Sports betting bankroll management is a set of rules for how much money you bet with in total and how much of it goes on each bet. Bet size decides how long you last through a cold streak, and cold streaks are guaranteed. This guide shows the arithmetic: flat unit sizes, the Kelly formula, and a risk-of-ruin simulation we ran ourselves, with the method laid out so you can check it.
By Jake Mercer, Lead Sports Betting Analyst. Last updated: October 15, 2026.
The Quick Version
- Your bankroll is money set aside only for betting. Rent money, grocery money and credit cards are not a bankroll. A unit is one standard bet, usually 1% to 3% of the pot.
- At -110 you must win 52.38% of your bets to break even. You risk $110 to win $100.
- Kelly math: the Kelly fraction is your win chance minus your loss chance divided by the payout. A true 55% winner at -110 gets 5.5%. A true 52.38% bettor gets 0%.
- Risk of ruin is real even at break-even. In our simulation, a bettor who wins exactly 52.38% at -110 and bets 5% units was left with less than one unit in 36% of 500-bet runs. At 1% units, none of the 100,000 runs ended that way.
- Overestimating your edge is the expensive mistake. If your true win rate is 53% but you bet as if it were 55%, the typical bankroll after 1,000 bets is about 0.48x the start.
- Our take: use flat units of 1% to 2%, and if you try Kelly, use a quarter of it and a deliberately low win-rate guess.
What Is Sports Betting Bankroll Management?
Bankroll management answers two questions: how much money do you bet with in total, and how much goes on any single bet? Your bankroll is the pot you set aside for betting and nothing else. Your unit is the standard amount you put on one bet, written as a percentage of that pot. If your bankroll is $500 and your unit is 2%, one unit is $10.
Nothing here helps you pick winners. What it buys you is enough bets to learn whether your picks are any good. If you are new to the basics, start with our Sports Betting 101 guide and come back for the money side.
Why Does Bet Size Matter More Than Your Picks?
Because losing streaks are normal, and bet size decides whether a streak is a bad week or the end of your bankroll. At -110, the most common price on a point spread, you risk $110 to win $100. The fee built into that price is called the vig. To break even you need 110 / 210 = 52.38% wins, so even a bettor who clears that bar loses about 47 of every 100 bets.
How common is an eight-bet losing streak? We computed it directly. If you win exactly 50% of your bets, the chance of at least one run of eight straight losses is 9.1% over 54 bets (about 3 bets a week for an 18-week NFL season) and 62.5% over 500 bets. If you win 52.38%, the numbers are 6.5% and 49.9%. An eight-loss streak is not rare luck. Over enough bets it is close to a coin flip.
Here is what one eight-loss streak costs at different unit sizes on a $500 bankroll (8 x unit):
| Unit size | Dollars per bet | Cost of 8 straight losses | Share of $500 bankroll |
|---|---|---|---|
| 1% | $5 | $40 | 8% |
| 2% | $10 | $80 | 16% |
| 3% | $15 | $120 | 24% |
| 5% | $25 | $200 | 40% |
Same picks in every row. The only change is the number on the ticket.
How Do You Set a Sports Betting Budget You Can Stick To?
Pick a dollar figure you could lose entirely without changing anything about your month. That figure is your bankroll. Then work through it in order:
- Choose a window. A single NFL season or one calendar month works. “Whenever” does not.
- Name the amount. Say $400 for the season. Write it down.
- Run the pain test. If losing all $400 would make a bill late, cut it to $200 and stop negotiating with yourself.
- Fund it once. Move the money in one or two deposits, then leave your card out of it.
- Set your unit. One percent to three percent of the pot. On $400, that is $4 to $12 a bet.
The funding step matters more at offshore books. The Michigan Gaming Control Board says players on unlicensed sites do not get the self-exclusion and deposit-limit tools that licensed Michigan operators must offer. So the limit has to come from you. A crypto transfer is not something you can ask your bank to reverse, so one practical fix is to keep the season bankroll in your own account and move over only the current month’s slice. Read the real risks of betting at offshore sportsbooks before you fund anything.
