Expected Value 101: Stop Saying “Good Bet” Without the Math

A “good bet” is not “I think they win.” A good bet is “the price is wrong.” That is the whole game.

Sportsbooks are not paying you for being right. They’re paying you for taking risk at a price. If your price is bad, you can be “right” a lot and still bleed out slowly through juice. If your price is good, you can lose plenty in the short term and still be on the profitable side long term.

This article gives you the minimum math you need to stop guessing and start filtering bets properly. Not super scientific, just the stuff that actually changes your decisions.

What Expected Value Actually Means

Expected Value (EV) is the average profit or loss you should expect if you placed the same bet over and over at the same odds with the same true win probability.

It’s not a guarantee for tonight. It’s a quality check on the bet.

A positive EV bet is a bet you would want to place 1,000 times. A negative EV bet is a bet you would want to avoid even if it “feels obvious.”

Odds Are Just Probabilities in Disguise

Most bettors stare at -110, +150, -200 like they’re vibes. They’re not. They’re breakeven rates.

Convert American odds to implied probability

Negative odds (favorite):
P = |odds| / (|odds| + 100)

Positive odds (underdog):
P = 100 / (odds + 100)

That probability is your breakeven. If you cannot beat it, you cannot win long term.

Quick breakeven table (copy-paste friendly)

If you take -110, you need 52.38% just to not lose. That is the vig tax in plain English.

The Vig Problem (Why Most “Decent” Bettors Still Lose)

The standard spread or total price is usually -110. That means both sides are priced as if they win 52.38% of the time, which is impossible. The “extra” probability is the sportsbook margin.

Example: -110 / -110 implies:
52.38% + 52.38% = 104.76%

That 4.76% over 100% is the overround. It is the house edge baked into the market.

This is why “I win around 51%” is not “pretty good.” It’s losing. Slowly, but consistently.

If you want an objective way to test whether your process is strong, you should care about prices and closing lines, not your last 20 bets. That’s exactly why we publish our computer betting plays and let the numbers speak.

EV Worked Examples (Spread and Moneyline)

These examples are the difference between betting and guessing.

Example 1: Spread at -110

You bet a spread at -110. Your true cover probability (your model, your projection, whatever you trust) is 54%.

Assume 1 unit stake. At -110, the profit on a 1 unit risk is 0.909 units.

EV = (0.54 × 0.909) − (0.46 × 1.00)
EV = 0.4909 − 0.46
EV = +0.0309 units

That’s about +3.1% EV per bet. It’s not sexy, but it compounds.

Now watch what line shopping does.

If you get -105 instead, profit becomes 0.952 units on a 1 unit risk.

EV = (0.54 × 0.952) − (0.46 × 1.00)
EV = 0.5141 − 0.46
EV = +0.0541 units

Same pick. Better price. EV almost doubles.

That is why execution matters. If you want a steady stream of edges, you should be living on an odds screen, not a highlights reel. If you want a daily example of a model-driven approach (no vibes), check the free sports predictions from our AI .

Example 2: Underdog moneyline (the price flips the bet)

You believe an underdog wins 41% of the time.

Compare three prices:

• +135 breakeven = 42.55%
• +140 breakeven = 41.67%
• +150 breakeven = 40.00%

At +135, you are negative EV (your 41% is below 42.55%).
At +140, you are still negative EV (41% below 41.67%).
At +150, you are positive EV (41% above 40%).

Same team. Same handicap. Two are bad bets, one is a good bet. That’s the reality of “price is the bet.”

Example 3: Favorite moneyline (you can be right and still lose)

A favorite is -200. Breakeven is 66.67%.
Your true win probability estimate is 62%.

If you risk 100, profit is 50.

EV = (0.62 × 50) − (0.38 × 100)
EV = 31 − 38
EV = -7

That is a -7% EV bet. You might win tonight. You are still making a losing trade.

This is why public bettors love “safe” favorites and still lose. The price is doing the damage, not the pick.

De-vigging (How to Compare Against a Fair Market)

If you compare your probability to a juiced line, you will overestimate your edge. The fix is simple for two-way markets.

Basic de-vig method for spreads and totals

1. Convert both sides to implied probabilities
2. Add them together
3. Divide each implied probability by the total

Example: -110 / -110
Each implied = 52.38%
Total = 104.76%
Fair probability each side = 52.38 / 104.76 = 50%

That gives you a cleaner “market truth” baseline. It’s also why sharp books and closing lines matter so much when you’re trying to measure skill.

The EV Sanity Check (A fast routine before you bet)

Use this every time. It will save you money immediately.

1. Convert odds to breakeven %
2. Write your true probability estimate (even if it’s rough)
3. If true probability is not above breakeven, pass
4. If it is above, line shop anyway and improve the EV
5. If the line moved and you are late, re-check EV at the new price

This is also where market movement literacy matters. If you chase steam and take a worse number, you can turn a good bet into a bad one without realizing it.

What Counts as “Good EV” in real life

You do not need monster edges. In efficient markets, small edges are normal.

General guidelines:

• NFL spreads and totals: even +1% to +2% is meaningful if you can repeat it and you’re getting decent CLV
• NBA: edges often come from information timing (injury news), so EV can show up quickly then disappear quickly
• MLB moneylines: “cents” matter, so a small edge repeated often is the whole strategy

If you want to act like an investor, your focus is repeatability. That means you need process, not hero stories.

Common EV Traps (Where “math” gets faked)

1) Odds boosts don’t automatically create EV

A boost makes a bad price less bad. It does not magically make it good. Compare boosted odds to your fair probability, not to your feelings.

2) Parlays are usually a vig multiplier

Even if you “like” each leg, the compounded hold is brutal. Parlays only make sense when each leg is already +EV, and even then, correlation and pricing matter.

3) Steam chasing is a silent EV killer

If the value was at the opener and you’re betting the post-move line, you are often paying for confirmation. That’s not edge. That’s FOMO with a receipt.

4) Public bet percentages are not EV inputs

“70% on X” is usually marketing data, not audited market truth. Even when it’s real, it’s not a probability model. Treat it as noise unless you know exactly what book the data comes from and what it excludes (parlays and SGPs are the usual hidden monster).

Thought leadership note (how Balazs approaches “good bets”)

Balazs has spent 10+ years on both sides of the iGaming ecosystem: sportsbooks, affiliates, and the content layer that sells narratives to bettors. The big advantage of that perspective is understanding where bettors get nudged into bad pricing and bad habits, and then building systems that ignore the noise.

That’s the intent behind AI betting content on BPS: focus on repeatable edges, clean execution, and transparent tracking instead of hot takes.

If you want to sharpen the execution side of this, it also helps to understand how vig and overround affect totals, props, and live lines and why alt lines can quietly turn “safe” bets into expensive ones.

People Also Ask: Expected Value FAQs