STATISTICAL AND NEURAL APPROACHES TO SMART-MONEY DETERMINATION

This paper deals with methods to estimate smart-money, a domain which is both open textured and vague. Here, smart-money estimation is studied in three ways: first, cases are statistically analyzed, second, some computerized methods used in DOLOR are studied, and finally, neural networks are introduced as a means to make reliable smart-money estimates. In terms of modelling, both the statistical formula as well as the neural networks can account highly for the variance found in smart-money granted. Therefore, neural networks should be added to the DOLOR system as an additional method to make smart-money estimates.