SPORTS BETTING · RISK
Sports Betting Bankroll Management: Unit Size, Kelly Criterion and How to Track Every Bet
At standard −110 odds you need to win 52.4% of your bets just to break even, and even a bettor who wins 53% will hit eight losses in a row at some point in a 500-bet season about half the time. Bankroll management is what keeps a normal losing streak from ending your season. Here is how to size a unit, when the Kelly criterion helps and when it hurts, which numbers to track, and a calculator to run your own.
THE SHORT ANSWER
What is bankroll management in sports betting?
Bankroll management is a set of rules for how much money you set aside for betting and how much of it you risk on each bet. You fund a separate bankroll with money you can afford to lose, bet a fixed slice of it called a unit (usually 1–2%), and never raise stakes to win back losses. The goal is not to win more on any single game. It is to survive the losing streaks that happen even to good bettors, and to produce clean records that show whether you have an edge at all.
52.4%
The win rate you need on standard spread and total bets just to stay even.
4.5%
The hold on a −110/−110 market: you risk $110 to win $100 on either side.
1–2%
Of the bankroll per bet. At 5% a real edge still busts about 4 in 10 bettors in our simulation.
~1 in 2
Chance of eight straight losses in 500 bets for a bettor who wins 52.4%.
THIS SEASON
Why bankroll management matters more in late 2026
The NFL, college football, the NBA and the NHL all run at once from late October, so many bettors place more bets per week now than at any other time of year. More bets means more streaks, and a bankroll without rules tends to get spread across too many games, parlays and live bets on a busy Sunday.
There are also more places to bet. Trackers count legal sports betting in 39 states plus Washington, D.C. as of October 2026, with online betting in about 30 of them; rules, minimum ages and available bet types differ by state, so check your state regulator. At the same time, CFTC-regulated prediction markets such as Kalshi list contracts on game outcomes. On February 4, 2026 the CFTC withdrew its 2024 proposal that would have banned sports and political event contracts and said it would write new event-contract rules instead, while a number of states have sued prediction market operators and federal courts have ruled both ways. The legal picture is still moving.
UNIT SIZE
What is a unit in sports betting, and how big should it be?
A unit is your standard stake, expressed as a share of the bankroll. If your bankroll is $2,000 and one unit is 1%, a "1-unit bet" is $20 and a "2-unit play" is $40. Units let bettors with different bankrolls compare results: "+12 units this month" means the same thing whether the unit is $10 or $500. When a tipster talks about a "5-unit" or "7-unit play", that is a confidence scale, and it only works if the base unit is small enough that five or seven of them is still a reasonable risk.
How big is too big? We simulated 4,000 seasons of 1,000 bets at −110 for a bettor who genuinely wins 53% of the time, which is a small but real edge (about 1.2% expected return per dollar bet). Every run used flat stakes set from the starting bankroll.
| Unit size (flat) | Typical worst drawdown | Chance of losing half the bankroll at some point | Chance of going broke |
|---|---|---|---|
| 1% of bankroll | 25% | 7% | under 1% |
| 2% of bankroll | 44% | 39% | 6% |
| 3% of bankroll | 58% | 61% | 17% |
| 5% of bankroll | 82% | 82% | 40% |
Monte Carlo simulation, 4,000 runs of 1,000 bets each at −110 with a true win rate of 53%; "typical worst drawdown" is the median of each run's largest peak-to-trough fall. With the same edge about one run in three still finished the 1,000 bets below the starting bankroll. Illustrative, not a forecast.
The lesson is the one traders learn with position sizing: the edge is the same in every row, and only the stake changes the outcome. A bettor without an edge, at 52.4%, does worse in every row.