What Is Flat Betting and Why Do Beginners Use It?
Flat betting means every bet is the same size, win or lose. If your unit is $10, you bet $10 on the game you love and $10 on the game you sort of like. No doubling after a loss. No tripling because a game is on national TV. Covers notes that flat betting usually uses 1% to 5% of the bankroll per play. Results become easy to read because every bet carries the same weight.
Here is how the common sizing methods compare:
| Method | How a bet is sized | Main risk | Good for beginners? |
|---|---|---|---|
| Flat units | Same dollar amount every bet | Does not shrink after a losing run | Yes |
| Percent of bankroll | Same percent of the current bankroll, recalculated | Bets shrink after losses and grow after wins | Yes, if recalculated monthly |
| Kelly | A fraction set by your win chance and the payout | Needs an honest win-rate estimate. Overestimating it is costly (see below) | Only fractional Kelly |
| Martingale | Double after every loss | Six straight losses starting at $10 cost $630, and the seventh bet would be $640 | No |
Martingale is not a strategy. It hides losses until one run takes everything. Add the sums: $10 + $20 + $40 + $80 + $160 + $320 = $630 lost, and the next bet is $640.
How Does the Kelly Criterion Work?
The Kelly criterion is a formula for the bet size that grows a bankroll fastest over the long run. John L. Kelly Jr. of Bell Labs described it in 1956, according to the Kelly criterion entry on Wikipedia. For a bet that either loses your whole stake or wins b dollars per dollar risked, the Kelly fraction is f = p – q / b, where p is your win probability and q = 1 – p.
At -110, you win $100 on a $110 bet, so b = 100 / 110 = 0.9091. Say your true win rate is 55%. Then f = 0.55 – 0.45 / 0.9091 = 0.55 – 0.495 = 0.055, or 5.5% of your bankroll. On a $1,000 bankroll that is $55. Half Kelly is $27.50 and quarter Kelly is $13.75. Here is how the fraction changes with your true win rate:
| True win rate at -110 | Expected profit per $1 risked | Kelly bet size |
|---|---|---|
| 50% | -4.5% | Bet nothing |
| 52.38% | +0.0% | Bet nothing |
| 53% | +1.2% | 1.3% |
| 55% | +5.0% | 5.5% |
| 57% | +8.8% | 9.7% |
Two things stand out. First, Kelly says to bet nothing at or below the 52.38% break-even line. Second, the sizes are bigger than most beginners expect. A 55% winner is told to bet 5.5%. That is also an ambitious win rate: it equals a 5% return on money risked, and Covers puts 5% to 7% at the top of what great long-term bettors reach.
Kelly also resizes after every bet. With a $1,000 bankroll and full Kelly at 5.5%, your first bet is $55.00. After five straight losses the bankroll is $753.63 ($1,000 x 0.945^5), and the next bet is $41.45. The bets shrink as you lose. That is Kelly’s built-in brake. In theory it means pure Kelly never reaches exactly zero, though large drawdowns still happen.
We simulated 40,000 bettors who each placed 1,000 bets at -110 with a true 55% win rate. Full Kelly cut the bankroll to half its start at some point in 42% of runs. Half Kelly did it in 10%, and quarter Kelly in 0.4%. The typical (median) finish was 3.97x the start for full Kelly, 2.81x for half and 1.83x for quarter. Wikipedia’s entry says gamblers use less than full Kelly to reduce volatility and account for model error. Our numbers show the trade: half Kelly finishes at 2.81x instead of 3.97x, but the chance of a 50% drawdown falls from 42% to 10%.
| Sizing (true 55% at -110, 1,000 bets) | Bet size | Chance of a 50% drawdown | Typical bankroll after 1,000 bets |
|---|---|---|---|
| Full Kelly | 5.50% | 42% | 3.97x |
| Half Kelly | 2.75% | 10% | 2.81x |
| Quarter Kelly | 1.38% | 0.4% | 1.83x |
What Is Risk of Ruin, and How Big Is Yours?
