Kelly Criterion for Sports Bettors: The Optimal Bet Sizing Guide

The Origin Story: From Bell Labs to the Betting Window

In 1956, a physicist named John Larry Kelly Jr. was working at AT&T's Bell Labs on a seemingly unrelated problem: how to maximize the signal-to-noise ratio in long-distance telephone lines. His insight was that information theory -- the mathematical framework Claude Shannon had developed to describe communication channels -- could also describe the optimal strategy for a gambler with an edge.

Kelly published a paper titled "A New Interpretation of Information Rate" in the Bell System Technical Journal. The core idea was deceptively simple: if you have an edge in a repeated bet, there is a mathematically optimal fraction of your bankroll to wager on each opportunity. Bet too little and you leave money on the table. Bet too much and you risk ruin, even with a genuine edge.

The formula Kelly derived became known as the Kelly Criterion, and it went on to influence everyone from blackjack card counters like Ed Thorp (who used it to beat casinos in the 1960s) to hedge fund managers like Warren Buffett and Bill Gross. Today, it is one of the most important concepts in sports betting bankroll management.

The Kelly Formula Explained

For a simple two-outcome bet (win or lose, no pushes), the Kelly Criterion says you should wager the following fraction of your bankroll:

f = (bp - q) / b*

Where:

• f* = The optimal fraction of your bankroll to wager
• b = The net odds received on the bet (decimal odds minus 1, or the profit per dollar wagered)
• p = Your estimated probability of winning
• q = The probability of losing (q = 1 - p)

If f* is positive, the bet has positive expected value and you should wager that fraction. If f* is zero or negative, the bet is not worth making.

Converting from American Odds

Since many bettors work with American odds, here is how to get b:

• For positive American odds (e.g., +150): b = odds / 100 = 1.50
• For negative American odds (e.g., -200): b = 100 / |odds| = 0.50

Detailed Worked Example

Let's apply the Kelly Criterion to a real betting scenario.

The Situation:

You have analyzed an NBA game and believe the Milwaukee Bucks have a 58% chance of covering the spread. The sportsbook is offering -110 on the Bucks spread.

Step 1: Identify the variables.

• p = 0.58 (your estimated win probability)
• q = 0.42 (1 - 0.58)
• American odds = -110, so b = 100/110 = 0.9091

Step 2: Apply the Kelly formula.

f* = (bp - q) / b f* = (0.9091 x 0.58 - 0.42) / 0.9091 f* = (0.5273 - 0.42) / 0.9091 f* = 0.1073 / 0.9091 f = 0.118, or 11.8% of your bankroll*

Step 3: Calculate the actual bet size.

If your bankroll is $5,000:

Bet size = $5,000 x 0.118 = $590

According to Kelly, you should wager $590 on this bet to maximize your long-term bankroll growth rate.

Verifying the Logic

Let's check that this makes intuitive sense. You believe you have a 58% chance of winning at -110 odds (which implies a break-even point of 52.38%). Your edge is approximately 5.62 percentage points above break-even, which is a meaningful edge. Kelly is telling you to bet about 12% of your bankroll, which reflects the confidence level -- a real but not enormous edge.

If your estimated probability were lower -- say 54% -- the Kelly fraction would shrink:

f* = (0.9091 x 0.54 - 0.46) / 0.9091 = (0.4909 - 0.46) / 0.9091 = 0.034, or 3.4%

With a smaller edge, Kelly recommends a proportionally smaller bet. This self-regulating property is one of Kelly's greatest strengths.

Full Kelly vs. Fractional Kelly

Here is where theory meets reality. Full Kelly is mathematically optimal for maximizing the geometric growth rate of your bankroll, but it comes with extreme volatility. The variance of full Kelly staking is brutal -- simulations show that a full Kelly bettor will regularly experience drawdowns of 50% or more, even with a genuine edge.

This leads to a critical insight: overbetting is far worse than underbetting.

Why Overbetting Destroys Bankrolls

The Kelly Criterion sits at the exact peak of the growth rate curve. On either side of this peak, your bankroll grows more slowly. But the curve is not symmetric:

• Underbetting (wagering less than Kelly) reduces your growth rate, but you still grow. At half Kelly, you achieve 75% of the maximum growth rate with significantly less variance.
• Overbetting (wagering more than Kelly) also reduces your growth rate, but the consequences are far worse. At double Kelly, your expected growth rate drops to zero -- you are statistically treading water despite having a real edge. Above double Kelly, your expected growth rate turns negative, meaning you will eventually go broke with certainty, even though every individual bet has positive expected value.

This asymmetry is why the vast majority of professional bettors use fractional Kelly -- wagering some fraction (typically 25-50%) of the full Kelly recommendation.

Fractional Kelly in Practice

The most common approach is half Kelly (also called Kelly/2), where you wager half of what the full Kelly formula recommends.

Using our earlier example:

• Full Kelly: 11.8% of bankroll = $590
• Half Kelly: 5.9% of bankroll = $295
• Quarter Kelly: 2.95% of bankroll = $147.50

Benefits of fractional Kelly:

1. Dramatically reduced variance. Half Kelly cuts the standard deviation of outcomes by roughly 50%, making the ride much smoother.
2. Protection against estimation errors. If your probability estimate is slightly off (and it usually is), fractional Kelly provides a safety buffer. Full Kelly with a wrong probability can easily become overbetting.
3. Psychological sustainability. Most bettors cannot emotionally handle the swings of full Kelly. If a 50% drawdown causes you to abandon the strategy, it does not matter that full Kelly is theoretically optimal.
4. Still captures most of the growth. Half Kelly delivers 75% of the theoretical maximum growth rate. Quarter Kelly still captures about 56%. The marginal benefit of going from half to full Kelly is small compared to the risk.

