How to Calculate Bankroll Units: 4 Methods (2026)

How to Calculate Bankroll Units: 4 Methods (2026)

You've read the "1% rule" twenty times. Bet 1 unit. Don't bet more than 2%. Use Kelly. Got it.

So your bankroll is $1,150. What's 1 unit? Is it $11.50? $12? Round to $25 because that's the table minimum? And does the answer change next week when you've won $200?

That's where every guide stops, right before the math actually starts. This one walks through the four ways real bettors compute their unit size in 2026, with the dollar-by-dollar arithmetic for each, the trap nobody warns you about (round numbers eating your edge), and the decision tree for picking the method that fits your bankroll size, edge, and patience.

If you're brand new to the concept of a betting bankroll, start with what is bankroll management and come back. This article assumes you already know why units exist and want to know how to compute them precisely.

TL;DR, The Four Methods at a Glance

Four different ways to calculate units, each suited to a different player profile. All assume a $1,000 starting bankroll for the comparison.

💡 Step back to the complete bankroll management guide for how units fit into the wider survival framework.

Key Numbers to Memorize

• 1% rule = $10 unit on $1,000 → can absorb a 50-bet losing streak
• 2% rule = $20 unit on $1,000 → can absorb a 25-bet losing streak
• 5% rule = $50 unit on $1,000 → real risk of ruin during the kind of downswings variance produces
• Anything above 5% = gambling, not bankroll management

What Counts as a "Unit" (and Why It Matters)

The One-Sentence Definition

A unit is the dollar amount that represents one standard bet, typically 1-2% of your bankroll, expressed as a single number you can multiply when sizing different plays.

Why "1 Unit" Is Better Than "$25"

Saying "I bet 1 unit" forces proportional thinking. Saying "I bet $25" lets the dollar amount drift upward when you're not paying attention. Two bettors with identical staking discipline, one tracking in units, one in dollars, will diverge dramatically over a year. The unit-tracker stays sized correctly to their bankroll. The dollar-tracker keeps betting $25 even after a 30% drawdown, which now represents 4% of their shrunken roll. Same habits, opposite outcomes, purely because of how the bet is named.

This is also why professional handicappers report results in units. It strips out the bankroll variable so any subscriber, regardless of size, can apply the same picks at their own scale. For deeper context on how this works in practice, see our bankroll management guide for sports betting.

The chart above shows what happens to a $1,000 bankroll over 1,000 bets at a +1% edge under each of the four methods. Notice that the disciplined methods (fixed 1%, Kelly) produce smoother growth curves with shallower drawdowns, while aggressive methods (5% flat) deliver the highest median end-bankroll but also the highest ruin probability. There is no free lunch. Bigger units = more variance, in both directions.

Method 1: Fixed Unit (Flat Percentage)

The simplest method. Pick a percentage. Multiply your starting bankroll. That's your unit, forever, or at least until your bankroll moves more than 25% from where it started.

The Math: Step by Step

The formula is:

$\text{Unit} = \text{Starting Bankroll} \times \text{Percentage}$

Plain English: pick your percentage (usually 1% or 2%), multiply by your bankroll, and that dollar number is your unit for every single bet you place going forward, regardless of confidence, edge, or current bankroll.

Worked Example: $1,000 Bankroll at 2%

You start with $1,000. You pick 2% as your sizing rule.

That's it. Every bet you place from now on is sized in $20 increments, $20 for a standard bet, $40 if you go 2u, $60 for 3u, and so on. You don't recalculate after wins or losses. You only adjust when your bankroll has moved 25% (so down to $750 or up to $1,250).

When to Reset Your Unit

The 25% rule prevents two problems. First: keeping a $20 unit on a bankroll that has shrunk to $600 means you're now betting 3.3% per play, silently ramping up risk. Second: keeping a $20 unit on a bankroll that has grown to $1,500 means you're under-utilizing growth and missing compounding.

When Fixed Unit Fits Best

• You're new to betting (under 100 bets tracked)
• You want simplicity over optimization
• You don't trust yourself to recalculate honestly after a big loss
• You bet across multiple sportsbooks and need one consistent number

Method 2: Variable Unit (Compounding)

Fixed units ignore your current bankroll. Variable units track it. The unit recalculates as your roll moves, so wins compound into bigger bets and losses shrink them. Same percentage rule, applied to a moving target.

The Math: Recalculate Each Session

$\text{Unit}_{\text{today}} = \text{Bankroll}_{\text{today}} \times \text{Percentage}$

Plain English: at the start of each session (or every 25 bets, or every Sunday, pick a cadence), look at your current bankroll, multiply by your percentage, and that's your unit for the next stretch.

Worked Example: 30 Sessions of Compounding

Start: $1,000 bankroll, 1% rule.

Notice: at session 30, your "1 unit" has grown from $10 to $17.20, a 72% increase that mirrors the 72% bankroll growth. Same rule (1% of bankroll), automatically scaling.

