Expected Value Explained: The Only Math That Matters
I burned through $2,400 before I understood expected value, and I thought I was being smart. I tracked every bet in a spreadsheet, celebrated my 58% win rate on NFL spreads, and genuinely believed I had figured something out. Then I did the EV calculation for the first time and realized I was losing $0.14 on every single dollar I risked. My winning percentage meant nothing because expected value explained the only truth that mattered: I was playing a negative expectation game and no amount of research was going to fix it.
Expected value is the average amount you win or lose per bet if you could repeat that exact wager thousands of times. Not what you hope to win. Not what feels right. The mathematical certainty of what happens when probability meets payout over the long run. Every bet you place has an EV, and if that number is negative, you are bleeding money even when you are winning.
The Math That Destroyed My Confidence
Here is the formula that made me question everything: EV = (Probability of Win × Amount Won) – (Probability of Loss × Amount Lost). Simple enough that I should have learned it before I ever placed my first bet, brutal enough that it exposes every bad decision you have ever made.
I was betting NFL spreads at -110 odds, which means I had to risk $110 to win $100. I was hitting 58% of those bets and feeling like a genius. The actual expected value told a different story:
| Win Rate | Win Amount | Loss Amount | EV Per $110 Bet | EV Percentage |
|---|---|---|---|---|
| 58% | $100 | $110 | -$4.20 | -3.8% |
| 55% | $100 | $110 | -$10.50 | -9.5% |
| 52.38% | $100 | $110 | $0.00 | 0% |
| 60% | $100 | $110 | $2.00 | +1.8% |
The breakeven point at standard -110 odds is 52.38%. I needed to win 52.38% of my bets just to break even after the juice. My 58% win rate sounds impressive until you calculate that I was making $2 per $110 risked at that rate, not the profit windfall I imagined. Over 200 bets that season, I averaged 56% and lost $380 because variance kept me below my average during critical stretches.
The vig is everything. That -110 price means the sportsbook has built in a 4.5% edge on a fair coin flip. You are not betting against the outcome, you are betting against math that has already tilted the table. Check an EV Calculator with your actual win rate and watch it confirm what you have been denying.
Why Win Rate Alone Is a Trap
I met a guy in a forum who was winning 62% of his MLB moneyline bets and still losing money. He was betting heavy favorites at -180 to -220, winning more often than he lost, and bleeding out because his average loss was $200 while his average win was $90. His EV per bet was -$6.40 despite a win rate that would impress most bettors.
Win rate without context is a vanity metric. You need to know your average odds, your average stake, and whether the combination produces positive expectation. I have seen 50% win rates make money and 60% win rates go broke. The difference is always expected value.
The Spreadsheet That Changed Everything
I started calculating EV before placing bets instead of after, and it eliminated 40% of the wagers I used to make. I built a simple tracker with these columns: Date, Bet Type, Odds, Stake, My Estimated Win Probability, Implied Probability from Odds, EV, Result, Profit/Loss.
The implied probability column is critical. If the sportsbook offers +150 odds, they are saying that outcome has a 40% implied probability (calculated as 100 / (150 + 100) = 0.40). If you believe the true probability is 50%, you have found a +EV bet. The math: (0.50 × $150) – (0.50 × $100) = $25 expected value on a $100 bet.
| Your Estimated Probability | Offered Odds | Implied Probability | Edge | EV on $100 Bet |
|---|---|---|---|---|
| 50% | +150 | 40% | +10% | +$25.00 |
| 55% | -110 | 52.4% | +2.6% | +$2.36 |
| 45% | +120 | 45.5% | -0.5% | -$1.00 |
| 60% | +200 | 33.3% | +26.7% | +$80.00 |
The problem with this approach is that your estimated probability is just a guess dressed up in a percentage. I thought I was being objective, but I was adjusting my estimates to justify bets I wanted to make. Over a six-week tracking period, I claimed to have found 34 +EV bets. I went 18-16 on those bets and lost $290 because my probability estimates were inflated by 8-12% on average.
Estimating true probability is the hardest part of this game. You can calculate expected value perfectly and still lose if your inputs are garbage. The Betting Data Lab approach uses historical data to calibrate estimates, which helps, but even professionals admit their edge is usually 2-4% on their best bets, not the 10-20% edges I convinced myself I was finding.
Where EV Breaks Down in Real Life
Expected value assumes infinite bankroll and infinite time. In reality, you have $2,000 in your account and a six-month attention span. Variance will destroy you before the long run arrives if you do not manage bet sizing correctly.
I found a bet with +5% EV and put 20% of my bankroll on it because the math said it was profitable. I lost. I found another +4% EV bet the next week and bet 18% because I needed to recover. Lost again. After five weeks of aggressive betting on positive EV spots, I was down 48% despite making mathematically correct decisions. My expected value was positive, but my risk management was suicidal.
