Last Updated: November 8, 2026
Quick Summary
To find positive EV bets, convert the sportsbook odds into an implied probability, build your own probability estimate, and bet only when your number is higher than the market's vig-free figure. Positive EV bets pay more than the true chance of the outcome justifies, so the price sits on your side over a large sample.
- Strip the vig by dividing each implied probability by the total, then compare your estimate to that fair number.
- A -150 favorite implies 60%, so you need to win 60% of the time just to break even.
- Test calibration, not accuracy: check whether your 60% predictions actually hit about 60% of the time.
- Line shopping matters because a point or two of price difference can turn a losing bet into a positive one.
- Use half or quarter Kelly rather than full Kelly, which is aggressive and can draw down hard during cold streaks.
Table of Contents
- What Positive EV Bets Actually Mean
- Step 1: Convert Sportsbook Odds Into Implied Probability
- Step 2: Build a Probability Estimate You Can Trust
- Step 3: Compare Your Number With the Market Price
- Step 4: Work the Math With a Positive EV Betting Calculator
- Step 5: How to Find Positive EV Bets Free and Online
- Step 6: How to Find Positive EV Bets NFL Week to Week
- Step 7: Size the Bet and Track Whether the Edge Is Real
- Frequently Asked Questions
What Positive EV Bets Actually Mean
Expected value is the average result of a bet if you could place it many times over (Expected value). A positive expected value bet has an average profit above zero.
Most bettors skip this step. They back a team because it looks good, then wonder why the account drifts down. The fix is a repeatable workflow.
- Convert the odds into a probability
- Build your own probability estimate
- Compare the two numbers
Below, we run that loop step by step, with the math worked out.
Step 1: Convert Sportsbook Odds Into Implied Probability
Every price a sportsbook shows hides a probability. Your first job is to pull it out.
American Odds to Implied Probability
For a favorite at -150, the formula is 150 divided by 150 plus 100, giving 0.60, or 60%.
If you would rather not do it by hand, use an odds converter to switch between American, decimal and fractional formats and see the implied win probability of any price.
The number you get is the break-even rate. At -150, you need to win 60% of the time just to break even.
Removing the Vig to Get Fair Odds
The implied probabilities on both sides of a market add up to more than 100%. That extra is the vig, also called the juice or overround.
Say a two-sided market prices both sides at -110. Each side implies 52.4%, so the pair adds to 104.8%. The book's hold is that 4.8% slice.
To find the fair odds, strip the vig out by dividing each implied probability by the total. A hold and no-vig calculator does this in one pass and shows the fair odds for both sides.
Now you have a baseline. The market's fair price is your starting point, not your answer.
Step 2: Build a Probability Estimate You Can Trust
This is the hardest part, and the one most guides skip. Many guides define "true probability" and move on, without showing you how to produce one or how to account for the fact that it is probably wrong.
A probability estimate is your own number for how often an outcome happens. It has to come from somewhere repeatable, not from a feeling.
Where the number comes from
Three sources work, and they are not equal:
- A model. Rate teams or players on the inputs that matter, then simulate or convert those ratings into a win chance. For an NFL spread that means offensive and defensive efficiency, pace, injuries, rest and venue; for an MLB moneyline, starting pitcher quality, bullpen state and park factors.
- A base rate. Look at how often similar situations have resolved in the past, then adjust for what is different this time. "Teams coming off a short week" is a starting point, not an answer.
- A sharp market. Use the closing line from a low-hold book as a proxy, then look for spots where another book is slower to move. This is the fastest route, but circular: your edge only exists if you can identify which book is lagging and why.
Calibrate before you trust it
The common mistake is building a model and trusting it too early. Test it on past data first, and test calibration rather than accuracy: when your model says 60%, does that outcome actually happen about 60% of the time?
A practical check: bucket your historical predictions into groups (50-55%, 55-60%, and so on), then compare the predicted rate in each bucket to the actual hit rate.
Account for uncertainty
A projection is not a fact. Every estimate carries error, and the size of that error should change how you bet. Two habits help:
- Use a range, not a point. If your model says 58%, ask whether the honest range is 54-62%. If the market's fair number sits inside that range, you have noise, not an edge.
- Shrink toward the market. When your estimate and the sharp market disagree by a lot, the market is usually right more often than a solo model. Pulling your number partway toward the market's fair probability avoids overbetting a bad projection.
A rough, well-calibrated estimate beats a precise, uncalibrated one. Track its predictions against results over a full season before you scale up.
Watch out: a model never tested on out-of-sample data is a guess with a spreadsheet attached. Hold back recent seasons as a test set so you are not grading the model on the data you built it from.
Step 3: Compare Your Number With the Market Price
Now the two numbers meet. Your estimate says 58%. The market's implied probability, after you strip the vig, says 54%.
That gap is your edge. It is small, and that is normal. Real edges are thin.
The rule: bet only when your probability is higher than the fair implied probability.
Line shopping matters here.
Watch out: comparing your number to the raw implied probability, vig included, hides real edges.
Step 4: Work the Math With a Positive EV Betting Calculator
Expected value is simple once you have the inputs. The formula is:
EV = (your probability × profit if you win) − (chance you lose × stake)
What most guides skip is the conversion step before the formula. Books quote prices in American, decimal or fractional format, and you need the same format on both sides of the equation. Here is the full loop, worked three ways.
Convert the price first
- American to decimal: for a positive price, divide by 100 and add 1 (+120 becomes 2.20). For a negative price, divide 100 by the absolute number and add 1 (-150 becomes 1.667).
- Fractional to decimal: divide the top number by the bottom and add 1 (5/2 becomes 3.50).
- Decimal to implied probability: divide 1 by the decimal (2.20 becomes 45.5%).