The Compounding Math

If your variable unit grows your bankroll 5% per month with disciplined 1% staking and a small +EV edge, the difference vs. fixed sizing over 12 months is roughly:

• Fixed 1% unit ($10/bet): end bankroll ≈ $1,500
• Variable 1% unit (recalculated monthly): end bankroll ≈ $1,795

Compounding adds roughly 20% over a year of consistent winning. It's also brutal in reverse: variable sizing during a losing run shrinks your unit faster than fixed sizing, which protects the bankroll but slows recovery.

When Variable Unit Fits Best

• You have at least 200 tracked bets and stable results
• You're comfortable doing a 30-second calculation per session
• You're playing for the long term (12+ months) where compounding matters
• You'll honestly shrink your unit during drawdowns instead of "betting it back"

Method 3: Confidence-Weighted Unit (Tiered Sizing)

Same unit math as fixed or variable, but you bet a multiple of the unit based on how confident you are in the play. This is how professional handicappers and sharp bettors structure their releases.

Setting Up Your Confidence Tiers

There are two common tier systems. Pick one and stick with it for at least 100 bets before judging.

The 1u/2u/3u Tier System

The most common setup, used by most published handicappers:

The 1u/2u/3u/5u "Lock" System

Adds an extreme tier for the rare highest-confidence plays. Use carefully, most bettors abuse the 5u tier.

The 5u tier is dangerous because human bettors massively overestimate their confidence. If you find yourself betting 5u more than once a week, your tiers are calibrated wrong.

Worked Example: $1,000 with Tiered Sizing

Bankroll: $1,000. Base unit: 1% = $10.

Notice the maximum exposure on a single bet is still 5% ($50), the same as the aggressive flat-betting cap. Confidence-weighting doesn't expand your max risk; it just concentrates more of your bankroll into the spots you actually believe in, and bets less on the spots you don't.

When Confidence-Weighted Fits Best

• You can honestly differentiate "good read" from "great read"
• You have at least 100 bets of data per tier (so you can verify the tiers actually correlate with results)
• You're disciplined enough to bet 1u on plays you only kinda like, instead of skipping or upgrading them
• You want a step toward Kelly without the math overhead

Method 4: Edge-Weighted Unit (Kelly Fraction)

Kelly sizes each bet to your measured edge, not your gut feeling about it. The bet grows with the edge and shrinks when the edge shrinks. It's the only method derived from first-principles probability theory rather than a rule of thumb.

The Math: Edge × Bankroll = Bet

The Kelly formula for a binary bet is:

$f^* = \frac{bp - q}{b}$

Where f* is the fraction of bankroll to bet, b is the decimal odds minus 1, p is your estimated win probability, and q = 1 − p.

Plain English: if you can estimate the true probability of winning (and you've checked yourself enough times to trust that estimate), Kelly tells you exactly what fraction of your bankroll maximizes long-term growth. The Kelly fraction is one of the four sizing modes built into our betting bankroll calculator, punch in your edge and odds and it returns the recommended stake.

Worked Example: $1,000 at +5% Edge on -110 Odds

Setup: you've found a bet with a 55% win probability priced at -110 (decimal 1.91). Your edge: 5%.

Step 1: convert odds. Decimal 1.91 → b = 0.91. Step 2: plug in. f* = (0.91 × 0.55 − 0.45) / 0.91 = 0.05 / 0.91 ≈ 0.055. Step 3: full Kelly bet = $1,000 × 0.055 = $55, or 5.5 units (where 1u = 1% = $10).

Full Kelly says risk 5.5% of your bankroll on this single bet. That's aggressive, and full Kelly almost guarantees a 50%+ drawdown at some point during normal variance, which most bettors emotionally cannot tolerate.

Why Most Bettors Use Half-Kelly

In practice, almost no serious bettor uses full Kelly. They use Half-Kelly or Quarter-Kelly, bet half (or a quarter) of what the formula recommends.

Half-Kelly captures most of the long-term growth (~75% of full Kelly) with roughly half the variance. Quarter-Kelly is what most disciplined bettors actually use day to day. The trade-off is real: if you can't measure your edge accurately within ±2 percentage points, Quarter-Kelly is the safe default. Full Kelly assumes perfect edge measurement, which is rare.

When Edge-Weighted Fits Best

• You have 500+ tracked bets and a measured edge to within 1-2 percentage points
• You're emotionally able to handle 30-50% drawdowns without tilting
• You can compute or look up Kelly fractions in real time before placing each bet
• You actively scout edge, you don't bet without one

If any of those don't fit, use confidence-weighting or fixed sizing instead. Kelly with a poorly measured edge is worse than flat 1% betting, you'll size up on bets that aren't actually +EV and bankrupt yourself faster than you would have flat-betting them.

Pick Your Method: Decision Tree

Walk through these questions in order. Stop at the first one that matches.

Beginners (0-100 Bets)

Use: Fixed 1% unit, no tiers, no Kelly.

You don't have enough data to know if you have an edge, let alone measure it. The goal at this stage is staying alive long enough to gather data, which means minimizing variance, not maximizing growth. Round to whole-dollar units. Bet flat. Track every bet.

Calculation: $1,000 bankroll → 1u = $10 → every bet is $10 until your bankroll moves 25%.

