March 30, 2026 | 9 min read
Table of Contents
What Is the Kelly Criterion? The Kelly Formula Example: NFL Moneyline Example: NBA Spread The Problem with Full Kelly Why Quarter Kelly Is the Move Kelly vs. Flat Betting Common Bet Sizing Mistakes How SharpEdge Calculates It for You FAQMost bettors spend all their time finding good bets and almost no time figuring out how much to bet. That is backwards. A great pick with terrible sizing will blow up your bankroll faster than a mediocre pick with smart sizing.
The Kelly Criterion solves this problem. Developed by John Kelly at Bell Labs in 1956, it is a formula that tells you the mathematically optimal percentage of your bankroll to wager based on your edge and the odds. It was originally designed for information theory, but professional gamblers and hedge fund managers adopted it because it maximizes long-term bankroll growth while minimizing the risk of ruin.
The concept is simple: bet more when your edge is bigger, bet less when your edge is smaller, and never bet when you have no edge at all. The formula handles the exact math so you do not have to guess.
Here is the formula in its simplest form:
Where:
f* = the fraction of your bankroll to bet
b = the decimal odds minus 1 (net profit per dollar wagered)
p = the probability of winning
q = the probability of losing (1 - p)
If the result is zero or negative, Kelly says do not bet. There is no edge. If the result is positive, it tells you what percentage of your bankroll to put down. The bigger your edge, the bigger the recommended stake.
Scenario: You find the Eagles moneyline at +150 on DraftKings.
SharpEdge's scanner shows the true probability (after de-vigging Pinnacle) is 42%. The sportsbook is pricing them as if they have a 40% chance. That gap is your edge.
Let's plug it in:
b = 2.50 - 1 = 1.50 (the +150 in decimal form minus 1)
p = 0.42 (true win probability)
q = 0.58 (1 - 0.42)
f* = (1.50 x 0.42 - 0.58) / 1.50
f* = (0.63 - 0.58) / 1.50
f* = 0.05 / 1.50 = 0.0333 = 3.33%
Kelly says bet 3.33% of your bankroll. On a $5,000 bankroll, that is $167.
Scenario: Celtics -4.5 at -108 on FanDuel.
After de-vigging, the true probability of covering is 54%. FanDuel's -108 implies 51.9%. You have roughly a 2% edge.
Plug it in:
b = 1.926 - 1 = 0.926
p = 0.54
q = 0.46
f* = (0.926 x 0.54 - 0.46) / 0.926
f* = (0.50 - 0.46) / 0.926
f* = 0.04 / 0.926 = 0.0432 = 4.32%
Kelly says 4.32% of your bankroll. On $5,000, that is $216.
Notice how the two examples give different bet sizes even though both are +EV. That is the entire point. Kelly scales your bet to match your edge. Bigger edge, bigger bet. Smaller edge, smaller bet.
Full Kelly maximizes long-term growth on paper. In practice, it is extremely aggressive. Here is why that matters:
Variance is brutal. Full Kelly can produce 50% drawdowns. If you start with $5,000, you might see your bankroll drop to $2,500 before it recovers. Most people cannot stomach that, and they quit right before the math starts paying off.
True probability is an estimate. Kelly assumes you know the exact probability of winning. You do not. You have a very good estimate based on sharp lines, but it is still an estimate. If your probability is off by even 1-2%, full Kelly overexposes you.
Correlated bets. If you are betting multiple games in the same night (which you should be, because volume matters), full Kelly on each game does not account for correlation between bets. Your actual exposure can exceed what Kelly intended.
The solution used by almost every professional bettor and fund manager: fractional Kelly. Specifically, quarter Kelly (25% of the full Kelly recommendation).
Full Kelly on the Eagles example: $167
Half Kelly: $83
Quarter Kelly: $42
Why quarter Kelly specifically?
1. It reduces max drawdown by roughly 75%. Instead of a 50% drawdown being normal, your worst stretches might be 12-15%. That is survivable.
2. It only sacrifices about 6% of long-term growth. You give up a small amount of theoretical maximum growth in exchange for massively less volatility. The risk-adjusted return is actually better.
3. It protects against estimation error. If your probability estimate is slightly wrong, quarter Kelly keeps you safe. Full Kelly with a wrong estimate can be catastrophic.
4. It lets you bet more games. With smaller individual stakes, you can comfortably bet 10-20 edges per day without overexposing your bankroll. More bets means faster convergence to your expected profit.
Think of it this way: full Kelly is like driving 200 mph on the highway. Theoretically fastest. But one pothole and you are done. Quarter Kelly is driving 150 mph with suspension that handles the bumps. You still get where you are going fast. You just survive the trip.
Some bettors use flat staking, which means betting the same dollar amount on every play regardless of edge size. Typically 1-2% of bankroll per bet.
Flat betting is simple and it works. But it has a fundamental flaw: it treats a 2% edge the same as a 10% edge. That is like paying the same price for a Honda and a Ferrari. Kelly fixes this by telling you to bet more on stronger edges and less on weaker ones.
Over 1,000 bets, quarter Kelly typically produces 30-50% more profit than flat betting because it concentrates capital where the edge is biggest. The tradeoff is slightly more complexity. But if your tool calculates it automatically, there is no tradeoff at all.
Already covered above. It works, but it leaves money on the table. Use Kelly to scale with your edge.
"I really like this pick, so I'll put $500 on it." This is how bankrolls die. Your confidence level is not a mathematical input. The odds and the true probability are. Let the formula decide.
Kelly naturally prevents this. After a loss, your bankroll is smaller, so Kelly automatically reduces your next bet. This is built-in bankroll protection. Flat bettors who chase losses are doing the opposite of what math says to do.
If you start with $5,000 and grow to $7,000, your bet sizes should increase proportionally. If you drop to $3,500, they should decrease. Kelly handles this automatically because it works in percentages, not fixed dollars.
Every alert SharpEdge sends includes a Kelly-sized bet recommendation calibrated to your bankroll. Here is what happens under the hood:
1. The scanner pulls live odds from major sportsbooks.
2. It identifies the sharpest line (usually Pinnacle) and de-vigs it to find the true probability.
3. It compares every book's odds against that true line.
4. When it finds a +EV edge, it runs the Kelly formula using the true probability and the offered odds.
5. It applies a fractional Kelly multiplier (quarter Kelly by default) for safety.
6. It sends you the alert with the exact dollar amount to bet based on your set bankroll.
You do not need to run the formula yourself. You do not need a spreadsheet. You get a Telegram message that says exactly what to bet, where to bet it, and how much to stake. The math is done.
The free tier gives you 1 edge per day with full Kelly sizing. Pro plans unlock all edges, AI analysis, and confidence grades.
Try Free on TelegramThat means the edge is massive, which is rare. Even then, quarter Kelly would bring it down to 5%, which is reasonable. If you are seeing extremely high Kelly percentages, double-check the odds. The line may have already moved.
Quarter Kelly for most people. If you have a large sample of verified edges and high confidence in your probability estimates, you can move to third or half Kelly. Full Kelly is almost never recommended in practice.
Kelly can be applied to parlays, but the calculation gets more complex because you need the true probability of all legs hitting. In general, Kelly will recommend very small parlay stakes because the win probability is low. This is mathematically correct. Parlays are high-variance by nature.
Your bankroll is the total amount you have allocated to sports betting and can afford to lose. It is not your rent money. It is not your savings. Set a number you are comfortable with, enter it into SharpEdge, and the Kelly formula handles the rest.