What positive EV actually means
Expected value (EV) is the long-run average profit per unit staked. A wager is "+EV" when the price you receive implies a lower probability than the event's true probability. Take a coin flip priced at $2.10. Implied probability is 1 / 2.10 = 47.6%, but the true probability is 50%. Stake $100 a thousand times and your average return is $2.40 per bet — positive expected value.
In real markets you never know the true probability with certainty. You estimate it. The whole game is making that estimate sharper than the bookmaker margin you're paying.
The +EV formula
For a decimal-odds wager:
EV per $1 stake = (p_true × decimal_odds) − 1 EV% = EV per $1 × 100
Where do you get p_true?
You need a sharp reference market — one that prices efficiently, has high limits, and accepts winning customers without restriction. The cleanest references in 2026 are international sharp markets and Betfair Exchange (post-commission). Both have margins under 2.5% on major markets, vs 5–8% at corporate AU books.
Krok Odds builds its sharp baseline from the lowest-margin international markets with Betfair Exchange as cross-validation. The combined "sharp line" is what +EV is measured against — read the full breakdown on the methodology page.
Devigging — removing the bookmaker margin
A 2-way market like Lakers vs Celtics priced 1.91 / 1.91 implies 52.4% + 52.4% = 104.8%. The extra 4.8% is the bookmaker's margin (the "vig" or "overround"). To get the true implied probabilities you devig — most commonly with the proportional method:
p_implied(Lakers) = 1 / 1.91 = 0.524 p_implied(Celtics) = 1 / 1.91 = 0.524 Sum = 1.048 p_true(Lakers) = 0.524 / 1.048 = 0.500 (50.0%) p_true(Celtics) = 0.524 / 1.048 = 0.500 (50.0%)
A worked +EV example
Sharp market (international sharp baseline, after devig) says the Sydney Roosters win at 58%. An AU corporate has them at $1.85.
Implied prob at AU book = 1 / 1.85 = 54.05% Sharp true prob = 58.0% EV per $1 = (0.58 × 1.85) − 1 = +0.073 EV% = +7.3% (long-run edge per dollar staked)
Reading a price as a probability
Every decimal price is a probability in disguise: implied probability = 1 / decimal odds. $1.50 is 66.7%, $3.00 is 33.3%, $8.00 is 12.5%. Do that for every runner in a market and the probabilities add to more than 100% — the excess is the bookmaker margin you are paying before a ball is bowled.
A bet is +EV when your estimate of the true probability is higher than the price implies; it is −EV when it is lower. The market does not need to be wrong for a bet to be −EV. Most bets are −EV by construction, because the price already includes the margin.
Implied probability = 1 / 1.85 = 54.1% If true probability is 58% -> EV = (0.58 x 1.85) - 1 = +7.3% (+EV, bet it) If true probability is 50% -> EV = (0.50 x 1.85) - 1 = -7.5% (-EV, pass) Same price, opposite decision: only your probability estimate separates them.
A repeatable +EV workflow
The bettors who last do the same seven things in the same order every time. Once a screen is open it takes about two minutes per market, and the order matters more than the speed.
The seven steps
- 1. Convert the price. Decimal to implied probability (1 / odds), so every book is speaking the same language.
- 2. Find a fair probability. A sharp baseline — international sharp markets, or Betfair Exchange after commission — not another corporate book.
- 3. Remove the margin. Devig the sharp market (proportional for two-way, Shin or worst-case for three-way) so the probabilities sum to 100%.
- 4. Shop the same bet. Compare the AU corporate price against the devigged fair price, not against the other corporates.
- 5. Calculate the edge. EV% = (p_true x decimal_odds - 1) x 100, and set a floor before you look: 2–3% at a corporate.
- 6. Size the stake. Quarter to half Kelly, or a flat 1–2% of bankroll. Never full Kelly on an estimate.
- 7. Log the closing price. Closing-line value is the only proof of edge you get inside a few hundred bets.
Which markets and sports actually pay
Edge does not distribute evenly. In Australia the softest prices are usually where the book is slowest to move: racing flucs in the last 20 minutes before a jump, niche sports with thin markets, player props, and anything a book prices off its own model rather than an exchange.
The hardest are the markets the books copy straight off a sharp feed — AFL and NRL head-to-head at the major corporates, the biggest NBA and EPL games. A price there is rarely wrong by more than the margin.
Where the edge usually sits (AU, 2026)
- Racing: fixed-price flucs that lag the exchange, especially in the last 20 minutes and on metro meetings.
- Player props and same-game legs: books price these off their own models with wide margins and slow updates.
- Niche sports and lower-tier competitions: fewer eyes, slower reaction to team news and late scratchings.
