Kelly Criterion for multiple simultaneous bets
The Kelly Criterion is a bankroll management method that links the size of a wager to two things: the probability of winning and the price offered by the bookmaker or casino. It is designed to maximise long-term logarithmic growth, rather than to produce a profit on every individual bet.
Multiple bets create an additional problem. Your wagers may overlap, depend on the same match, or react to the same market information. Treating each selection as an isolated opportunity can result in too much money being exposed at once. A sound approach considers the combined portfolio, available bankroll, probability estimates and the level of risk that suits your circumstances.
| Betting situation |
Main calculation issue |
Sensible starting approach |
| One independent wager |
Compare estimated probability with decimal odds |
Calculate the single-bet Kelly fraction |
| Several unrelated bets |
Total exposure can become excessive |
Calculate each edge, then check the combined stake |
| Correlated bets |
Wins and losses may arrive together |
Reduce the allocation or use a covariance-aware model |
| Opposing or mutually exclusive bets |
Outcomes interact directly |
Analyse the whole market rather than each bet separately |
| Uncertain probability estimates |
Small errors can create large stakes |
Use fractional Kelly and conservative probabilities |
What the Kelly calculation measures
For a single wager using decimal odds, convert the price into net odds by subtracting one. The standard formula is:
f* = (bp − q) / b
Here, f* is the proportion of your bankroll to stake, b is the net decimal price, p is your estimated probability of winning, and q is the probability of losing, calculated as 1 − p. If the result is zero or negative, the theoretical Kelly stake is zero because the price does not justify a wager.
Suppose a tennis player is offered decimal odds of 2.20 and your analysis gives the player a 50% chance of winning. The net odds are 1.20, so the full Kelly fraction is approximately 8.3% of the bankroll. That is mathematically aggressive for many recreational bettors. Half Kelly would reduce the stake to about 4.2%, while quarter Kelly would use roughly 2.1%.
The calculation depends heavily on the quality of the probability estimate. A model based on injury news, surface performance and reliable data may be useful, while a quick hunch can make the output look more precise than it really is. Educational material and platform comparisons on CasinoBBG guides can help separate probability assessment from emotional selection.
Why simultaneous wagers need a portfolio view
If you place three bets at the same time, adding three individual Kelly percentages is only a rough shortcut. The bets may all be connected to one event. Backing a football team to win, backing it to lead at half-time and backing its striker to score are separate markets, but a single match result can determine the performance of all three positions.
Even apparently unrelated wagers can share hidden risk. A tennis bet and a baseball bet may be independent in sporting terms, yet both could be based on a model that overestimates home advantage. The practical result is that your estimated edge may be wrong across several selections at once.
A more complete multi-bet Kelly model maximises the expected logarithm of final wealth across all possible outcome combinations. In simplified form, it seeks the stake vector that maximises:
E[log(1 + total portfolio return)]
For small returns, analysts may use a matrix approach such as f = Σ⁻¹μ, where μ represents expected returns and Σ represents the covariance between bets. This is useful conceptually: higher expected edge increases allocation, while greater variance and correlation reduce it. The model becomes difficult when outcomes are mutually exclusive, prices move quickly or probabilities are uncertain.
For practical betting, calculate each positive edge, inspect the relationships between positions and set an overall exposure limit. This is safer than automatically staking every full Kelly recommendation.
A practical method for allocating the bankroll
Start with a fixed bankroll dedicated to wagering. Do not include rent, emergency savings or money needed for everyday costs. In Australia, using an account in Australian dollars can make the process easier because the amount at risk is clear and there are fewer surprises from exchange-rate movements.
Next, record each wager’s odds, estimated probability, market and stake. Convert all prices to decimal format and calculate the single-bet Kelly fraction. Then classify the wagers as independent, related or directly opposed. For example, two bets on matches played in Melbourne and Perth may be broadly independent, while two selections from the same AFL fixture are likely to be connected.
After that, apply a reduction factor. Full Kelly can create severe swings even when the underlying model is accurate. Half Kelly or quarter Kelly is commonly used because it sacrifices some theoretical growth in exchange for lower volatility. If you have limited data, quarter Kelly is often a more defensible starting point than trusting a highly exact probability.
Finally, check the total amount exposed if every position loses. A portfolio cap might be set as a small percentage of the bankroll, with lower limits for correlated markets. Recalculate after a result changes the bankroll; do not increase stakes simply to recover a losing session. A spreadsheet can automate the maths while keeping assumptions visible.
Correlation matters more than the number of bets
Correlation describes how bets tend to move together. Positive correlation means several positions are likely to win or lose at the same time. Negative correlation can reduce portfolio swings, although it does not automatically create value. The key question is whether the combined price compensates for the full risk.
Consider a soccer match in Sydney. A bet on the home side, a bet on the home side at half-time and a bet on over a certain number of goals might all be supported by the same attacking narrative. A model that treats them as three separate opportunities could recommend a much larger combined stake than intended. Reducing each fraction or selecting the strongest edge may be more appropriate.
Mutually exclusive bets require special care. Backing several runners in a horse race is not equivalent to placing several independent wagers because only one can win, and the outcomes are linked through the same market. Hedging can lower variance, but it may also remove the advantage if the prices are inefficient after commissions and margins.
Correlation estimates are often noisy in sports betting, particularly when sample sizes are small. It is better to use a conservative adjustment than to pretend the relationships are known precisely. Weather, team news and late market movement can also make previously separate positions converge.
Applying the method to Australian betting conditions
Australian punters encounter a mixture of local and international markets. Licensed online bookmakers commonly display prices in decimal odds, and many bettors compare football, NRL, AFL, cricket, tennis and horse racing in the same account. The Kelly formula works across these markets, but the probability estimate must reflect the sport, competition and available information.
Legal conditions also matter. Online wagering services offered to Australians are regulated differently from online casino products, and in-play online betting restrictions can affect how a strategy is executed. Casino players should check whether a platform is authorised to serve their location, understand withdrawal terms and avoid assuming that a familiar payment method provides consumer protection. Australian rules can change, so current official guidance should take priority over old betting discussions.
Promotions should not be treated as guaranteed edge. Turnover requirements, maximum winnings, market restrictions and expiring odds can alter the real value of an offer. The same applies to digital assets: a crypto balance can change in Australian-dollar value before a bet settles. A comparison such as this Bitcoin casino guide may help explain platform features, but it should not replace checks on licensing, identity verification, fees and local availability.
For casino games, the Kelly method is usually relevant only when a player has a measurable advantage, such as a legitimate promotional value or a game with a known positive expectation. Standard house-edge games do not become favourable because a staking formula is applied. Gambling should remain discretionary, with deposit limits and loss limits set before play.
The safest use of Kelly is as a ceiling rather than a command. Compare the calculated fraction with your personal tolerance, reduce it when probabilities are uncertain and keep simultaneous exposure under control. Reviewing educational betting material from Rockerhieu resources alongside bookmaker terms can support better decisions, while Australian players should use available self-exclusion and responsible gambling tools whenever betting stops feeling controlled.
Calculate the edge, account for correlation, use fractional stakes and record every wager. Treat the Kelly Criterion as a disciplined framework for managing risk—not as a guarantee of profit—and keep your betting within a bankroll you can afford to lose.