Self-Correcting Kelly Strategies for Skeptical Traders

The Kelly criterion gives the appropriate bet size in idealized situations with known parameters. In financial trading situations parameters are generally unknown and the mathematical assumptions underlying the Kelly proof are not met precisely. Moreover a risk manager typically must cooperate with a trader who may be skeptical about both the Kelly criterion specifically and the concept of mathematical optimization of bet size in general. This presentation tackles the problem of designing a Kelly-based system for setting trade risk management parameters that is both self-correcting (the system delivers good results even if initial parameter are misestimated or parameters change) and acceptable to skeptics (its performance can be continuously validated)