- Promos and boosts: real EV only once the terms are converted to a cash-equivalent value first.
The trade-off is limits. The softest markets are also the ones a corporate book will cut fastest, which is why volume spread across many books beats size at one.
What a realistic return looks like
A well-executed +EV strategy at corporate bookmakers returns roughly 2–5% on turnover over a full season. That is turnover, not bankroll: $50,000 turned over at 3% is $1,500, and it does not arrive in a straight line.
Variance is the part that ends most attempts. At a 4% edge and $100 stakes you can be 20 units down after 100 bets and still be running the strategy correctly. It typically takes 300–500 settled bets before results separate from luck, which is why CLV matters more early than profit does.
1,000 bets x $100 at 4% EV = +$4,000 expected Standard deviation over 1,000 bets at ~$1.91 prices = ~$1,580 A -$1,000 year sits inside one SD of a genuinely profitable strategy.
Doing the steps with tools
You can run this workflow by hand on one market. You cannot run it across eight bookmakers and forty markets before the price moves, which is what the scanner is for: it devigs a sharp baseline, compares it to every AU corporate price on the same selection, and ranks by edge.
The four tools in the loop
- Odds screen — every AU corporate price on the same selection, side by side (/odds).
- No-vig calculator — turn a sharp market into fair probabilities (/free-tools/ev-calculator).
- EV calculator — convert a price and a fair probability into EV% before you stake.
- Bet tracker — log the price taken and the closing price so CLV is measurable.
The live +EV scanner at /ev-bets shows the sharp baseline source and the EV% on every signal, so you can sanity-check the math instead of trusting it.
How big should each bet be?
Kelly Criterion is the bankroll-optimal staking rule. Full Kelly is volatile, so most professional bettors use quarter-Kelly or half-Kelly. A simple flat-stake fraction (1–2% of bankroll per bet) also works.
Practical staking rules
- Never bet more than 5% of your bankroll on a single +EV signal — even a 20%+ EV bet can lose.
- Use Kelly for sizing direction, then halve it. Variance in real bets is brutal.
- Re-baseline bankroll weekly, not after every win or loss.
- Skip bets under 2% EV — corporate books restrict winning accounts, so smaller edges aren't worth the heat.
Common mistakes that destroy edge
Most retail +EV strategies fail not because the math is wrong, but because of execution errors:
- Comparing AU corporate prices against each other instead of against a sharp baseline (international sharp markets / Exchange).
- Ignoring boost/promo terms — bonus bets often pay only profit, not stake, so EV math has to be adjusted.
- Chasing stale lines after a sharp move. If the sharp baseline has moved, the AU "lagging" price may already be priced in by the time you click.
- Not tracking closing-line value. CLV is the only proxy for long-run edge that doesn't require thousands of bets to be statistically meaningful.
- Betting too big. Variance is heavy in the short run. A 20-bet losing streak at 5% EV is normal.
FAQ
Is positive EV betting legal in Australia?
Yes. Placing bets at licensed Australian bookmakers is legal regardless of strategy. AU corporate books are private businesses and can restrict or close winning accounts, which is the practical limit on scaling a +EV strategy.
How much EV is enough to bet?
Most professional bettors require at least 2–3% EV after accounting for staking limits and account-restriction risk. Below 2% the long-run edge often does not compensate for variance and book heat.
Why do my +EV bets lose?
Variance. EV is a long-run average. Over 20–50 bets, a 5%-EV strategy can lose money. Hundreds of bets are typically required to be confident the model is profitable, which is why closing-line value tracking matters.
What's the difference between +EV and arbitrage?
Arbitrage locks in a small guaranteed profit by covering all outcomes at different bookmakers. +EV is a one-sided positive-expectation bet that wins on average but can lose individually. Arb returns are smaller and more reliable; +EV returns are larger but variance-heavy.
Do bookmakers restrict +EV bettors?
Yes. AU corporate bookmakers routinely cap or close accounts that consistently take +EV prices. Betfair Exchange and offshore sharp markets (AU access restricted) are the main exceptions. Spreading volume across many books and avoiding consistent post-move bets reduces heat.
What does +EV and −EV mean in plain English?
+EV means the price pays more than the true probability justifies, so the bet profits on average over many repeats. −EV means the price pays less than fair, so it loses on average. The same price can be either — what separates them is your estimate of the true probability, not the odds on the screen.
Where do I see live +EV bets right now?
Krok Odds publishes a continuously refreshed +EV feed across AU corporate bookmakers and Betfair Exchange. Open the EV Bets scanner — every signal includes the sharp baseline source and EV% so you can sanity-check the math.