How much of your bankroll should you bet on each play?
Size bets as a fixed percentage of bankroll, usually 1-2% as one unit, and cap even your best plays at 1 unit. On +EV plays around +250, ten-bet losing streaks are statistically routine, so survival math beats conviction math. The edge only pays if the bankroll lives long enough to collect it.
Most bettors who find an edge still go broke, and it’s almost never the edge that fails. It’s the sizing. Here’s the boring system that keeps the math alive long enough to pay, and the streak arithmetic that justifies it.
Units: the whole system in three sentences
Pick a bankroll number you can truly afford to swing. One unit is 1-2% of it. Every play gets sized in units, flat, and your best play of the week gets the same cap as your Tuesday afternoon one.
That’s it. A $2,500 bankroll at 1u = $50 (2%) means a normal play risks $50 and nothing risks more than $50. Note the two moving parts: the unit itself is a fixed dollar amount, while individual plays get fractions of it. The desk runs exactly this shape: recommended sizes come out between 0.3u and 1.0u scaled by the measured edge, with 1u as a hard cap no play crosses, ever. You’ll see it on every card as “Rec” next to a cap line.
The streak math that makes flat sizing non-negotiable
+EV betting at boosted prices means most plays are underdogs on purpose. A +250 play with a real edge still loses about 70% of the time. Run the numbers on what that does in sequence:
At a 30% hit rate, the chance any specific 10 plays all lose is 0.7^10, about 2.8%. Sounds rare. But across 300 plays you get dozens of overlapping 10-bet windows, and the chance you hit at least one 10-loss streak somewhere is near certainty. Twelve and fifteen-loss streaks are not exotic either. This is what a healthy, profitable, correctly-priced betting operation looks like from inside a bad month.
Now size that streak. At 2% flat, ten straight losses costs 20% of bankroll: painful, survivable, recoverable. At 10% flat, “I’m sure about these” sizing, the same routine streak is your entire bankroll. Same edge, same bets, same luck. Only the sizing differed.
Why not just use Kelly?
The Kelly criterion is the mathematically optimal growth formula, and worth knowing. It also assumes you know your true edge precisely. You don’t: fair prices are estimates, devigged from markets that move. Overestimate your edge with full Kelly and you systematically overbet, which is worse than underbetting. That’s why practitioners use fractional Kelly, and why a flat 1-2% with an edge-scaled cap is the version of it that survives contact with reality. The desk’s sizing is exactly that philosophy in code.
The quiet benefit: your record means something
Flat sizing makes a ledger honest. If stakes swing with mood, a record can be green on dollars while terrible on decisions, or vice versa, and nobody can audit which. Fixed units make every play comparable, which is why the public ledger can show flat-stake results that mean what they appear to mean. It’s also why “how much did you bet” is the first question to ask anyone showing you a winning screenshot.
The free Discord posts one +EV play a day with the recommended size in units on the card, math shown: join free. Bring a bankroll number you can defend, and let the sizing be the most boring part of your betting life. Boring is what compounding feels like from the inside.
Quick answers
What is a unit in sports betting?
A fixed slice of your bankroll used as the standard bet size, typically 1-2%. A $2,000 bankroll at 1u = $25 means every play risks $25 regardless of confidence. Units make records comparable and remove the nightly temptation to size by feeling.
Should I bet more on plays I'm confident in?
Confidence should come from measured edge, and even then only slightly. The desk expresses it as fractions between 0.3u and a hard 1u cap, computed from the edge, never vibes. The moment sizing becomes emotional, one bad night can erase a month of correct decisions.
What is the Kelly criterion and should I use it?
Kelly computes the bankroll fraction that maximizes long-term growth given your edge and odds. Full Kelly assumes your edge estimate is perfect, which it never is, so practitioners bet a fraction of it. A flat 1-2% with an edge-scaled cap gets you most of the benefit with far less of the blowup risk.
How big a losing streak should I plan for?
Longer than feels possible. At plus-money prices where each play hits around 30%, the chance of some 10-bet losing streak across a few hundred bets is close to certain. Plan sizing so that streak is an annoyance, not an ending.
Wanna see which boosts cleared the bar today?
The bot checks every boost at every book each morning, posts the ones worth taking with the math shown, and grades everything in public. Watching is free.