Risk of ruin is the chance your bankroll runs out before you stop betting. Here is the method so you can check it. We set a -110 price on every bet, so a win pays 100 / 110 = 0.909 units and a loss costs 1 unit. The bettor starts with 100, 50, 20 or 10 units (that is, units of 1%, 2%, 5% or 10% of the bankroll) and bets one flat unit every time. We simulated 100,000 bettors for each win rate and unit size, with 500 bets each. A bettor is ruined if the balance falls below one unit, because they can no longer place a full bet.
| Unit size | Wins 50%: ruin / halved | Wins 52.38% (break-even): ruin / halved | Wins 55%: ruin / halved |
|---|---|---|---|
| 1% units (100 units) | 0.02% / 15% | none seen / 2% | none seen / 0.1% |
| 2% units (50 units) | 16% / 60% | 2% / 23% | 0.1% / 4% |
| 5% units (20 units) | 73% / 89% | 36% / 63% | 9% / 29% |
| 10% units (10 units) | 90% / 95% | 66% / 80% | 32% / 53% |
“Halved” means the balance touched half the starting bankroll at some point. Read the middle column first: a bettor who wins exactly 52.38%, the break-even rate, is still ruined (under one unit left) in 36% of runs at 5% units, and in 66% at 10% units. Even a true 55% winner is ruined in 9% of runs at 5% units. At 1% units, ruin was nearly absent: 0.02% of runs for the 50% bettor and none of the 100,000 for the other two. None seen is not the same as impossible, though, and there is a bigger catch. A coin-flip bettor at -110 still loses 0.0455 units a bet on average, or about 22.7 units (22.7% of the bankroll) over 500 bets. Small units do not create an edge. They buy time.
As a check, we compared the 50% column with a standard random-walk-with-drift approximation. It gives 73.8% at 20 units, 16.1% at 50 units and 0.03% at 100 units. Our simulation gave 73.2%, 15.6% and 0.02%. The two methods agree within about a point. Limits: the simulation assumes independent bets, a fixed -110 price and a known win rate. Real bets are messier.
What Happens If You Overestimate Your Edge?
Kelly assumes you know your true win rate. You do not. Suppose you believe you win 55% and bet full Kelly (5.5%), but your true rate is 53%, where full Kelly is only 1.3%. We simulated 40,000 bettors for 1,000 bets at several bet sizes:
| Bet size (true win rate 53%) | Typical bankroll after 1,000 bets | Share of runs below the start |
|---|---|---|
| 0.65% (half of true Kelly) | 1.06x | 39% |
| 1.3% (true Kelly) | 1.08x | 41% |
| 2.0% | 1.06x | 46% |
| 5.5% (Kelly for a believed 55%) | 0.48x | 66% |
The cliff on the right is the point. Betting as if you had a 55% edge when you really have 53% leaves the typical bettor with 0.48x the starting bankroll, and 66% of runs end below the start. The curve is lopsided: betting too little costs you a little growth, but betting too much can cost you the bankroll. Even at the best size, 41% of runs still finished below the start after 1,000 bets, because a 3% edge is small next to normal swings. Our take: if you use Kelly at all, use quarter Kelly and plug in a win rate lower than your gut says.
How Should You Size Bets for an NFL Season?
Divide your season bankroll by the 18 weeks of the NFL regular season and set a weekly cap. Take a $600 bankroll and a 2% unit, which is $12. $600 / 18 = $33.33 a week, so a $33 cap is 2.75 units, or two or three bets per Sunday.
| Item | Number | Math |
|---|---|---|
| Bankroll | $600 | |
| Unit (2%) | $12 | $600 x 0.02 |
| Weekly cap | $33 | $600 / 18 = $33.33, rounded down |
| Cap in units | 2.75 | $33 / $12 |
| Total risked in 18 weeks | $594 | 18 x $33 |
| Weeks of cap you can lose in a row | 18 | $600 / $33 = 18.2 |
Leave the unit alone until the month ends. Recalculating after every result is how people end up betting $80 in Week 3 to get back to even. Line shopping stretches the same bankroll. If one book prices a side at -110 and another at -115, a $50 bet wins $45.45 at the first and $43.48 at the second. Two dollars on one ticket is small, but it compounds across a season, which is why accounts at two or three books beat loyalty to one.
