to closing line |
| Market | Sport, league, game, bet type | Identifies which markets you edge in |
| Odds taken | American or decimal odds at time of bet | Core EV input |
| Closing odds | Odds at kick-off from a sharp book | CLV calculation |
| Your probability estimate | Your pre-bet win probability | Tracks calibration over time |
| Stake | Dollar amount wagered | P&L tracking |
| Calculated EV | EV at time of bet | Compares expected vs. actual results |
| Result | Win / loss / push | P&L tracking |
After 500+ bets, compare your actual ROI to your predicted ROI (sum of EV% across all bets). If actual exceeds predicted consistently, your estimates are conservative. If actual lags predicted significantly, your probability estimates may be overconfident.
Common EV mistakes
- Confusing short-term results with EV. A 10-bet losing run on +EV bets is entirely normal. Abandoning a sound process after a losing streak destroys long-term profitability.
- Overestimating your win probability. Most bettors estimate higher win probabilities than are justified. A 5% overestimate in probability can turn a +3% EV bet into a −2% EV bet.
- Ignoring the vig on accumulated bets. Each bet carries vig. Even a slight positive edge per bet compounds positively; a slight negative edge compounds negatively. Small errors in probability estimation matter over large samples.
- Betting without a probability estimate. Placing bets based on team preference, narrative, or streaks without a prior probability estimate means you have no EV framework at all.
- Treating win rate as the success metric. As shown above, win rate without odds context is meaningless. Track expected ROI, not win percentage.
- Betting high-vig markets without extra edge. Same-game parlays and futures require substantially larger edges to overcome the built-in margin. Most bettors applying normal-market EV logic to these products are losing money without realising it.
Need an AI tool to generate probability estimates?
EV betting requires win probability estimates before you can apply the formula. ZCode System uses historical data and AI models across NFL, NBA, MLB, NHL, and soccer to generate probability scores — the inputs EV and Kelly sizing require. It does not guarantee profit, and no AI system does. But it provides a systematic starting point for probability estimation.
Explore ZCode SystemAffiliate disclosure: we earn a commission if you sign up via this link at no extra cost to you. We recommend ZCode because it provides probability-based outputs, not because of commission rate.
FAQ
Frequently asked questions
What is a good expected value in sports betting?
Any positive expected value is mathematically better than negative. In practical terms, professional bettors typically target edges of 2–5% or more. An EV under 1% is marginal — small errors in your probability estimate can flip it negative. An EV of +5% on a $100 bet means you expect to average $5 profit per bet over a large sample.
Can you lose money on positive EV bets?
Yes. Positive EV means you will profit over a large enough sample — not that you win every bet. A +EV bettor can lose money for weeks or even months due to variance. This is why sample size matters: at a 5% edge, you need approximately 1,500 bets to be 95% confident your results reflect genuine edge rather than luck.
How do I calculate expected value for a sports bet?
Use the formula: EV = (probability of winning × amount won per bet) − (probability of losing × amount lost per bet). For example, if you estimate a 55% win probability on a bet at +110 odds (win $110 per $100 staked): EV = (0.55 × $110) − (0.45 × $100) = $60.50 − $45 = +$15.50 per $100 bet. This is a +15.5% edge.
What is the difference between EV and CLV?
Expected Value is calculated at the time of bet placement using your probability estimate. Closing Line Value (CLV) measures the difference between the odds you took and the closing odds at kick-off — a proxy for how efficiently the market priced the outcome. CLV is useful because it provides market validation of your edge without requiring your own probability model. Consistently beating the closing line is one of the strongest indicators of genuine long-term +EV betting.
How accurate do my probability estimates need to be?
Very accurate at the margin. A 2–3% error in probability estimation can eliminate a small edge entirely. At −110 odds, you need 52.38% win rate to break even. If your model estimates 54% but the true rate is 51%, you are betting with negative EV despite thinking you have an edge. This is why calibration tracking — comparing your estimates to actual outcomes over large samples — is essential.
Does expected value work for all sports?
The EV framework applies to any sport with quantifiable odds. However, finding genuine edges varies by market. High-profile NFL, NBA, and soccer markets are highly efficient — sharp money corrects mispricing quickly. Lower-profile markets (lower football leagues, minor tennis tournaments) tend to be less efficient, offering more opportunities for well-researched bettors to find +EV situations. The trade-off is smaller markets with lower betting limits.
Responsible gambling notice. Expected Value is a mathematical framework, not a guarantee of profit. Sports betting involves variance — even correctly identified +EV bets lose. Never bet more than you can afford to lose. If gambling is causing harm: NCPG | BeGambleAware | Gambling Therapy