Bankroll management and unit sizing
A simple sizing method so one bad week doesn't turn into a bad month. How to define your bankroll, size a unit, and stay in the game long enough for your edge (if you have one) to actually show up.
Your bankroll is a fixed, dedicated amount you can afford to lose in full - never rent, bills, or savings.
A unit is a fixed percentage of your bankroll, usually 1–2%, so bet size scales automatically as your bankroll changes.
The single biggest driver of long-term results isn't picking winners - it's not going broke before your edge plays out.
1. What Bankroll Management Actually Is
Bankroll management is the set of rules you use to decide how much money to risk on any individual bet, based on a fixed, dedicated pool of money set aside specifically for betting. It has almost nothing to do with picking winners and everything to do with surviving the losing streaks that happen to every bettor, including the best ones, so that your results over a full season actually reflect your skill (or lack of it) rather than the size of the wagers you happened to place during a rough patch.
This is a separate skill from handicapping. A bettor with genuine, sustainable edge who sizes bets badly can still go broke. A bettor with no edge at all who sizes bets conservatively will simply lose slowly, in line with the vig, rather than catastrophically. Bankroll management is the part of betting that determines whether you're still playing by the time your long-run edge, if you have one, has a chance to materialize.
2. Why Betting Fixed Dollar Amounts Fails
Most new bettors size bets in raw dollar terms - "I'll bet $50 on this game" - chosen based on how confident they feel in the moment rather than any consistent underlying rule. This approach has two compounding problems. First, confidence is a notoriously poor guide to actual win probability; bettors tend to feel most confident right after a big win, which is exactly when a bankroll can least afford a large bet gone wrong. Second, a fixed dollar amount doesn't adjust as the bankroll itself grows or shrinks, so the same $50 bet represents a wildly different level of risk depending on whether the bankroll is $2,000 or $200.
Sizing bets as a percentage of a current, tracked bankroll - the unit system covered throughout this guide - solves both problems at once. It removes the emotional, in-the-moment decision about bet size, and it automatically scales bets up as a bankroll grows and down as it shrinks, which is precisely the behavior that keeps a downswing from turning into a wipeout.
3. Defining Your Bankroll
A bankroll is a specific, dedicated amount of money set aside exclusively for betting, kept entirely separate from money needed for rent, bills, savings, or anything else load-bearing in your life. The single most important rule of bankroll management is deciding this number in advance, while thinking clearly, and never treating it as a number that can flex upward under pressure to chase a loss.
A reasonable starting bankroll is an amount you could lose in its entirety without it affecting your actual financial life in any meaningful way - not money you're counting on, not money borrowed for the purpose, and not a number you'll quietly top up from your checking account the moment it runs low. If losing the whole bankroll would be a real financial problem, the bankroll is too large relative to your actual situation, regardless of how good your picks are.
4. The Unit: Betting's Basic Unit Of Measurement
A unit is simply a fixed percentage of your current bankroll - most commonly somewhere between 1% and 2% - used as the standard building block for every bet size you place. Instead of saying "I bet $50," a bettor using units says "I bet one unit," and the actual dollar figure that represents is whatever 1–2% of the current bankroll happens to be.
The appeal of thinking in units rather than dollars is that it's portable and self-scaling. A one-unit bet automatically means something proportionally sensible whether your bankroll is $500 or $50,000, and it automatically shrinks in real dollar terms if your bankroll has taken a hit, which is exactly the built-in protection a fixed-dollar approach doesn't offer.
5. How To Choose Your Unit Size
Most experienced recreational bettors settle somewhere between 1% and 3% of bankroll per unit, with 1–2% being the more common, more conservative range. The right number for you depends on how much variance you're comfortable absorbing and how confident you genuinely are in your process, not on how quickly you'd like to grow the bankroll - a larger unit size doesn't create more skill, it just means each individual result swings the bankroll further in either direction.
- 1% per unit - conservative. A long losing streak barely dents the bankroll, at the cost of slower growth during winning stretches.
- 2% per unit - the most commonly recommended default for a recreational bettor with a reasonably tracked process.
- 3%+ per unit - more aggressive, generally reserved for bettors with a longer track record and a higher, well-documented tolerance for variance.
A newer bettor with limited betting history to draw on is generally better served starting conservative - 1% or so - and only considering a larger unit size after accumulating enough tracked results (see section 10) to have real evidence about their own process, rather than assuming skill that hasn't yet been demonstrated over a meaningful sample.
