Sports Betting Bankroll: Unit Size, Staking Plans, and Discipline

It was a sunny Saturday. You liked three sides, felt strong, and doubled your stakes “just this once.” The first pick lost on a weird call. You chased a live bet. By the night game, you were down and angry. The picks were not that bad. The plan was. This is how smart people lose. Not from bad ideas, but from no rules.

What a bankroll is (and what it is not)

Your bankroll is money set aside only for betting. It is not rent, food cash, or a card balance. Ring-fence it. Treat it like a small shop fund. If it hits zero, the shop closes.

We use a “unit” to size each bet. A unit is a fixed slice of your bankroll. It keeps your risk steady and emotions low. For most non-pros, a good start is 0.5% to 2% of your roll per unit. If you have $2,000, a 1% unit is $20. One game, one unit. Clear and calm.

Units let you compare results without dollar bias. A +7-unit month reads the same for a $2k or a $20k roll. Units also force you to think in risk, not in “I feel great today.”

Math pit stop: EV, variance, and why units exist

Expected value (EV) is the long-run average of a bet. If you could play the same spot many times, EV tells you the mean gain or loss per play. For a simple intro to expected value, this resource is clear and short.

Even good bets swing. That swing is variance. In plain words: your results will bounce around the mean. Runs happen. You can be right and still lose five in a row. Or be lucky and win with bad picks for a week. Here is more on variance if you want a deeper idea in simple terms.

Quick EV example: say a book posts +110 (decimal 2.10). You think the true win chance is 49%. EV per $100 is (0.49 × $110) − (0.51 × $100) = $53.9 − $51 = +$2.9. That is +2.9% EV. Nice. But you can still lose three of the next three. Units help you live to see the average show up.

The real choice: which staking plan fits you

There is no one “best” plan. You match the plan to your goal, your real edge (if any), how wild your markets are, and your nerves. The plan must keep you in action when you are cold and keep you from going wild when you are hot.

Staking plans in practice

Below are common plans, what they do, and how they feel on the way up and down. We also compare them in a table you can scan fast.

Flat staking

You bet one unit per pick. Simple and strong for most people. Typical unit: 0.5% to 1.5% of bankroll. Pros: easy, steady, low stress. Cons: you do not “scale up” when your roll grows, unless you reset the unit now and then.

Percent-of-bankroll (proportional)

Your unit is a set percent of your current roll, so it floats. If the roll goes up, your unit grows. If you draw down, your unit gets smaller. Pros: natural compounding, self-braking on downswings. Cons: can feel slow after a hit, and it can whipsaw if your roll moves a lot.

Kelly and fractional Kelly

Kelly uses your edge and the odds to set the “optimal” stake for growth. The Kelly criterion paper gives the core idea: bet a fraction of bankroll equal to an edge-based formula. The math looks neat, but errors in your edge guess hurt fast.

Many smart bettors use a cut of Kelly, like 0.25× to 0.5×. This smooths the ride. For more on why that helps in real life, see Thorp’s treatment of Kelly.

Quick Kelly example on that +110 spot with 49% win chance: b = 2.10 − 1 = 1.10; p = 0.49; q = 0.51. Kelly fraction f* = (b×p − q) / b = (1.10×0.49 − 0.51)/1.10 = (0.539 − 0.51)/1.10 ≈ 0.026. Full Kelly says risk ~2.6% of roll. Half Kelly would be ~1.3%. If your edge guess is high by even a little, full Kelly can be too hot.

Confidence scaling (A/B/C or 0.5/1/1.5 units)

You rate your pick and bet 0.5, 1, or 1.5 units based on set rules. This is fine if your signal grades are real and backtested. It is bad if “A” just means “I like this team.” Build clear rules: market moves, model gap, injury info, etc.

Progressions (Martingale, Fibonacci) — why to pass

These increase stake after a loss to “win it back.” This looks safe on paper and deadly in life. A long streak burns your roll or hits table limits. See a clear note on risk of ruin math. The short point: progressions do not change EV. They only reshape how fast you can go broke.

Staking plans at a glance

Discipline is not a mood. It is a system.

Set rules when you are calm. Follow them when you are not. Here is a clean, small system you can use today.

• Pre-bet checks: source of edge, unit size rule, and max daily exposure (for example, cap risk at 5% of roll per day).
• No live chase: do not add units to “get even.” If live is part of your plan, size it like any bet with a unit.
• Cool-off triggers: after a tilt sign (two fast impulsive bets, heart rate up), take 30 minutes. Drink water. Re-read rules.
• Stop rules: at −20% roll, cut unit by 25–50%. At −35%, pause 72 hours and audit your data and model.
• Weekly review: did you beat the close? Did you keep unit size? Any leaks?

