Kelly Criterion: The Math Behind Smart Bankroll Management

What Is the Kelly Criterion?

The Kelly Criterion is a mathematical formula developed by John Kelly Jr. at Bell Labs in 1956. Originally designed for information theory and signal processing, it was quickly adopted by gamblers and investors as a method for determining optimal bet sizes.

The core idea is straightforward: if you have an edge, the Kelly Criterion tells you exactly what percentage of your bankroll to wager to maximize long-term growth. Bet too little, and you leave money on the table. Bet too much, and you risk catastrophic drawdowns. Kelly finds the sweet spot.

The formula has been used by some of the most successful bettors and investors in history, including Edward Thorp (who used it to beat blackjack and the stock market) and Bill Benter (who built a horse racing model that generated over a billion dollars in profit).

The Formula Explained Simply

The Kelly Criterion formula is:

f = (bp - q) / b

Where:

• f = the fraction of your bankroll to wager
• b = the decimal odds minus 1 (or the net profit per dollar wagered)
• p = the probability of winning
• q = the probability of losing (which is 1 - p)

Let us work through an example. Suppose you find a bet at +150 odds (decimal odds of 2.50) and you estimate you have a 45% chance of winning.

• b = 2.50 - 1 = 1.50
• p = 0.45
• q = 0.55

f = (1.50 x 0.45 - 0.55) / 1.50

f = (0.675 - 0.55) / 1.50

f = 0.125 / 1.50

f = 0.0833

Kelly says to bet 8.33% of your bankroll on this wager.

For a standard -110 bet where you estimate a 55% win probability:

• b = 1.909 - 1 = 0.909
• p = 0.55
• q = 0.45

f = (0.909 x 0.55 - 0.45) / 0.909

f = (0.50 - 0.45) / 0.909

f = 0.05 / 0.909

f = 0.055

Kelly recommends 5.5% of your bankroll.

Notice something important: when your edge is small (55% at -110), Kelly recommends a modest bet size. When your edge is zero or negative, the formula returns zero or a negative number, telling you not to bet at all.

Full Kelly vs. Fractional Kelly

Here is where theory meets reality. Full Kelly is mathematically optimal for maximizing long-term growth, but it comes with severe drawdowns that most bettors cannot stomach.

Running full Kelly simulations, you will regularly experience drawdowns of 40-60% of your bankroll. Even with a legitimate edge, you will see stretches where your bankroll drops by half. Most people panic, abandon their strategy, or go broke before the math has a chance to work.

This is why virtually every professional bettor uses fractional Kelly, typically between one-quarter Kelly and one-half Kelly.

Quarter Kelly (1/4 Kelly): Take the Kelly recommendation and divide by 4. If Kelly says 8%, you bet 2%. This is the most conservative approach and is recommended for bettors who are uncertain about the accuracy of their edge estimates. Half Kelly (1/2 Kelly): Take the Kelly recommendation and divide by 2. If Kelly says 8%, you bet 4%. This offers a good balance between growth and drawdown management. It is the most common choice among professional bettors.

The reduction in growth rate from using fractional Kelly is surprisingly small. Half Kelly achieves approximately 75% of the growth rate of full Kelly, while dramatically reducing variance and maximum drawdown. Quarter Kelly achieves about 50% of full Kelly's growth rate with even smoother performance.

Why Most Bettors Go Broke

The number one reason bettors go broke is overbetting -- risking too large a percentage of their bankroll on individual wagers. This is not an opinion; it is a mathematical certainty.

Consider a bettor with a genuine 55% win rate at -110 odds. Full Kelly would suggest about 5.5% per bet. But many recreational bettors routinely risk 10%, 15%, or even 25% of their bankroll on a single game.

At 25% per bet with a 55% win rate, the probability of going broke before doubling your bankroll is staggeringly high. Even with a real edge, betting this aggressively virtually guarantees eventual ruin during an inevitable losing streak.

Here is the math that makes this clear. A bettor risking 25% per bet who loses four in a row sees their bankroll drop to:

$1,000 x 0.75 x 0.75 x 0.75 x 0.75 = $316

They have lost 68% of their bankroll. Four consecutive losses at -110 is not unusual. At a 55% win rate, you will experience a 4-game losing streak approximately every 30-40 bets.

The same four-loss streak at 3% per bet:

$1,000 x 0.97 x 0.97 x 0.97 x 0.97 = $885

A manageable 11.5% drawdown. The bettor barely notices. This is the power of proper position sizing.