Unit sizing example
| Bankroll | 1% unit | 1.5% unit | 2% unit |
|---|---|---|---|
| $250 | $2.50 | $3.75 | $5.00 |
| $500 | $5.00 | $7.50 | $10.00 |
| $1,000 | $10.00 | $15.00 | $20.00 |
| $2,500 | $25.00 | $37.50 | $50.00 |
6. Flat Betting vs. Variable ("Confidence") Sizing
FLAT BETTING means risking the same number of units - most often exactly one - on every single bet, regardless of how strongly you feel about any individual pick. This is the simplest system to follow consistently and removes almost all of the emotional decision-making around bet sizing, since the sizing question is answered identically every time.
VARIABLE SIZING, sometimes called confidence-based sizing, allows the unit size to move within a defined range - commonly 1 to 5 units - based on a bettor's genuine conviction in a specific pick. This can better reflect real differences in edge between bets, but it introduces a serious risk: without real discipline, "confidence" quietly becomes a stand-in for whichever bets simply feel exciting in the moment, which is not the same thing as genuine analytical edge.
For most recreational bettors, flat betting at one unit per bet is the more sustainable long-term default. Variable sizing is worth considering only once a bettor has enough of a track record to know that their stated confidence levels actually correlate with real results, rather than with recency bias or excitement about a particular matchup.
7. The Kelly Criterion, In Plain English
The Kelly Criterion is a mathematical formula, originally developed for information theory, that calculates the theoretically optimal fraction of a bankroll to wager given a bettor's estimated edge and the offered odds. In its full form it can recommend surprisingly large bet sizes when a bettor's estimated edge is significant - but it depends entirely on that edge estimate being accurate, and most bettors, even skilled ones, tend to overestimate their own edge.
Because of this, most practical users of Kelly-based sizing use FRACTIONAL KELLY - commonly a quarter or half of what the full formula recommends - specifically to build in a margin of safety against the near-certainty that any individual bettor's edge estimate is at least somewhat too optimistic. Applied this way, Kelly-based thinking mostly serves as a reminder that bet size should scale with genuine, demonstrated edge rather than with excitement, and that even a real edge doesn't justify betting an enormous share of a bankroll on a single outcome.
For the large majority of recreational bettors without a rigorously tested edge estimate, flat unit betting (section 6) remains a simpler, more forgiving starting point than attempting to apply Kelly sizing directly.
8. Sizing Parlays, Teasers, And Props
Multi-leg bets like parlays and teasers, covered in depth in our parlays guide and teasers guide, carry meaningfully more variance per dollar risked than a single straight bet, since every leg needs to hit for the ticket to pay out at all. A sensible bankroll approach generally sizes these bets smaller in unit terms than a straight bet reflecting similar underlying confidence, precisely because the realistic probability of the full combination landing is lower than any single leg's individual probability.
The same logic applies to player and game props, covered in our props guide: props often carry wider inherent variance and, in some cases, a larger built-in house margin than a core spread or moneyline market, which is worth factoring into sizing decisions rather than betting props at the same unit size as your highest-confidence straight bets by default.
9. Setting A Stop-Loss And A Stop-Win
A STOP-LOSS is a predetermined limit - a specific number of units or a specific dollar figure - beyond which you commit in advance to stop betting for the day or week, regardless of how good the next opportunity looks in the moment. This exists specifically to counter a well-documented pattern where a losing session creates pressure to bet larger or more frequently in an attempt to "get back to even," which tends to compound losses rather than recover them.
A STOP-WIN works the same way in the other direction: a predetermined point at which you commit to banking your winnings rather than continuing to press a hot streak, guarding against the equally common pattern of giving back a big win by betting increasingly large amounts on the assumption that a streak of good results reflects skill rather than, at least in part, variance.
10. Tracking Results Over Time
Consistent record-keeping - the sport, the bet type, the odds, the unit size, and the result of every single bet - is what turns a string of anecdotes ("I remember hitting a big parlay in October") into an actual, honest picture of long-term performance. Without a real tracked record, it's extremely easy to remember the wins vividly and quietly forget or minimize the losses, a well-documented bias that makes self-assessed betting performance unreliable without hard numbers behind it.
A tracked record also makes it possible to evaluate performance by bet type and by sport, which often reveals that a bettor is genuinely stronger in one specific market - say, NFL unders - than in others, information that's essentially invisible without a spreadsheet or tracking app logging every wager over enough of a sample to mean something, generally at least several dozen bets before drawing any real conclusions.
11. Rebuilding After A Losing Streak
Every bettor, regardless of skill, will experience losing streaks - that's the nature of variance layered on top of a house-edge product, as covered in our vig guide. The unit system already provides some built-in protection here, since a shrinking bankroll automatically reduces the dollar size of future units. But the bigger risk during a losing streak isn't mathematical, it's behavioral: the temptation to abandon the sizing system entirely and bet larger to recover losses faster.
The more sustainable response to a losing streak is the opposite of increasing bet size: temporarily reducing unit size, taking a short break to review whether anything in the process has actually changed, and resuming normal sizing only once there's a clear, considered reason to believe the underlying process is still sound - not simply because the losing streak has gone on long enough that a win "feels due."