Intermediate (100-500 Bets, Mixed Results)

Use: Variable 1-2% unit OR confidence-weighted (1u/2u/3u tiers at 1% base).

You have enough data to know roughly how your picks perform. Variable sizing lets you compound steady growth without forcing you to estimate edge per bet. Confidence-weighting works if you can honestly differentiate your reads, but only if your "Best Bet" tier actually wins more often than your "Standard Play" tier when you check the data after 100 bets.

Calculation: $1,000 bankroll → 1u = $10–$20 (recalculated weekly) OR 1u = $10 with 2u = $20 / 3u = $30 multipliers.

Advanced (500+ Bets, Tracked Edge)

Use: Quarter-Kelly or Half-Kelly based on measured per-bet edge.

You've earned the right to size by edge. Just remember: Kelly only works if your edge measurement is accurate. Quarter-Kelly is forgiving of measurement error. Full Kelly assumes you're a probability god, which you're not.

Calculation: Per bet: f* = (b·p − q)/b, then bet f*/4 of current bankroll.

The Round-Number Trap

This catches every method, especially intermediate bettors who like clean numbers. If your unit calculation says $23, do not bet $25. Here's why:

• $23 → $25 = 8.7% increase in bet size
• Compound that over 500 bets, all rounded up
• You're now betting at 2.18% effective rate when you intended 2%
• Over a year, that's roughly a 9% larger drawdown when variance hits

If you must round, round down, never up. $23 → $20, not $25. Or just bet the exact number. Most modern sportsbooks accept odd amounts. The discipline of betting $23.47 instead of $25 is, no exaggeration, worth thousands of dollars over a serious bettor's career.

Tracking Units Over 100 Bets

Once you've picked a method, the only thing left is tracking. Without tracking, you can't tell if your method is working, and after a few months, you can't even tell what your unit was supposed to be.

A minimal tracking spreadsheet has six columns:

That's it. Date, what, how much, what odds, did it win, what's your running tally. After 100 bets, look at the running total. If it's +5u or better, your method is working. If it's between -5u and +5u, you're at break-even and the method is fine. If it's worse than -5u over a sample of 100, your edge is missing or your method is misaligned with your skill, re-read the decision tree and reset.

For an automatic version of this calculation that includes risk-of-ruin and recommended unit sizing, run your numbers through our universal bankroll calculator. It handles the math so you can focus on the picks.

The calculator above takes your bankroll, sizing method, confidence tier (if applicable), and edge percentage (if Kelly), and returns your 1u, 2u, 3u, 5u dollar amounts plus the recommended bet size for the current play. Use it before every betting session, twenty seconds of math saves a lot of guessing.

Common Mistakes That Wreck Unit Sizing

Forgetting to Recalculate

The most common mistake. You set your unit at $20 (2% of $1,000). After a bad month your bankroll is $700. Your $20 unit is now 2.86% of your roll. You're staking heavier than you intended without realizing it. Over a 50-bet recovery period, that's the difference between climbing back to even and digging deeper.

Set a recurring calendar reminder: "Friday night, recheck bankroll, recalculate unit." Five seconds of work, prevents months of accidental over-staking.

Confidence Inflation

Every confidence-weighted bettor goes through this. Month 1: 60% standard plays, 25% strong, 15% best bets. Month 6: 30% standard, 40% strong, 30% best bets. Same picks. Same data. The bettor just got bored of betting 1u and started "upgrading" everything to feel more engaged.

The fix: track tier results separately. If your 3u "best bets" win at the same rate as your 1u "standard plays" after 100 bets per tier, your confidence calibration is broken. Either retire the tier system and go flat, or rebuild your tiers from the data, the lower-confidence tier is right and the higher-confidence tier is your imagination.

Reloading From Personal Money

Your bankroll drops to $500. You add $500 from checking to "get back to $1,000" and resume betting at the original $20 unit. You've now broken every rule of bankroll management at once: ignored the recalculation, restored old unit math to a now-mixed-money pot, and erased the discipline that the smaller bankroll was teaching you. Don't reload. Recalculate.

Sizing Up After a Hot Streak

You've won 10 in a row. Surely your edge is bigger than you thought. Time to bet 5u instead of 2u, right?

Wrong, unless you have data showing your edge has actually changed. Hot streaks are mathematically expected. A 55% bettor will hit 10-in-a-row roughly once every 600 bets. It feels rare but it isn't. The correct response to a hot streak is to recalculate your variable unit (which naturally grows because the bankroll grew) and continue at the same percentage. The math doesn't care about feelings.

Mixing Methods Mid-Session

You start the session betting flat 1%. Halfway through, you take a "lock" play at 5u because it feels different. Now you don't know which method you're using. Whatever your results are at the end of the night, you can't attribute them to the strategy because you didn't use one consistently.

Pick a method. Use it for 100 bets minimum. Evaluate. Then iterate. Don't blend methods inside a session. If you want a permanent reference for which method matches your numbers, save our free bankroll tool and run it before every session, same input, same output, no temptation to "feel" your way to a different answer mid-stream.