Flat Betting vs Percentage Betting

Use flat betting if you're new, bet casually or don't log every wager, and use percentage betting if you're experienced, track your balance before each bet and have a bankroll large enough for the recalculating to matter. The whole difference is whether your stake follows your bankroll after each result, and at the 1 to 2% stakes most guides recommend, it's smaller than the arguments about it suggest: over 10 straight losses from $1,000 at 2%, percentage betting saves you $17.07.

The figures quoted for this comparison often seem to contradict each other, usually because they mix the two methods. The tables below track both from the same $1,000, using stated assumptions and arithmetic you can check.

All of those results are calculated, with no simulation involved. Each losing bet costs the full stake, each winning bet at -110 pays about $0.91 profit per $1 staked, and percentage stakes are carried unrounded and shown to the cent.

Flat betting means the same dollar amount on every bet. A worked example on DEV Community uses a $1,000 bankroll and a $20 stake, 2% of the starting figure, and that $20 stays put through wins and losses alike.

The best case for it is behavioral. The Spread calls it "simple, disciplined, and resistant to tilt," and that last word counts for more than it seems to. There's no calculation to fudge after a bad night of player props, because the number was settled before the night began. Action Network credits flat betting with keeping bettors disciplined through losing streaks for much the same reason.

Its weakness is that your stake stays fixed while your bankroll moves. A $50 bet on $1,000 is 5% of your money, but after 10 straight losses that same $50 is 10% of a $500 bankroll, the size DeucesCracked warns a single cold run could wipe out. So your risk rises at the moment you can least afford it, and if you never resize, 20 straight losses at $50 leave you with nothing. It also gives up upside, since a growing bankroll keeps placing the same bet.

The Spread's fix is to resize the unit only after the bankroll has moved roughly 25 to 33%. Say you started at $1,000 with a $20 stake. If you fall to $750, a 25% move, 2% becomes $15, and if you climb to $1,250 it becomes $25. Without that review, the $20 stake would sit at about 2.7% of the $750 bankroll and keep climbing.

Flat betting suits anyone who wants one number to remember and will honor a review point.

Percentage betting sets each stake as a fixed share of your current bankroll, so it moves after every result. The Spread's example starts with 1% of $3,000, or $30. After a $300 drawdown, the next bet is $27, while the flat bettor is still staking $30.

SharpAI calls the method "mathematically superior," and there's a serious argument behind that. Wizard of Odds ran a simulation of an even-money bet that wins 51% of the time, and the number of bets needed to double a bankroll depended heavily on the percentage staked:

• 0.5% stake: 7,901 bets
• 1% stake: 4,617 bets
• 2% stake: 3,496 bets
• 3% stake: 4,477 bets

That's strong evidence that percentage stakes have a best size and that overshooting it costs you. As reported, though, it doesn't put flat stakes alongside them, and it assumes a known edge, which a prop bettor doesn't get handed. The claim that the method is superior holds best for a bettor with a real edge and enough bets for compounding to work.

The cost is the admin. SharpAI's own caveat is that you have to recalculate before every wager, which means knowing your exact bankroll every time. That gets awkward quickly if you line shop, because your money is spread across several books. A slate of props that all lock around the same tip-off adds another question: does the stake on bet four count the unsettled money on bets one to three? Pick a rule and keep it. The simplest is to size the whole slate from your total balance before the first bet, pending stakes included.

Percentage betting suits a bettor who already keeps a complete record, and a prop tracker can hold the bet log alongside the balance you maintain.

This table also clears up a disagreement between two guides. Bookmaker100 says 10 losses at 5% leave you with 50% of your bankroll, while DeucesCracked says the same streak removes about 40%. Both are right about different methods: the 50% figure is flat math, and the 40% figure matches percentage math, where $401.26 is gone. At Action Network's 3% middle ground, the same streak ends at $700.00 flat and $737.42 by percentage.

The streak itself is no freak event. DeucesCracked says a bettor winning 55% of standard -110 bets will still periodically lose 10 or more in a row, and that such a run doesn't necessarily mean the edge has gone. How you behave during that cold stretch is a separate problem, but the stake is what decides how much money you have left to behave with.

Here's the mirror image, with 10 straight wins at -110.

This is compounding at work. The flat $50 bettor makes $45.45 on every win. The 5% bettor makes $45.45 on the first and $67.81 on the tenth, finishing $105.19 ahead. At 2% the lead is $15.62, and at 1% it's $3.81.

Combine the two streaks and the protection turns out to have a price. After 10 losses and 10 wins at -110, in either order, the percentage bettor is behind at every size:

• 1%: $990 by percentage, $991 flat
• 2%: $978 by percentage, $982 flat
• 5%: $934 by percentage, $955 flat

The arithmetic is simple. Each loss and each win changes the bankroll by a percentage, and a percentage lost takes a bigger percentage gain to recover. That makes percentage betting a trade rather than a free upgrade: it gives you shallower drawdowns and faster compounding on sustained runs, and a slower climb out of a hole.

