Kelly Bet Sizing Calculator

The Origin: John Kelly at Bell Labs

The Kelly Criterion was first published in 1956 by John L. Kelly Jr., a researcher at Bell Labs in Murray Hill, New Jersey. Kelly was not working on gambling — he was solving a problem in information theory related to signal noise on long-distance telephone lines. His key insight was that a gambler receiving noisy tips over a phone line should size their bets in proportion to the reliability of the information, and he derived a formula for the wager size that maximizes the long-term geometric growth rate of wealth.

The paper, titled “A New Interpretation of Information Rate,” appeared in the Bell System Technical Journal and initially attracted more attention from mathematicians than from gamblers. That changed when Edward O. Thorp — the MIT professor famous for card-counting in blackjack — recognized the formula's power and applied it both at casino tables and on Wall Street. Thorp went on to run one of the most successful quantitative hedge funds in history, and his use of Kelly-based position sizing was central to his approach.

The Mathematical Foundation

At its heart, the Kelly Criterion maximizes the expected logarithm of wealth. This is different from maximizing expected wealth itself (which would suggest betting everything on any +EV proposition). By maximizing the log of wealth, the Kelly strategy has two remarkable properties:

• It never goes broke. Because you are always wagering a fraction of your bankroll, you theoretically can never reach zero (though in practice, a small bankroll becomes impractical).
• It outgrows any other strategy. In the long run, a Kelly bettor's bankroll will, with probability 1, surpass the bankroll of anyone using a different fixed-fraction strategy.

The trade-off is volatility. Full Kelly can produce drawdowns of 50% or more before recovering. For a bettor with a $10,000 bankroll, watching it dip to $5,000 during a cold streak — even while maintaining a genuine edge — tests psychological limits. This is why fractional Kelly (typically half Kelly) is the standard in professional circles: it captures most of the growth while dramatically smoothing the ride.

Practical Application in Sports Betting

To use the Kelly Criterion for sports betting, you need two things: a probability estimate for the outcome you plan to bet on, and the decimal odds the bookmaker is offering. The formula f* = (bp − q) / b converts these into a recommended stake size as a percentage of your total bankroll.

For example, suppose you estimate that a tennis player has a 60% chance of winning a match, and the bookmaker offers odds of 1.90. Plugging into the formula: b = 0.90, p = 0.60, q = 0.40. So f* = (0.90 × 0.60 − 0.40) / 0.90 = (0.54 − 0.40) / 0.90 = 0.1556, or roughly 15.6% of your bankroll. At half Kelly, that becomes 7.8% — still a sizeable bet reflecting the significant edge.

Now suppose the odds drop to 1.50. Recalculating: b = 0.50, f* = (0.50 × 0.60 − 0.40) / 0.50 = (0.30 − 0.40) / 0.50 = −0.20. The negative result tells you this is a −EV situation — despite a 60% win probability, the odds are too short to provide an edge. This is where many recreational bettors go wrong: they back likely winners at insufficient prices.

Key Assumptions and Limitations

The Kelly Criterion rests on several assumptions that are rarely fully met in practice. First, it assumes your probability estimate is accurate. In sports betting, probabilities are always estimates, not exact measurements. Even sophisticated models have calibration errors. Second, it assumes bets are independent. In reality, many sports bets are correlated (for example, multiple bets on the same team across different markets). Third, it assumes unlimited betting opportunities at the stated odds, ignoring liquidity constraints and bookmaker limits.

Because of these limitations, the fractional Kelly approach is not merely a preference — it is a practical necessity. Operating at half or quarter Kelly provides a buffer against the inevitable imprecision in your inputs. Many professional bettors cap their Kelly fraction at a maximum of 3-5% regardless of the calculated output, adding another layer of protection against overexposure on any single event.

Kelly Criterion vs. Flat Staking

The simplest alternative to Kelly is flat staking — betting the same fixed dollar amount (or the same percentage of your starting bankroll) on every wager. Flat staking is easy to implement and psychologically comfortable. However, it ignores the size of your edge: a bet with a 10% edge receives the same stake as a bet with a 1% edge, leaving growth on the table.

Kelly-based staking, by contrast, is edge-proportional. Bigger edges get bigger stakes; smaller edges get smaller stakes. Over hundreds and thousands of bets, this leads to dramatically faster bankroll growth. Academic simulations consistently show that Kelly growth outpaces flat staking by a wide margin, even when account is taken of real-world imprecisions. The practical recommendation for most bettors is to start with flat staking while building confidence in their probability estimates, then transition to half-Kelly once they have established a positive track record over at least 500 graded bets.

Bankroll Management and the Kelly Criterion

The Kelly Criterion is not a standalone tool — it functions best as part of a broader bankroll management framework. Your bankroll should be a segregated amount of money that you can afford to lose entirely. Never use rent money or savings as a betting bankroll. Serious bettors typically start with a bankroll equivalent to 50–100 flat-bet units (e.g., $5,000 with a $50 base unit). Kelly then dynamically adjusts your unit size as your bankroll grows or shrinks, naturally compounding winners and protecting during losing streaks.

Pair your Kelly calculations with meticulous record-keeping. Track every bet: the event, your estimated probability, the odds, the Kelly-recommended stake, and the outcome. Over time, this data reveals whether your probability estimates are well-calibrated, which sports or markets give you the strongest edge, and whether you're disciplined enough to follow the system through inevitable drawdowns. The Kelly Criterion gives you the mathematics; discipline and data give you the execution.