How do you decide your bet size as a percent of your bankroll?

What percent of bankroll should one bet be?

For most bettors, a single bet should be small enough that a losing streak does not damage decision quality. Many disciplined bettors cap one play around 1% to 3% of bankroll, even when a formula flashes something larger.

That cap is not cowardice. It is variance management with a cleaner haircut.

How does Kelly convert edge into bet size?

The input matters more than the calculator. If your edge estimate is inflated because you used a juiced sportsbook price instead of a no-vig baseline, the stake will be too aggressive.

Kelly converts edge into bet size only after the edge input has been cleaned up. First cap any single bet near the written bankroll limit, then ask whether the Kelly output still makes sense at the number currently available. If the answer depends on stale odds, a missing injury assumption, or an emotional read of the matchup, the bet is not ready for sizing.

Why do bettors use quarter Kelly or half Kelly?

Full Kelly maximizes theoretical bankroll growth when your probabilities are accurate. Real bettors have imperfect models, changing markets, and bad Tuesdays, so smaller fractions are usually smarter.

Quarter Kelly and half Kelly reduce drawdowns and make model error less expensive. Sharkie respects the math, then trims the fins.

When should you override the Kelly output?

Override Kelly when the market is thin, the injury news is unstable, the line moved away from your entry, or the bet is correlated with other exposure you already hold.

A hard cap keeps one bad estimate from doing too much damage. Bankroll rules should be written before the bet, not negotiated after the espresso hits.

How should bankroll percentage change with edge and uncertainty?

Bet size should be tied to bankroll, price, and estimated edge, not confidence language. A larger edge can justify a larger stake, but only when the probability estimate is credible. The Kelly framework is useful because it connects those pieces directly: expected advantage relative to payout determines the theoretically optimal fraction of bankroll. In practice, the full Kelly answer is usually too aggressive for sports betting because the edge estimate is rarely exact.

The first step is to use an honest edge input. That means devigging the market before comparing it to a model probability. If a bettor treats the sportsbook's posted line as the fair baseline, the edge can look larger than it is because the sportsbook margin is still embedded in the price. Overstated edge leads directly to overstated bet size.

Fractional Kelly is the safer operating mode. Quarter Kelly or half Kelly reduces variance and protects against model error, stale information, and normal losing streaks. The cost is slower theoretical growth, but the benefit is a much lower chance of severe drawdown. Many disciplined bettors also cap any one position, such as keeping single bets near 1-3% of bankroll even when a formula suggests more.

Flat betting can also be reasonable when the model is early, sample size is limited, or the bettor wants simpler risk control. The important point is consistency. A staking plan should prevent one strong opinion from dominating the bankroll. Bet size is part of the model process: estimate fair probability, compare against the no-vig market, translate the edge into a fraction, then reduce that number to account for uncertainty.