THE ODDS MATH
Implied probability, the vig and your break-even win rate
Every staking rule starts from one number: the probability the odds imply. American odds convert like this:
| What | Formula | Example |
|---|---|---|
| Implied probability, negative odds | |odds| ÷ (|odds| + 100) | −110 → 110 ÷ 210 = 52.4% |
| Implied probability, positive odds | 100 ÷ (odds + 100) | +150 → 100 ÷ 250 = 40.0% |
| Decimal odds | 1 + 100 ÷ |odds| (negative) or 1 + odds ÷ 100 (positive) | −110 → 1.909; +150 → 2.50 |
| Book's margin (overround) | Sum of both sides' implied probabilities − 100% | −110/−110 → 52.4% + 52.4% = 104.8% |
| Hold | 1 − 1 ÷ overround | 1 − 1 ÷ 1.048 = 4.5% |
| Expected value per $1 | p × (decimal − 1) − (1 − p) | 55% at −110 → 0.55 × 0.909 − 0.45 = +$0.05 |
The implied probability is also your break-even win rate at those odds. At −110 you need 52.4%; at +150 you need 40%. A bet only has positive expected value when your honest estimate of the probability is above that number, and that estimate is exactly what most bettors overrate.
THE KELLY CRITERION
Kelly criterion for sports betting, and why bettors use a fraction of it
The Kelly criterion, from John Kelly's 1956 paper and popularized for betting by Edward Thorp, gives the share of the bankroll that maximizes long-run growth when you know your edge:
| Your estimate | Full Kelly | Half Kelly | Quarter Kelly |
|---|---|---|---|
| 55% at −110 (b = 0.909) | 5.5% | 2.8% | 1.4% |
| 45% at +150 (b = 1.5) | 8.3% | 4.2% | 2.1% |
| 53% at −110 | 1.3% | 0.65% | 0.33% |
| 52.4% at −110 (break-even) | 0% | 0% | 0% |
Full Kelly is only optimal if your probability is right, and in sports it rarely is. Suppose you think a −110 bet wins 55% of the time and stake full Kelly (5.5%), but the true probability is 53%. Your expected long-run growth turns negative: you lose money betting on a game where you actually have an edge, purely because the stake is too big. Half Kelly at the same overestimate roughly breaks even, and quarter Kelly still grows. If the true probability is 55% after all, quarter Kelly keeps about 43% of full Kelly's growth rate with far smaller swings.
VARIANCE
How likely is a losing streak?
Streaks feel like signals. Usually they are noise. This is the probability of seeing at least one run of losses of a given length, for a bettor who wins 52.4% of bets at −110 (exactly break-even) and one who wins 55%:
| Bets placed | 5 losses in a row | 7 in a row | 8 in a row | 10 in a row |
|---|---|---|---|---|
| 100 bets, 52.4% win rate | 74% | 25% | 12% | 3% |
| 250 bets, 52.4% | 97% | 52% | 29% | 7% |
| 500 bets, 52.4% | over 99% | 77% | 50% | 14% |
| 500 bets, 55% | over 99% | 64% | 37% | 9% |
Five bets a day through football season is about 500 bets. At that pace an eight-loss streak is a coin flip even for a break-even bettor, and still more than one chance in three for a good one. With 2% flat units, eight losses cost 16% of the bankroll; with 5% units they cost 40%. Plan the unit for the streak you will get, not for the average.
A streak is also when bad decisions happen: bigger stakes to "get it back", extra live bets, a late parlay. Traders call this revenge trading, and the same psychology and fixes apply to betting: a daily or weekly stop, stakes set before the games, and a written log.
BANKROLL CALCULATOR
Unit, Kelly and streak risk
Enter your bankroll, unit and one bet. The calculator converts the odds, shows your break-even and edge, the Kelly stake at the fraction you choose, the chance of a losing streak over the season and, if you add the closing odds, your closing line value.
The streak chance treats every bet as independent at your win probability. Your probability is an estimate; if it's too high, every number here is too optimistic. Results stay in your browser.
MEASURE IT
ROI, yield and closing line value: are you actually good?