12. Common Bankroll Sizing Mistakes
Chasing losses. Increasing bet size specifically to recover a prior loss faster, rather than sizing each new bet independently based on the same consistent rules used before the loss occurred.
Redefining the bankroll under pressure. Quietly adding fresh money to a depleted bankroll outside of a planned, predetermined schedule, which defeats the entire purpose of setting a fixed bankroll in the first place.
Sizing every bet at maximum confidence. Treating nearly every pick as a 3-to-5-unit "lock," which - if genuine - would be statistically unusual, and more often reflects excitement rather than a real, differentiated edge assessment.
Ignoring parlay and prop variance. Applying the same unit size to a five-leg same-game parlay as to a core spread bet, without adjusting for the meaningfully different variance and realistic hit rate covered in section 8.
No tracked record. Relying on memory rather than logged results to assess whether a betting approach is actually working, as covered in section 10.
13. Worked Examples, Start To Finish
Example 1 - Setting up a fresh system. A bettor sets aside a $600 bankroll they're fully comfortable losing, chooses a 2% unit size ($12 per unit), and commits to flat one-unit bets on every pick for the first month while tracking every result in a spreadsheet.
Example 2 - Recalculating after a winning month. That same bankroll grows to $780 after a strong month. The bettor recalculates: one unit is now $15.60, rounded to $15, and future bets are sized against the new total rather than staying frozen at the original $12.
Example 3 - Recalculating after a losing month. Alternatively, the bankroll falls to $420 after a rough month. Sticking with the 2% rule, one unit shrinks to $8.40, automatically reducing risk during the downswing without requiring any separate decision.
Example 4 - Sizing a parlay appropriately. The same bettor likes a four-leg same-game parlay and, per their own house rule from section 8, caps parlay bets at half their normal straight-bet unit size - 0.5 units instead of 1 - reflecting the higher variance involved.
Example 5 - Hitting a stop-loss. After landing on the wrong side of three bets in a row on a Sunday slate, the bettor reaches their predetermined 3-unit daily stop-loss and closes the app for the day, resuming with a clear head the following week rather than trying to force a recovery in the same session.
14. Bankroll Management Across Multiple Sportsbooks
Bettors who maintain accounts at more than one sportsbook specifically to compare pricing, a practice covered in depth in our line shopping guide, need one additional layer of discipline: treating the combined total across every account as a single bankroll for sizing purposes, rather than accidentally betting a full unit at each of three different books on the same pick and effectively tripling the intended risk.
A simple way to manage this is to keep a single master tracking sheet showing the combined balance across every sportsbook account, and to calculate unit size against that combined total rather than against any one individual book's balance in isolation.
15. Frequently Asked Questions
What percentage of my bankroll should one unit be?
Most recreational bettors use somewhere between 1% and 2% per unit. Starting conservative, around 1%, is generally the safer choice for anyone without an established tracked record.
Should I increase my bet size after a big win?
Only in the sense that your recalculated bankroll - and therefore your unit size - has grown, per section 4. Increasing units specifically because you're feeling confident after a streak is a common mistake covered in section 12.
Is the Kelly Criterion worth using as a recreational bettor?
Full Kelly sizing depends on an accurate edge estimate that most bettors don't reliably have. A conservative fractional Kelly approach, or simple flat unit betting, is usually more practical, as covered in section 7.
How much money should I start with as a bankroll?
An amount you could fully lose without it affecting your actual financial life, decided in advance and kept separate from other funds, as covered in section 3.
Should I use the same unit size for parlays as for straight bets?
Many bettors size parlays and other high-variance bets smaller than their standard straight-bet unit, reflecting the lower realistic hit rate, as covered in section 8.
What should I do if I lose my entire bankroll?
Stop and reassess before adding new funds. Review your tracked record, per section 10, to understand what happened, and only resume with a clear, considered plan rather than an immediate reload.
16. Quick-Reference Glossary
- BANKROLL
- A fixed, dedicated amount of money set aside exclusively for betting, separate from other funds.
- FLAT BETTING
- Risking the same number of units on every bet, regardless of individual confidence level.
- FRACTIONAL KELLY
- Betting a reduced fraction, often a quarter or half, of what the full Kelly Criterion formula recommends.
- KELLY CRITERION
- A formula for calculating a theoretically optimal bet size based on estimated edge and offered odds.
- STOP-LOSS
- A predetermined limit at which a bettor commits in advance to stop betting for a set period.
- UNIT
- A fixed percentage, commonly 1–2%, of a current bankroll, used as the standard measure of bet size.
- VARIABLE (CONFIDENCE) SIZING
- Adjusting unit size within a defined range based on a bettor's stated conviction in a specific pick.