Treat any parlay as part of your normal unit, never as a bonus bet on top of it. Multi-leg bets pay more because they win less often. Our breakdown of same-game parlay risk and our guide to what a parlay bet is show what those tickets cost, and how NFL odds work covers the pricing behind them.
How Is Sports Bankroll Management Different From Casino Bankroll Management?
In sports betting you are trying to find a price better than the market. In casino games the house edge is fixed and known, so the question becomes how many hours your money lasts. Our casino bankroll management guide turns that into hours of play by game, and says to keep at least 50 bets in your bankroll for a two-hour table session. The two ideas share one rule: size the bet so a normal losing run cannot end your plan. If a bonus is part of your bankroll, read how casino bonus terms and wagering requirements work first, because rollover rules decide whether bonus money is really yours.
What Are the Biggest Bankroll Mistakes Bettors Make?
The most expensive mistake is chasing: raising your bet size to win back money you already lost. It shows up in five familiar forms.
- Doubling after a loss. Your worst bet becomes your biggest bet.
- Depositing mid-slate. A second deposit on game day is an emotional decision wearing a math costume.
- Betting a fixed dollar amount with no pot behind it. “$50 on every game” is not a plan if you never decided what $50 is a slice of.
- Counting bonus money as bankroll. A deposit match usually carries a rollover, meaning you must wager the bonus several times before it becomes withdrawable cash.
- Betting more sports than you follow. A $400 bankroll spread over four leagues is four small samples instead of one useful one.
Keep the scale in view. The AGA Commercial Gaming Revenue Tracker reports U.S. sports betting handle of $11.60 billion and revenue of $1.27 billion in July 2026. That is a 10.9% hold ($1.27 / $11.60). The books keep about 11 cents of every dollar bet. We cannot tell from AGA’s release what drives the gap above the 4.5% fee on a straight -110 bet. The point is that the fee is priced into every ticket. For the picking side, NFL betting strategies for beginners and how to bet on the NFL cover what to do once your money rules are set.
When Should You Step Back and Reevaluate?
Step back the moment betting stops being money you set aside and starts being money you need back. Clear signals to stop and reset:
- You have lost 30% or more of your bankroll and you are betting bigger, not smaller.
- You deposited outside your plan more than once this month.
- You are hiding the amount from a partner, or guessing instead of knowing your total.
- You are betting games you would not watch, just to have action.
The reset is boring on purpose. Withdraw what is left, take a week off, then rebuild the pot at half the old size with a smaller unit. Nothing about your record changes if you sit out seven days. If any of those signals feel familiar, call or text 1-800-MY-RESET, or reach the National Council on Problem Gambling helpline at 1-800-522-4700. Offshore sportsbooks are not licensed by U.S. state regulators, so read their terms and payout rules before you deposit. Never bet money you cannot afford to lose.
References
- “Kelly criterion.” Wikipedia. Cites John L. Kelly Jr., Bell System Technical Journal (1956), and Edward O. Thorp. Checked October 15, 2026. https://en.wikipedia.org/wiki/Kelly_criterion
- Covers. “The Art of Winning: A Practical Guide to Sports Betting Bankroll Management and ROI.” Checked October 15, 2026. https://www.covers.com/guides/sports-betting-bank-roll-management-and-roi
- American Gaming Association. “Commercial Gaming Revenue Tracker,” July 2026 data. Checked October 15, 2026. https://www.americangaming.org/resources/commercial-gaming-revenue-tracker/
- Michigan Gaming Control Board. “MGCB issues 33 cease-and-desist letters, warns offshore sites offer no safety net for players trying to stop.” Oct. 8, 2026. Checked October 15, 2026. https://www.michigan.gov/mgcb/news/2026/10/08/mgcb-issues-33-cease-and-desist-letters
- National Council on Problem Gambling. “Help & Treatment.” Checked October 15, 2026. https://www.ncpgambling.org/help-treatment/