Stake size matters far more than method. Across 10 straight losses, moving from a $20 flat stake to a $50 flat stake costs you an extra $300. Switching method saves $17.07 at 2% and $98.74 at 5%. A bettor agonizing over flat versus percentage while betting 5% has the decision backward.

The second difference is risk drift, and it's the real mark against flat betting. A flat stake grows as a share of a shrinking bankroll, and that only stays harmless if someone resizes it. The Spread's 25 to 33% trigger does that in occasional steps, which leaves flat betting behaving like a slower version of percentage betting. A flat bettor who won't resize has the worst version of the method.

The third is whether you'll do the work. Percentage betting only protects you if every stake comes from an accurate balance, and that depends on your record-keeping across every book you use. Matter of Stats notes that advisors favor level staking because bettors generally can't judge when one bet's edge is above or below average. Its own conservative practice applies fractional Kelly to the original bankroll instead of the current one, which removes the stake swings that compounding would otherwise cause. So even a site that uses edge-based sizing has chosen not to recalculate against every result.

The commonly cited range runs from 1 to 5% of bankroll, but the sources lean lower than the top of that range suggests:

• Action Network: 1 to 5%, with 3% as a middle ground.
• TopEndSports: 1 to 2% conservative, 2 to 3% standard, 4 to 5% aggressive, and never above 5%.
• DeucesCracked: 1 to 2%, with 5% used only as a stress test.
• The Spread: 0.5 to 2%, with 1% a popular starting point.

None of these sets a figure specifically for player props. Bookmaker100 comes closest, assigning 1% to low-confidence, speculative or unfamiliar-market bets. Props with thin histories fit that description well enough to justify staying at 1 to 2%, and the loss table shows what 5% does during a bad run.

Flat bettors read this table once at the start and again at each resize. Percentage bettors are effectively reading it before every bet. The calculator below works out flat-unit and fractional-Kelly stakes from your own bankroll, along with the losing streak to expect over a season.

The Kelly Criterion keeps the idea of staking against your current bankroll but drops the fixed percentage. The Spread puts the principle neatly: "Bet more when your edge is bigger, less when it's smaller." Wizard of Odds says that sizing this way maximizes expected bankroll growth over the long run. If you have no edge, the formula says not to bet.

The weak spot is the 55%. Break-even at -110 is about 52.4%, so your whole edge is a small gap that you've estimated yourself. Suppose your true rate is 53%. Now the correct full Kelly stake is about 1.3%, and the 5.5% you just calculated is more than four times that. Wizard of Odds shows why this matters: double Kelly produces zero expected growth, and anything above it means expected decline. A small estimating error can turn a bettor with a real edge into one whose stake sizing shrinks the bankroll. Even half Kelly on the 55% estimate, at 2.75%, is over double the correct stake in that scenario. Quarter Kelly, at 1.375%, is about right.

This is why The Spread calls full Kelly "aggressive and error-sensitive," and why most practitioners scale it down. Wizard of Odds puts the trade at half the volatility for about a quarter less growth, and Matter of Stats warns that full Kelly can call for one-fifth to one-third of a bankroll on a single bet. Kelly also doesn't find the edge for you: DeucesCracked stresses that it's a staking method only, and it needs well-tested probabilities behind it.

One correction if you're comparing sources: SharpAI gives a 5% Kelly stake for a 55% bet at +100. It reaches that by treating the 5-point gap between true and implied probability as the edge, but the formula above gives 10% at those odds (1 × 0.55 − 0.45 = 0.10). The edge calculator will run the sum at your own price and projection, and there's more on sizing with fractional Kelly if you've reached that stage.

New or casual bettors should use flat betting. Set the stake at 1 to 2% of your starting bankroll, write it down, and resize only when the bankroll has moved 25%. On $1,000, that means $10 or $20 a bet until you reach $750 or $1,250. Flat betting at 1% gives up $4.38 of protection over 10 straight losses, and that's a fair price for having nothing to recalculate.

Experienced bettors with a larger bankroll and a complete bet log should use percentage betting at the same 1 to 2%. It takes the risk drift out automatically and lets a growing bankroll compound. If you can't state your exact balance across all your books before you bet, you don't yet meet the condition.

Kelly belongs to bettors whose logged results show their probability estimates hold up. Even then, use a fraction of it, and quarter Kelly is the sensible place to begin. For everyone else, The Spread's guidance applies: without robust edge estimates, flat betting or a flat 1% per wager is safer than full Kelly.

Whichever you pick, stay away from 5% as an everyday stake, because that decision outweighs this one. The step-by-step setup covers how to put either method in place.