Win rate alone says little, because a 45% win rate on +150 underdogs is profitable and a 55% win rate on −150 favorites is not. Three numbers do the work:
| Metric | Formula | Example |
|---|---|---|
| Yield (ROI on stakes) | Net profit ÷ total amount staked | $160 profit on 200 bets of $20 ($4,000 staked) = 4.0% |
| Return on bankroll | Net profit ÷ starting bankroll | $160 on a $1,000 bankroll = 16% |
| Profit in units | Net profit ÷ unit size | $160 ÷ $20 = +8 units |
| Closing line value (CLV) | Your decimal odds ÷ closing decimal odds − 1 | Bet at −110 (1.909), game closes −130 (1.769): +7.9% |
Sportsbooks often quote "ROI" when they mean yield; say which one you use and keep it consistent. A yield of a few percent over hundreds of bets is good; double digits over a short run is mostly luck.
Closing line value is the most useful early signal. The closing line is the market's final price, after all the information is in. If you regularly get better odds than the close (−110 on a side that closes −130), you are buying probability cheaper than the market's final estimate, which is the closest thing betting has to a measurable edge. Record the closing odds for every bet and look at the average after a few hundred bets.
THE STAKING PLAN
Bankroll rules that hold up through a season
- Fund a separate bankroll. Money you can lose without touching rent, bills or savings. Don't top it up after losses; that's chasing with extra steps.
- Pick one unit and write it down. 1% if you are new or bet many games a day, up to 2% with a long record. Recalculate monthly or when the bankroll moves 20–25%.
- Cap any single bet. 2–3 units at most, however strong the play looks. Kelly sizes above that are a sign your probability is too high.
- Set a daily and weekly loss limit. For example 5 units a day and 10 a week. Use the deposit and loss limits in your sportsbook account; legal books are required to offer responsible-gaming tools.
- Treat parlays and live bets as a small, separate budget. Their margin is higher, and they're where stakes drift.
- Shop for the best price. −105 instead of −110 lowers your break-even from 52.4% to 51.2%. It's the cheapest edge there is.
- Log every bet the moment you place it, including the ones you'd rather forget.
THE RECORD
What to track for every bet
A sportsbook's history screen shows wins and losses. It doesn't show your closing line value, your results by bet type across several books, or whether you kept to your unit. A spreadsheet does. The columns that matter:
| Column | Why it matters | Formula (example) |
|---|---|---|
| Date, sport, league, matchup | Lets you split results by sport and season. | typed in |
| Bet type and sportsbook | Shows whether spreads, totals, props or parlays make or lose money, and which book gives the best prices. | dropdown |
| American odds and decimal odds | Every other calculation runs on decimal odds. | =IF(D2<0,1+100/ABS(D2),1+D2/100) |
| Stake and stake in units | Checks that you kept to the unit. | =F2/$B$1 |
| Result and profit/loss | Win, loss, push. Profit = stake × (decimal − 1) on a win. | =IF(G2="W",F2*(E2-1),IF(G2="L",-F2,0)) |
| Closing odds and CLV | The early signal of an edge. | =E2/I2-1 |
| Running bankroll and drawdown | Shows the current streak and how deep it is. | =MAX($J$2:J2)-J2 |
| Rule followed? (Y/N) | Separates bad luck from bad discipline. | dropdown |
The structure is the same as a trading journal in Excel: one row per position, formulas for the results, and a summary that answers "where do I actually make money?". After a few hundred rows, filter by sport, bet type and book; the losing segments are usually obvious, and cutting them is the quickest improvement available.
READY-MADE TEMPLATE
A sports betting journal
If you'd rather start with the columns, formulas and reports already built, there is a ready-made Sports Betting Journal for Excel and Google Sheets; check the product page for its current features.
Sports Betting Journal Template for Google Sheets & Excel
The current listing describes logging the date, sport, league, matchup, bet type, odds, stake, sportsbook and status of each bet, with automatic profit/loss, ROI, yield and win rate, a bankroll view with risk per bet and maximum drawdown, closing line value and juice by sportsbook, a profit calendar with streaks, and breakdowns by sport, bet type, book and odds range. American, decimal and fractional odds are supported.
- Bankroll, risk per bet and drawdown
- CLV and results by sportsbook
- Works offline, no subscription
Prefer a dark interface? The same journal is available as the Sports Betting Journal Template (Dark Mode).
