Bet 1 to 2 units per game, with 1 unit equal to 1% to 2% of your total bankroll, and cap any single wager at 5% even on your highest-confidence pick. The number that headline rule leaves out: unit size has to shrink as your bankroll shrinks, so a "2-unit bet" after a losing week is a smaller dollar amount than a 2-unit bet the week before — not the same figure carried forward out of habit.
- How many units should you bet per game: 1 to 2 units, with 1 unit equal to 1-2% of bankroll.
- Flat betting caps every wager at 1 unit regardless of confidence level.
- Confidence-based betting scales stake to 1-3 units based on graded pick strength.
- Kelly Criterion sizing ties unit size to edge size, often landing between 0.25 and 2 units.
- VigWatch's bet tracker logs unit size against results so drift from your plan gets caught early.
Why this matters
Unit sizing decides whether a bad month ends your bankroll or just dents it. A bettor risking 2% per game survives a 10-game losing streak with 80% of their bankroll intact; a bettor risking 10% per game is broke after 10 straight losses. VigWatch exists because most bettors lose money to bad sizing decisions, not bad picks — the math on unit size is more predictable than any single game outcome in 2026.
The question "how many units should you bet per game" only has one wrong answer: a number so large that three or four losses in a row wipe out the bankroll. Everything inside the 1-5% range is a defensible choice depending on your method.
How many units should you bet per game?
The right answer depends on which sizing method you use. Here's how the three common approaches compare on a standard game:
| Method | Unit size per game | Best for |
|---|---|---|
| Flat betting | 1 unit every game | Beginners, high-volume bettors |
| Confidence-based | 1 to 3 units | Bettors who grade picks before placing them |
| Kelly Criterion | 0.25 to 2 units (edge-dependent) | +EV bettors with tracked win-rate data |
All three methods define 1 unit the same way: 1% to 2% of your current total bankroll, recalculated periodically rather than fixed to a dollar amount forever. A $2,000 bankroll at 1% means a 1-unit bet is $20; if that bankroll grows to $2,500, the unit grows to $25.
Flat betting: 1 unit per game
Flat betting means every wager, regardless of how good the number looks, gets the same 1-unit stake. It removes the temptation to chase losses with bigger bets or get overconfident after a win streak.
Best for: bettors who place 20+ wagers a month and don't have a reliable way to grade edge size yet. The tradeoff is you leave some value on the table when a bet is genuinely a bigger mismatch than your other plays.
Confidence-based betting: 1 to 3 units per game
Confidence-based sizing lets you bet 1 unit on a coin-flip line and 3 units on a spot where your model or research shows a clear edge. The catch is discipline — self-graded confidence drifts toward "everything is a 3-unit play" without a tracking system to check it.
Best for: bettors who log results and can prove, over 100+ graded bets, that their 3-unit plays actually hit more often than their 1-unit plays. Without that proof, confidence-based sizing is just flat betting with extra steps and extra risk.
Kelly Criterion betting: 0.25 to 2 units per game
The Kelly Criterion calculates stake size from your estimated win probability and the odds offered, producing a mathematically "optimal" bet size for long-run bankroll growth. Most bettors use fractional Kelly — a quarter or half of the full Kelly number — because full Kelly stakes swing too violently game to game.
Best for: +EV bettors who can quote a specific edge percentage on a bet, not a gut feeling. Running the calculation by hand every game is tedious, which is why most professional bettors use a Kelly Criterion calculator instead of guessing at the fraction.
Why unit size varies
No single unit percentage fits every bettor or every bet. The number that's right for you depends on:
- Total bankroll size — a $500 bankroll and a $50,000 bankroll both use 1-2% units, but sportsbook minimum bet sizes can force the smaller bankroll toward a higher effective percentage.
- Bet type — parlays and same-game parlays carry more variance than single-game moneylines, which is a reason to size them below your standard unit even when the potential payout looks attractive.
- Edge size — a bet you can quantify as +EV by 3% deserves a different unit size than a bet you can only quantify as +EV by 0.5%.
- Sportsbook limits — some books cap winning bettors at stakes far below a full unit, which forces smaller effective sizing regardless of what your model says.
- Current drawdown — a bankroll down 20% from its peak should be betting smaller dollar-value units even if the percentage stays the same, because the base number shrank.
- Sample size confidence — a new model with 50 graded bets should size smaller than the same model at 500 graded bets, because the edge estimate is less reliable early on.
Tracking these factors by memory is how unit sizing drifts. A bankroll management tool recalculates your current unit value automatically as your bankroll moves, instead of leaving you to guess whether last month's unit size still applies.
Keep unit size tied to your real bankroll
Log every bet and let unit size update automatically as your bankroll changes.
What is a betting unit?
A betting unit is a fixed percentage of your bankroll — typically 1% to 2% — used as the base stake size for every wager, so "3 units" always means 3 times that percentage of your current bankroll, not a fixed dollar figure. Defining stakes in units instead of dollars keeps sizing consistent as your bankroll grows or shrinks over a season.
Should you increase unit size after a losing streak?
No — increasing unit size after a losing streak (chasing losses) is the single fastest way to turn a manageable drawdown into a blown bankroll. A bettor who doubles unit size after each loss needs only 4-5 straight losses to lose the entire bankroll, versus 30-50+ losses at a steady 2% unit size.
How many units is too many to bet on one game?
More than 5 units (5% of bankroll) on a single game is generally considered too aggressive for anyone other than a bettor with a verified, large edge and a tolerance for sharp bankroll swings. Sportsbooks also start limiting or reducing max bet sizes for players who consistently stake heavily on the same side, which makes oversized single-game bets a liability beyond just variance.
FAQ
How many units should you bet per game in 2026?
Bet 1 to 2 units per game in 2026, with 1 unit equal to 1% to 2% of your total bankroll. Scale up to 3 units only on bets where you can quantify a larger edge.
What percentage of bankroll is 1 unit?
1 unit is typically 1% to 2% of your total bankroll. Some conservative bettors use 0.5% per unit, while aggressive bettors go as high as 3%.
Is flat betting or confidence-based betting better?
Flat betting is better for bettors without proven grading accuracy, since it caps risk at 1 unit on every game. Confidence-based betting only outperforms flat betting when tracked results prove higher-confidence picks actually win more often.
How much should a beginner bet per game?
Beginners should bet 1 unit per game, sized at 1% of total bankroll, using flat betting until they have at least 50-100 tracked bets to evaluate.
Does unit size change with parlays?
Yes, parlays should generally use a smaller unit size than single-game bets because correlated legs and lower hit rates add extra variance on top of standard game variance.
What is the Kelly Criterion for bet sizing?
The Kelly Criterion is a formula that sizes bets based on your estimated edge and the odds offered, producing a stake between 0.25 and 2 units depending on edge size. Most bettors use fractional Kelly rather than the full calculated stake.
How many units is too risky per game?
More than 5 units, or 5% of bankroll, on one game is considered too risky for most bettors. That level of exposure only makes sense with a verified large edge and a plan to survive several consecutive losses.
Should unit size stay the same all season?
No, unit size should recalculate as your bankroll changes so the dollar value of 1 unit shrinks during a drawdown and grows during an upswing. Keeping the same dollar amount fixed all season ignores actual bankroll risk.
One last thing
Run the math on a real losing streak: at 2% flat units, five straight losses cost 10% of the bankroll, and the bettor is still in the game with 90% left to work with. Bump that to 5% units and five straight losses cost 25% — and a bettor who starts increasing stakes to "get back to even" after each loss can burn through 40% or more of the bankroll in the same stretch. The unit size decision matters more than any single pick, in 2026 or any other year.