Track every bet. Log the stake, odds, result, and the line at close. “Closing line value (CLV)” is a key check. If your price is often better than the final price, your process may be sound, even if short-term results are rough. For a market view of this idea, see this paper on closing line value (CLV) and betting efficiency.

• Mon: +0.6u at +105. CLV +3 cents.
• Tue: −1.0u at −110. CLV −2 cents. Note: late entry.
• Wed: +1.0u at +120. CLV +6 cents.
• Thu: 0u, passed two leans (no edge).
• Fri: −0.5u at +100. CLV 0.
• Sat: +0.8u at −102. CLV +4 cents.
• Sun: −0.4u at +115. CLV +1 cent. Net week: +0.5u.

Edges, limits, and real-world checks

Sports markets can be sharp. Lines often move fast on good info. Some work finds parts of the market are near efficient, at least in top leagues. A good primer on market efficiency in sports betting gives the big picture. What this means for you: you must earn your edge with slow, steady work, and you must protect it with strong risk rules.

Books have limits. Some cut bet size if you win or if you beat the close a lot. Some offer worse prices on niche leagues. Before you think “my model died,” check the shop. Are you getting second-best lines? Are limits small at key times? Independent reviews of operators help you avoid traps with price, caps, and payouts. If you also play on your phone or mix in casino games, you can view the guide to vetted mobile operators for safe play habits and smoother banking. One careful choice up front can save months of grind.

Tools you can set up in two minutes

Make a Google Sheet with these columns: Date, Sport/Market, Pick, Odds (price), Stake (units), Result (W/L/Push), P/L (units), Bankroll ($), Close Line, CLV (Close − Your Price), Notes.

• Formula tip: Bankroll after bet = prior bankroll + (units × unit_size × profit/loss at odds).
• Backfill 25 past bets to see real swings. If a −10u run would have kept you calm, your unit is fine. If not, lower it.
• Once a month, if your roll changed by 20%+, reset your unit to 1% of the new roll (flat staking) or keep percent-of-roll.

Quick answers to hard questions

What if my bankroll is only $300?

Use flat 0.5%–1% units. That is $1.50–$3. Keep fees low. Bet fewer markets. Your edge must beat juice first. Size is not pride; it is survival.

Should I raise unit size after a hot streak?

No sudden jumps. If you use flat staking, review monthly and adjust unit to 1% of the new roll. If you use percent-of-roll, the unit grows on its own.

Is 1% always right?

No. If your markets swing a lot (long shots, props), try 0.5% units. If you have a stable, proven edge and a cool head, 1%–1.5% can be fine.

Can I mix staking plans?

Yes, but be clear. For example, flat for sides/totals, and 0.5× Kelly for props you have modeled. Do not switch mid-tilt.

How many bets until I know if I am +EV?

Often 500+ bets in one market to feel sure. Track CLV and sample size. If you do not beat close over time, question the edge.

What one habit protects a bankroll most?

Never chase. If you feel heat, stop. Rules made in calm save you in chaos.

Responsible gambling and compliance

Bet only what you can afford to lose. Keep betting money apart from life money. If betting feels out of control, get help. In the U.S., see the National Council on Problem Gambling. In the U.K., see GamCare. They offer free, private support.

Know your local laws and age rules (often 18+ or 21+). Operators must follow rules, but you have duties too. For official guidance, see the UK Gambling Commission website or your local regulator.

Author’s note and method

Author: Jordan Hale, sports analytics researcher. I have 8+ years working on models for NFL, NBA, and soccer markets, and five years in risk control in finance. I do not take fees from sportsbooks. I test ideas with simple sims and with old bet logs. I update this guide every 3–6 months. Math examples use public odds and simple compounding logic. This is not financial advice. It is risk management for a risky hobby.

TL;DR

• Pick a unit (0.5%–1.5%) and stick to it. Flat or percent-of-roll both work if you follow rules.
• Have a drawdown plan: cut size at −20%; pause and audit at −35%.
• Track CLV and results. If you cannot beat close, fix process before you raise size.

Sources

• Khan Academy: Probability and expected value basics — https://www.khanacademy.org/math/statistics-probability/probability-library
• Encyclopedia Britannica: Variance — https://www.britannica.com/science/variance-statistics
• Kelly (1956) original paper — https://www.princeton.edu/~wbialek/rome/refs/kelly_56.pdf
• Edward O. Thorp on Kelly — https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2042658
• Wizard of Odds: Risk of ruin — https://wizardofodds.com/gambling/house-edge/#risk-of-ruin
• CLV and market efficiency study — https://papers.ssrn.com/sol3/papers.cfm?abstract_id=907103
• Market efficiency in sports betting (Levitt) — https://www.jstor.org/stable/3216860
• National Council on Problem Gambling — https://www.ncpgambling.org/
• GamCare — https://www.gamcare.org.uk/
• UK Gambling Commission — https://www.gamblingcommission.gov.uk/