Building a Betting System
If you're serious about betting, you need a system. Not some magical formula, but a structured way to find bets with positive expected value. This means understanding implied probability, calculating your own probabilities, and tracking whether you're actually beating the market.
Implied Probability: What the Odds Really Mean
Every line has an implied probability—what the sportsbook thinks the chances are of that outcome happening.
Convert American odds to implied probability:
For negative odds:
Example: -150 odds
For positive odds:
Example: +200 odds
If you add up the implied probabilities for both sides of a bet, they'll total more than 100%. That extra percentage is the vig.
Example: Lakers -110 vs. Suns +110
- Lakers: 52.4%
- Suns: 52.4%
- Total: 104.8%
The extra 4.8% is the sportsbook's edge.
Expected Value: The Only Number That Matters
Expected value (EV) tells you the average profit or loss on a bet if you could make it thousands of times.
Example: You think the Suns have a 40% chance to win at +200 odds.
- Profit if you win: $200 (from a $100 bet)
- Loss if you lose: $100
This bet has +$20 EV. On average, you expect to profit $20 per $100 bet.
If your probability estimate matches or is lower than implied probability, your EV is zero or negative. Don't bet.
Finding Your Edge
Your edge is the difference between your probability estimate and the market's implied probability.
If the market says 35% and you think it's 40%, you have a 5% edge.
The bigger the edge, the more you should bet. But you need to be honest about whether your probability estimate is better than the market's. Most of the time, it's not.
Building a Basic Model
You don't need machine learning. A simple model beats no model.
Step 1: Pick Predictive Variables
For NBA, good starting variables:
- Offensive/defensive efficiency (points per 100 possessions)
- Recent form (last 10 games)
- Home/away splits
- Rest days
- Injuries to key players
Step 2: Estimate Impact
How much is each variable worth? Start with rough estimates:
- Home court: ~2.5 points
- Back-to-back game: ~3-4 point penalty
- Missing a star player: ~5-8 points depending on the player
Step 3: Calculate Predicted Spread
Add up all your adjustments to get a predicted point differential.
Example:
- Celtics have +8 point differential
- Warriors have +5 point differential
- Game in Boston (Celtics +2.5 home court)
Predicted: Celtics by 5.5 points
If the market line is Celtics -3, there's potential value on the Celtics.
Step 4: Test and Refine
After 100+ games, check:
- Are you predicting winners correctly more than 52.4% of the time?
- Are your probability estimates calibrated? (If you say 60%, does it happen ~60% of the time?)
- Are you beating the closing line?
If not, your model needs work. Either your variables are wrong or your weights are off.
Line Shopping is Non-Negotiable
The difference between getting -3 and -3.5 is huge over time.
If you bet $100 per game at -110 odds and you win 53% of the time (a good win rate):
- At -3: you might win 53 of 100 bets
- At -3.5: you might win 51 of 100 bets (losing 2% because of the extra half-point)
That's the difference between profitability and losing money.
Use an odds aggregator or check multiple books manually before every bet.
Closing Line Value (CLV)
The closing line is the final line right before the game starts. It's the sharpest line—the market's best guess at the true probability after absorbing all available information.
If you consistently beat the closing line (bet early, line moves in your direction), you're likely finding +EV bets.
Example:
- You bet Celtics -3 on Monday
- Closing line on game day: Celtics -5.5
- You gained 2.5 points of CLV
Track this. If you're consistently getting worse than the closing line, you're betting into a smarter market. That's a losing strategy.
Bet Sizing: Kelly Criterion
Once you've found a +EV bet, how much should you bet?
The Kelly Criterion gives you the optimal bet size to maximize long-term bankroll growth:
Where:
- = fraction of bankroll to bet
- = your win probability
- = decimal odds - 1
Example: You think the Celtics at +150 (2.50 decimal) have a 45% chance to win.
Full Kelly says bet 8.3% of your bankroll.
Problem: Full Kelly is aggressive and leads to big swings. Most people use fractional Kelly (25-50% of the full amount) to reduce variance.
Half Kelly in this example: 4.15% of your bankroll.
Key Metrics to Track
ROI (Return on Investment):
Good bettors hit 3-8% ROI long-term. Anything above 5% is excellent.
Win Rate: The percentage of bets you win. At -110 odds, you need 52.4% to break even.
Closing Line Value: The average difference between your bet and the closing line. If this is positive, you're beating the market.
Sharpe Ratio: Risk-adjusted return. Higher is better. If you're winning 5% ROI but with huge swings, your Sharpe is low.
When to Bet
Early: Sharp bettors often bet early before the market fully adjusts. If you have good information or a good model, betting early can get you value.
Late: Casual bettors bet late. If you're fading the public (betting against heavy public action), wait until close to game time when public money has inflated the line.
Track which works for you. Some bettors are great at early lines. Others are better at live betting or late line movement.
Building Discipline
The hardest part isn't finding +EV bets—it's not betting when you don't have an edge.
If you're bored and want action, bet $5 for fun. Don't bet $100 on a game you haven't analyzed.
If you have a bad week, don't increase bet size to "make it back." Variance happens. Trust the process.