Example Calculations with Realistic Scenarios

Let us run through several realistic scenarios using half Kelly:

Scenario 1: Strong Edge on an Underdog

• Bet: Underdog at +180
• Your estimated win probability: 42%
• b = 1.80, p = 0.42, q = 0.58
• Full Kelly: (1.80 x 0.42 - 0.58) / 1.80 = (0.756 - 0.58) / 1.80 = 9.78%
• Half Kelly: 4.89%
• On a $2,000 bankroll: $97.80 wager

Scenario 2: Moderate Edge on a Spread

• Bet: Team -4.5 at -110
• Your estimated win probability: 55%
• b = 0.909, p = 0.55, q = 0.45
• Full Kelly: (0.909 x 0.55 - 0.45) / 0.909 = 5.5%
• Half Kelly: 2.75%
• On a $2,000 bankroll: $55.00 wager

Scenario 3: Small Edge

• Bet: Over 217.5 at -108
• Your estimated win probability: 54%
• b = 0.926, p = 0.54, q = 0.46
• Full Kelly: (0.926 x 0.54 - 0.46) / 0.926 = 4.0%
• Half Kelly: 2.0%
• On a $2,000 bankroll: $40.00 wager

Notice the pattern: as the edge decreases, so does the recommended bet size. This is exactly right. You should bet more when your edge is larger and less when it is smaller. This is the opposite of what most recreational bettors do (they bet the same amount on every game regardless of confidence).

When NOT to Use Full Kelly

Full Kelly should never be used in the following situations:

When your edge estimate is uncertain. The Kelly formula is extremely sensitive to the accuracy of your probability estimate. If you think you have a 58% chance but the true probability is 52%, full Kelly will have you dramatically overbetting. Since we can never know our true edge with certainty, fractional Kelly provides a margin of safety. When you are betting correlated events. If you bet three NBA games on the same night and all three could be affected by the same factor (league-wide referee trend, schedule quirk), your bets are correlated. Kelly assumes independent bets. When your bankroll is small relative to bet minimums. If Kelly says to bet 2% of a $500 bankroll ($10), but the sportsbook minimum is $10, you are effectively forced into full Kelly territory. Consider a smaller number of bets or building your bankroll before applying Kelly sizing. When you are emotionally unable to handle the swings. The best strategy is the one you can actually follow. If half Kelly drawdowns cause you to abandon your strategy, use quarter Kelly. Consistent execution of a slightly suboptimal strategy beats inconsistent execution of the optimal one.

Practical Bankroll Management Rules

Beyond Kelly, here are rules that every serious bettor should follow:

Never bet more than 5% of your bankroll on a single bet. Even if Kelly suggests more, cap yourself at 5%. This protects against the inevitable errors in your probability estimates. For most bettors, 1-3% per bet is ideal. This range allows for meaningful growth while keeping drawdowns manageable. At 2% per bet, even a brutal 10-game losing streak only costs you about 18% of your bankroll. Track every bet in a spreadsheet or tracking app. Record the date, sport, league, bet type, odds, your estimated probability, Kelly recommendation, actual stake, and result. Review this data monthly. Maintain a dedicated bankroll. Your betting bankroll should be completely separate from your living expenses. Never add to it from funds you need, and do not withdraw from it until you have established a clear, documented edge over hundreds of bets. Reassess your bankroll monthly. If your bankroll has grown, your bet sizes should grow proportionally. If it has shrunk, bet sizes should shrink. This is automatic with Kelly-based staking, but it needs to be explicit if you use flat staking. Set a stop-loss for the day or week. If you lose 10% of your bankroll in a single day, stop betting for the rest of that day. This prevents tilt-driven decisions from compounding a bad day into a catastrophic one.

How PIPER Helps with Bankroll Management

Every pick PIPER issues includes a recommended stake size based on fractional Kelly calculations. We estimate the edge on each play using our 5-path confluence system, then apply quarter-to-half Kelly to determine the appropriate bet size.

Our dashboard tracks your running bankroll, shows cumulative profit and loss, and alerts you if your betting patterns deviate from Kelly-optimal sizing. If you start overbetting, you will see a warning.

We also provide drawdown analysis, showing you the worst-case historical drawdowns for your betting style and stake sizes. This helps set realistic expectations: if you know that a 15% drawdown is normal and expected, you are far less likely to panic and abandon your strategy when it happens.

Smart bankroll management is not glamorous. It does not produce exciting stories about "putting it all on the line." But it is the single most important factor separating bettors who survive long enough to realize their edge from those who go broke before they get the chance.