Kelly Criterion - Citations

No staking system is perfect. Some (Martingale et al.) are not even close. However, Kelly Criterion is proven to be the best way to grow your wealth. John L. Kelly was a colleague of the inventor of information theory, Claude Shannon. Kelly used information theory to show the optimal way to grow a bankroll.

The Kelly Criterion formula tells the trader the optimal bankroll percentage to risk. Percentage stakes is superior to level stakes but choose to high or too low a percentage and you will not profit as much from a winning system as you can with Kelly.

Here is a by no means complete list of Kelly Criterion citations, starting with Kelly's seminal work.

Kelly, John. "A new interpretation of information rate." IRE Transactions on Information Theory 2.3 (1956): 185-189.

Breiman, L. "Optimal gambling systems for favorable games." The Kelly Capital Growth Investment Criterion: Theory and Practice (2011): 47-60.

Chapman, S. J. "The Kelly criterion for spread bets." IMA journal of applied mathematics 72.1 (2007): 43-51.

Noon, Edmund. "Extending Kelly Staking Strategies to Peer-to-Peer Betting Exchanges." (2014).

Noon, Edmund, William J. Knottenbelt, and Daniel Kuhn. "Kelly's fractional staking updated for betting exchanges." IMA Journal of Management Mathematics 24.3 (2013): 283-299.

Noon, Edmund, and W. Knottenbelt. "Market making with an inverse Kelly strategy." 4th International Conference on Mathematics in Sport. 2013.

More articles on Kelly Criterion


  1. Kelly criterion is based on the fact we know odds and probability. In reality, we calculate probability from the odds. The formula is very smart and also very useless for calculating 'right' probabilities

    1. Kelly Criterion does not calculate a probability.

      The inputs are probabilities and odds, the output is a fraction.

    2. We know odds for sure. We do not know probability. We often calculate probability as 1/odds. If we do so - the criterion is useless.

    3. You have to create your own odds. Either fundamentally with past performance data to create an odds line or technically with strike rate for a particular technical trading strategy.

      Also, you can use fractional Kelly's to account for variance. It can be done. I do it and so do others.

  2. I cannot calculate odds. Odds are external thing that comes from exchange. I can try to calculate probability, but it's extremely difficult task. Anyway, punter and trader behave in different ways. Punter thinks - Federer will win, I will back at 1.05. Trader thinks - Federer will win, but is it very likely to have complications and coefs grow, I will lay 1.05. They are both right and may have good bet making opposite decisions

    1. I refered to technical trading where you have (for example) a 50% strike rate and your yield gives you an edge of say 5%. That is the same as odds, the odds put on your ability as a trader and not the odds of the underlying event.