The Kelly Criterion Reality Check
The Kelly Criterion tells you how much to bet based on your edge and your bankroll. The formula is: (Edge / Odds) = Bet Percentage. If you have a 5% edge on a bet at +150 odds (1.5 decimal), Kelly says to bet 3.3% of your bankroll. Most professionals use quarter-Kelly or half-Kelly because full Kelly produces wild swings.
| Edge | Odds (Decimal) | Full Kelly | Half Kelly | Quarter Kelly |
|---|---|---|---|---|
| 5% | 2.50 | 3.3% | 1.7% | 0.8% |
| 3% | 2.00 | 3.0% | 1.5% | 0.8% |
| 8% | 3.00 | 5.3% | 2.7% | 1.3% |
| 2% | 1.91 | 2.2% | 1.1% | 0.6% |
I ran a simulation with 1,000 bets at +3% EV each, betting full Kelly versus flat 2% stakes. Full Kelly ended with 340% profit but experienced a maximum drawdown of 52%. Flat betting ended with 180% profit and a maximum drawdown of 28%. The aggressive approach made more money but required surviving multiple stretches where my bankroll was cut in half. An Kelly Criterion Calculator will show you these tradeoffs, but it cannot prepare you for what it feels like to watch half your money disappear while making +EV bets.
The Bets I Stopped Making After Learning EV
Parlays were the first to go. A two-leg parlay at -110 each leg pays +264 but requires you to win both. If each leg has 52.4% true probability (breakeven at -110), your combined probability is 27.5%. The expected value: (0.275 × $264) – (0.725 × $100) = -$0.04 per $100 bet. Even with positive EV on each individual leg, the compounding vig destroys you.
Live betting was next. The odds move too fast for me to calculate EV accurately, and I was making emotional decisions based on what I was watching rather than what the math supported. Over a two-month period, my pre-game bets returned -2.1% EV while my live bets returned -7.8% EV. I was significantly worse at estimating probability when the clock was running.
Same game parlays are poison. The sportsbook adjusts the correlations in their favor, and you cannot calculate the true combined probability without knowing their correlation model. I tracked 50 same game parlays with what I estimated was +8% EV each. I hit 6 of them and lost $780. My probability estimates were wrong because I did not account for the correlation penalty the book had built in.
The Only Bets Worth Making
You need +3% EV minimum to overcome estimation error and variance. Anything less than that, you are gambling on your ability to estimate probability accurately, which you probably cannot do. I now track my closing line value instead of trying to estimate true probability. If I bet +6.5 and the line moves to +4.5 by kickoff, I have captured value regardless of the outcome.
Focus on markets where you have information edges, not opinion edges. I stopped betting NFL because everyone has access to the same injury reports and stats. I started betting lower-tier soccer where line movement is slower and inefficiencies last longer. My hit rate dropped from 56% to 53%, but my average EV per bet increased from -1% to +2.2% because I was finding actual pricing errors instead of convincing myself I was smarter than the consensus.
What the Long Run Actually Looks Like
I am 1,400 bets into tracking EV religiously. My overall record is 718-682, which is 51.3%. My total profit is $1,840 on $58,000 total wagered, which is a 3.2% ROI. My average EV per bet is +1.8%, meaning I am underperforming my expectation by 1.4%, which is probably due to overestimating my edge on 20-30% of the bets I tracked as +EV.
The graph of my bankroll over that period looks like a mountain range. I had an eight-week stretch where I went 42-38 on +EV bets and lost $680. I had a five-week stretch where I went 28-14 and made $1,240. Expected value tells you what will happen on average, but variance determines what actually happens in any given sample.
The reality is that even with perfect EV calculation, you need 500+ bets minimum to have confidence that your results reflect your true edge. Most bettors do not have the bankroll or discipline to survive that long. I was down 18% at bet 200, up 8% at bet 400, down 5% at bet 600, and finally stabilized around my expected ROI after bet 800. You cannot evaluate a betting strategy with 50 bets, no matter how sure you are about the math.
Frequently Asked Questions
Can you be profitable long-term with negative EV bets?
No. If your expected value is negative, you will lose money as your sample size increases. You might get lucky for 100 bets or even 500 bets, but the math always wins eventually. Every casino game has negative EV, which is why casinos make money and players lose over time.
How do I know if my estimated win probability is accurate?
Track your calibration by grouping bets into probability buckets. If you estimate 60% probability on 100 bets, you should win approximately 60 of them. If you are consistently winning fewer than your estimates predict, you are overestimating your edge and making negative EV bets while thinking they are positive.
Is it possible to calculate EV without knowing the true probability?
You can use closing line value as a proxy. If you consistently beat the closing line, you are likely capturing positive EV even if you cannot calculate it precisely. An ROI Calculator helps track whether your betting approach is actually profitable regardless of your EV estimates.
Explore more strategies in our I Lost $840 Testing Baccarat Tie Bets So You Don’t Have To: Why The 14% House Edge Destroys Bankrolls.