A Rule of Thumb

If you are not a quantitative professional with a highly accurate probability model, quarter Kelly or less is almost certainly the right choice. Many professional sports bettors operate at 1-3% of bankroll per bet regardless of edge size, which implicitly functions as a very conservative fractional Kelly approach.

Practical Application for Sports Bettors

Estimating Your Edge

The Kelly formula requires you to input p, your true probability estimate. This is simultaneously the most important and the most difficult input to get right. Every error in your probability estimate flows directly into your bet sizing.

Sources for probability estimates include:

• Your own model built from statistical analysis
• Sharp market consensus (removing vig from Pinnacle or other sharp books)
• Closing line comparison (if you consistently beat the close by X%, you can estimate your edge)
• Situational analysis combining quantitative and qualitative factors

Be conservative. If you think a team has a 55% chance, consider using 53% in your Kelly calculation. The cost of slight underbetting is trivial compared to the cost of overbetting.

Handling Simultaneous Bets

One complication in sports betting is that you often have multiple bets open at the same time. The basic Kelly formula assumes sequential bets with no overlap. When you have concurrent bets, several approaches exist:

• Reduce Kelly fraction further (e.g., from half to quarter) when you have many open bets
• Use a total exposure limit (e.g., never have more than 20% of bankroll at risk at any time)
• Apply the simultaneous Kelly formula, which accounts for correlations between bets (though this is computationally complex)

For most bettors, the simplest approach is to use a conservative fractional Kelly (quarter or less) and monitor your total exposure.

Adjusting for Odds Format

The basic Kelly formula works directly with decimal odds (subtract 1 to get b). For American odds, always convert first. For fractional odds (e.g., 5/2), b is simply the fraction as a decimal (2.5).

When Kelly Says "Don't Bet"

If the Kelly formula returns zero or a negative number, it is telling you the bet has no edge or negative expected value. Listen to it. One of Kelly's most valuable functions is not just telling you how much to bet, but telling you when not to bet at all.

Many bettors struggle with this because they want action. They will talk themselves into a probability estimate that justifies a bet, rather than honestly estimating probability and accepting the Kelly verdict. Discipline here separates long-term winners from everyone else.

Kelly Criterion for Parlays and Other Bet Types

The Kelly formula extends to parlays and other multi-outcome bets, though the calculation becomes more involved. For a parlay, you need to estimate the joint probability of all legs hitting (accounting for any correlation between them) and use the combined payout odds.

For a two-leg parlay:

• p = P(leg 1 wins) x P(leg 2 wins) -- assuming independence
• b = combined decimal odds - 1

Since parlays amplify both the payout and the probability of losing, Kelly will typically recommend very small bet sizes for parlays, which aligns with common sense: most of your action should be on single bets where your edge estimate is most reliable.

Common Kelly Mistakes

Using the Sportsbook's Implied Probability as "p"

If you use the book's own implied probability as your win estimate, Kelly will always return zero or negative (because the vig ensures the odds are worse than fair). You must have an independent edge estimate to use Kelly.

Ignoring the Vig in Your Edge Calculation

Your edge exists relative to the true probability, not relative to a no-vig line. Make sure you are accounting for the juice when estimating whether you have an edge and how large it is.

Applying Full Kelly with Uncertain Estimates

Full Kelly assumes your probability estimate is perfectly accurate. In sports betting, it never is. Always use a fractional Kelly approach unless you have extraordinary confidence in your model.

Not Adjusting for Bankroll Changes

Kelly is a proportional system -- you bet a percentage of your current bankroll, not your starting bankroll. As your bankroll grows, your bets grow. As it shrinks, your bets shrink. This automatic scaling is a feature, not a bug: it makes Kelly naturally resistant to ruin.

How HedgeSlider's Kelly Calculator Helps

Applying the Kelly Criterion correctly requires converting odds formats, running the formula, and then scaling to your chosen fraction. HedgeSlider's Kelly Calculator handles all of this:

• Enter odds in any format -- American, decimal, or fractional -- and the calculator handles conversion
• Input your estimated probability and see the exact Kelly fraction and recommended bet size
• Adjust your Kelly fraction (full, half, quarter, or custom) with a slider to see how it affects your recommended stake
• Enter your bankroll and get a dollar amount you should wager, not just a percentage
• Instantly see the EV of the bet alongside the Kelly recommendation, so you can confirm the bet is actually +EV before sizing it

The calculator takes the mental arithmetic out of the process, letting you focus on the judgment calls -- estimating probabilities and choosing your risk tolerance -- rather than grinding through formulas.

Building Your Kelly Strategy

Here is a practical framework for incorporating Kelly into your sports betting:

1. Set your bankroll. This is the total amount dedicated to betting. Do not include rent money, savings, or any funds you cannot afford to lose.
2. Choose your Kelly fraction. Start with quarter Kelly. You can always increase later as you gain confidence in your estimates.
3. Develop probability estimates. Use models, sharp lines, or a combination. Be conservative.
4. Run the Kelly formula for each bet using a calculator. If Kelly says do not bet, do not bet.
5. Track everything. Record your estimated probability, the odds, the Kelly recommendation, your actual bet size, and the result.
6. Review monthly. Compare your estimated probabilities to actual outcomes. If your estimates are consistently too high, you are overbetting. If they are consistently too low, you may be leaving growth on the table.

The Kelly Criterion is not a magic formula that guarantees profits. It is a bankroll management system that, when combined with a genuine edge and accurate probability estimates, maximizes your long-term growth while protecting you from ruin. It requires patience, discipline, and honesty about the limits of your knowledge. Used correctly, it is one of the most powerful tools in a sports bettor's arsenal.