Product features and compatibility are listed on the product page and may change. A journal records results; it doesn't make bets profitable.
RESPONSIBLE GAMBLING
When to stop, and where to get help
No staking plan beats the sportsbook's margin on its own, and no bankroll is worth your health or your relationships. Warning signs include betting more to win back losses, borrowing or using bill money to bet, hiding bets from people close to you, and feeling restless when you try to cut back.
- National Problem Gambling Helpline (NCPG): call or text 1-800-MY-RESET (1-800-697-3738), 24/7, free and confidential. The older number, 1-800-522-4700, still works, and you can chat online.
- Self-exclusion and limits: every legal sportsbook lets you set deposit, wager and time limits or exclude yourself; most states also run a statewide self-exclusion list.
- In a crisis: call or text 988, or 911 in an emergency.
Since January 2026 the national helpline number is 1-800-MY-RESET; 1-800-GAMBLER now belongs to the Council on Compulsive Gambling of New Jersey and is still printed on some ads.
FAQ
Sports betting bankroll questions
What is a unit in sports betting?
A unit is your standard bet size, set as a percentage of your bankroll, usually 1–2%. With a $2,000 bankroll and a 1% unit, one unit is $20. Bettors report results in units so that records are comparable whatever the bankroll size.
What percentage of my bankroll should I bet?
Most bankroll guides suggest 1–2% per bet, and no more than 2–3% on any single bet. In our simulation of 1,000 bets with a real 53% edge at −110, 1% stakes had a typical worst drawdown of 25%, while 5% stakes went broke in about 40% of runs.
How much money do I need to start a bankroll?
There is no minimum; the bankroll is whatever you can afford to lose entirely. What matters is that the unit fits it: at 1% per bet, a $500 bankroll means $5 bets. If that feels too small to be interesting, that is a signal to bet less often, not to raise the unit.
Is the Kelly criterion good for sports betting?
It is the right way to size bets when you know your edge, but in sports you only have an estimate. If you overestimate your win probability, full Kelly can lose money even on bets with a real edge, so most bettors use a quarter to half Kelly and cap each bet at 2–3% of the bankroll.
What win rate do you need to be profitable in sports betting?
At standard −110 odds you need 52.4% just to break even, because you risk $110 to win $100. At −105 the break-even falls to 51.2%. On underdogs at +150 you need only 40%, so compare win rate with the odds, not on its own.
What is a good ROI in sports betting?
Over hundreds of bets, a yield (profit divided by total amount staked) of a few percent is good, and most bettors are negative because of the sportsbook's margin. Double-digit returns over a short run are usually luck; over 500 bets the random error on a win rate is still about ±2.2 percentage points.
What is closing line value (CLV)?
CLV compares the odds you got with the final odds before the game. If you bet at −110 and the line closes at −130, you got about 7.9% better odds than the market's final price. Consistently positive CLV is one of the best early signs of an edge.
How many units should you bet per day?
Set a daily limit rather than a target, for example no more than 5 units at risk in a day and 10 in a week. More bets do not create an edge; they only make the swings faster if you don't have one.
Sources and further reading
- J. L. Kelly Jr.: A New Interpretation of Information Rate (Bell System Technical Journal, 1956)
- Edward O. Thorp: The Kelly Criterion in Blackjack, Sports Betting and the Stock Market (2006)
- CFTC: Withdraws event contracts rule proposal and staff sports event contracts advisory (February 4, 2026)
- American Gaming Association: Interactive map of legal sports betting by state
- National Council on Problem Gambling: 1-800-MY-RESET adopted as the National Problem Gambling Helpline number (January 2026)
- National Council on Problem Gambling: National Problem Gambling Helpline (call, text, chat)
Checked October 29, 2026. Odds formulas, Kelly stakes, streak probabilities and the drawdown simulation were calculated by SpreadsheetsHub; the simulation is illustrative, not a forecast. State laws and prediction market rules change often. This article is general information about managing risk, not betting, legal, tax or financial advice, and not an encouragement to gamble. Must be 21+ (18+ in some states) and physically located in a legal state to bet.
