Mechanics · 8 min read

Bankroll management: fixed unit, percentage, and Kelly

Three staking systems, plus when each one fails. The simplest answer (fixed unit) is usually the right one.

What a bankroll is, and why it matters

A bankroll is the money you have set aside, in advance, that you can afford to lose without affecting the rest of your life. Two things follow:

1. The bankroll is bounded. It is not your bank account. It is not what you might earn next month. It is a number you wrote down before you started.
2. Stake size is a fraction of that number, not a number you pull out of the air. “Bet $200 on this one because I feel good about it” is not bankroll management.

The discipline of staking a fraction of a fixed bankroll is what keeps a bettor solvent long enough for any real edge to show up. Without it, even a profitable system can be wiped out by a normal losing streak (see variance).

The three staking systems worth knowing

Fixed unit (recommended for most punters)

Decide a unit size as a fixed dollar amount — typically 1% or 2% of your starting bankroll. Bet that amount on every pick, regardless of confidence.

• $2,000 starting bankroll, 1% unit = $20 per bet.
• Every bet is $20 unless you re-evaluate the bankroll (e.g. quarterly).

Why this works: it removes the largest single source of variance — your stake-size decision. A real edge can compound. Tilt cannot scale you out of business in a single day.

Why it might not feel right: every losing bet feels small, but every winning bet also feels small. Punters who want to be entertained sometimes bristle at fixed unit. The answer is to decide upfront which one you want — entertainment or compounding edge — and stick to it.

Percentage of current bankroll

Same idea but the unit floats: always bet 1% of your current bankroll, not your starting bankroll. Stakes shrink during drawdowns and grow during winning runs.

• $2,000 → bet $20.
• Down to $1,500 → bet $15.
• Up to $2,500 → bet $25.

Pros: builds in a soft stop-loss. The bankroll mathematically cannot hit zero from proportional staking alone (though in practice you bet integer amounts and bookmakers have minimum stakes, so you can run aground at small balances).

Cons: you compound losses by reducing the stakes that could recover them, and you compound wins by inflating stakes during what might be a variance-driven upswing. For most punters this is a wash vs fixed unit.

Kelly Criterion (and why most punters should not run it raw)

Kelly is the mathematically optimal stake size if you know the true probability and the offered odds. The formula:

Kelly fraction = (probability × decimal odds − 1) / (decimal odds − 1)

Example: you back a market priced at 2.10 that you believe has a 55% true probability. Kelly = (0.55 × 2.10 − 1) / (2.10 − 1) = (1.155 − 1) / 1.10 = 0.155 / 1.10 ≈ 14.1%. Kelly says stake 14.1% of bankroll. That is a huge stake.

The catch: Kelly is optimal only if your probability estimate is exactly right. If you are off — even by a small amount — Kelly stakes can ruin you. In practice, almost every sharp bettor uses fractional Kelly: stake one-quarter or one-half of what raw Kelly recommends. That cushions against estimation error.

Recommendation: do not run raw Kelly unless you have a back-tested model with well-calibrated probabilities and hundreds of bets of evidence. For the average bettor, fixed unit at 1-2% of bankroll is functionally close to half-Kelly for typical edges and far harder to mis-execute.

Bankroll size vs unit size

The right starting bankroll is the one that lets you absorb a normal drawdown without emotional pressure. Rules of thumb:

• A 5% edge bettor at flat 1% stakes can see a drawdown of 30+ units in normal variance. Plan for a 50-unit drawdown.
• That means: starting bankroll = 50 × unit. If you can spare $1,000 to lose without it affecting anything else, your unit is $20 and you flat-stake $20 per bet.

Punters who size their unit relative to last weekend's wages tend to find themselves out of bankroll just as the edge would have started showing.

Re-evaluation cadence

Decide upfront when you will recompute the unit. Quarterly is reasonable: at the end of each quarter, look at your bankroll and adjust the unit accordingly. Weekly is too often (you adjust to noise). Annually is too rare (you miss real changes in your bankroll). Whatever cadence you pick, write it down and stick to it.

What bankroll discipline does not do

• It does not turn a -EV bettor into a +EV one. The math compounds in either direction. Discipline preserves edge if it exists; it does not invent it.
• It does not make winning streaks more frequent. It just makes you alive long enough for them to occur.
• It does not eliminate the discomfort of losses. Five units down still feels like something. The point is that it does not threaten the bankroll's ability to keep taking bets.

Practical stop-rules

Worth pre-committing to:

• Daily stop-loss. A maximum number of units lost in a day before you stop betting that day. 5 units is a common figure. The point is not the number; it is having a number you have decided in advance.
• Tilt stop. If you find yourself wanting to stake more than your unit on the next bet, that is the tilt signal. The rule is “step away for an hour before placing another bet”. Cheap to implement; saves bankrolls.
• Bankroll re-evaluation. If you have lost more than 30 units in a stretch, sit with the spreadsheet. Is the edge still there in CLV terms? Or has the process broken down? Don't conclude one way or the other from a single day.

The boring truth

Bankroll management is easier to read than to live. Almost every losing punter we have spoken to could quote the rules above; very few of them actually held to fixed-unit staking when they were down for the night and one bigger bet would have got them back to even. The discipline only counts when it is uncomfortable.

If a fixed-unit rule feels too restrictive, that is signal — not noise. The restrictiveness is the feature. It is what keeps you betting in a year when free-styling punters have run through their bankroll twice and given up.

Educational content only — not personal financial advice. Sports are uncertain and any bet can lose. Past results do not predict future results. 18+. Gamble responsibly. Responsible gambling resources.