An odds converter does all three in one pass and shows the implied win probability of any price, which is the number you actually compare against your estimate.
Worked example: a $100 bet at +120
Your model says the true chance is 50%. If you win, you profit $120. If you lose, you drop $100.
- Win side: 0.50 × 120 = 60
- Lose side: 0.50 × 100 = 50
- EV = 60 − 50 = +10
That bet is worth $10 on average per $100 staked. In decimal (2.20), the EV is (0.50 × 1.20) − (0.50 × 1.00) = +0.10 per unit staked, the identical answer, expressed per unit.
Worked example: a $100 bet at -150
Your model says the true chance is 65%. At -150 the decimal is 1.667, so a $100 stake returns $66.70 in profit.
- Win side: 0.65 × 66.70 = 43.36
- Lose side: 0.35 × 100 = 35.00
- EV = 43.36 − 35.00 = +8.36
Positive, but notice how much thinner the margin is than the +120 example. The bigger the gap between your estimate and the implied probability, the larger the EV, and the more a small error in your estimate can erase it.
Worked example: a $100 bet at +250
Your model says the true chance is 32%. Decimal is 3.50, so profit on a win is $250.
- Win side: 0.32 × 250 = 80
- Lose side: 0.68 × 100 = 68
- EV = 80 − 68 = +12
A positive EV betting calculator runs any of these in seconds and also shows the break-even rate at the listed price. Break-even at +250 is 28.6%; your 32% estimate sits 3.4 points above it, and that gap produces the +$12.
The sensitivity check most bettors skip
Before you stake, ask how wrong your estimate can be before the EV turns negative. In the +120 example, break-even is 45.5%, so a 50% estimate has 4.5 points of cushion. If your model is routinely off by more than that, the bet is not as safe as the headline EV suggests.
Run the math before every bet. If the EV is negative, the price is against you, however good the team looks.
Step 5: How to Find Positive EV Bets Free and Online
You can run this process without paying for anything. The free route takes more time, but it works.
How to Find Positive EV Bets Online Without Paid Tools
Start with free odds screens that show prices across several books side by side.
Then apply the steps above to each outlier. Most will be noise. A few will be real.
- Use free calculators for implied probability, hold and EV
- Track every bet in a simple spreadsheet
- Log the price, your estimate, the stake and the result
For a deeper walkthrough of the full process, read our guide on how to find positive expected value in betting.
The trade-off is speed. Manual screens take time and lines move while you work, so accept that you will miss some.
Step 6: How to Find Positive EV Bets NFL Week to Week
Football brings the most volume and the most noise. Here is how to find positive EV bets NFL weeks reward.
Start with the lines that move most: injuries, weather and quarterback news. When a starting quarterback is ruled out, the market reprices fast.
Next, track the closing line.
A few habits that help:
- Check injury reports before you build your number
- Compare the opening line with the current line
- Note which books move first and which lag
The same logic carries to college football, college basketball, the NBA, MLB and golf.
Step 7: Size the Bet and Track Whether the Edge Is Real
Finding the edge is half the job.
The Kelly criterion sets the stake from your edge and your bankroll (Kelly criterion). A Kelly criterion calculator turns your win probability, the odds and your bankroll into a suggested stake, with a fractional setting for lower variance.
Many bettors use half or quarter Kelly.
Then track. Keep a log with four columns:
| Field | What to Record | Why It Matters |
|---|---|---|
| Price | The odds you took | Shows whether you beat the close |
| Your estimate | Your probability figure | Tests your model's accuracy |
| Stake | Amount risked | Confirms sizing discipline |
| Result | Win or loss | Builds the sample |
Sample size is everything. Twenty bets tell you nothing.
Watch out: a short winning run can feel like proof but is not.
At Edge, we send every pick with its line, edge and a suggested stake by email and Telegram, and every pick is logged and graded in public. You can see every graded pick on the Records page rather than take our word for it. Our models cover the NFL, college football, college basketball, the NBA, MLB and golf.
The real challenge is not understanding expected value.
Edge is built for that gap.
Get started with Edge and put a repeatable process behind every bet.
Frequently Asked Questions
How do you identify positive EV bets?
You need two numbers: the implied probability in the sportsbook price and your own estimate of the true probability. Convert American odds to implied probability, strip out the vig to get fair odds, then compare your estimate with the market price. If your number says an outcome wins more often than the price implies, the bet is positive EV. A positive EV betting calculator does the arithmetic in seconds once you have both inputs.
What are some free tools for finding positive EV bets?
Free odds screens and odds comparison pages show prices across books so you can spot the outlier. Free calculators handle the math: an expected value tool at https://www.sportbettingedge.com/tools#expected-value, a no-vig tool at https://www.sportbettingedge.com/tools#hold-novig-odds and an odds converter at https://www.sportbettingedge.com/tools#odds-converter. The part no free tool does for you is the probability estimate, which is where the work sits. Edge also offers a 10-day free trial if you want model-driven numbers.
Does a positive EV bet guarantee a win?
No. Positive EV means the price is better than the true probability over many repetitions, not that the single bet cashes. A +EV bet at a 55% true probability still loses 45% of the time. Variance is why bankroll management and stake sizing matter as much as finding the edge. Judge a betting process on a large sample size, not on one weekend, and treat any single result as noise.
Can you find positive EV bets by comparing sportsbook odds?
Yes, and it is the most repeatable manual method. Line shopping across several books surfaces the best available price on the same outcome. When one book lags the market consensus, that stale price is often where the edge sits. Track line movement too: a number that has already moved tells you the market has repriced. Comparing prices is the easy half; estimating true probability is the half that decides whether the edge is real.
