Investment portfolios are into a dynamic process of optimisation to maximize expected profits. Under the diminishing expectations that the crisis produces, investors seek reliable information to evaluate their assets. Accounting statements include significant valuable data on the economic health of companies. The vast amount of accounting data and financial indices, can only be analyzed with sharp techniques of econometrics, or artificial intelligence. Hybrid neural-genetic Time Lag Recurrent Network is quite capable to process temporal information such as financial-accounting data. Further comparisons take place to the Multi Layer Perceptrons and Logistic Regressions. Additionally hybrid neural-genetic Time Lag Recurrent Network with Cross Validation and no hidden layers performed significantly higher than networks of the same architecture but with less hidden layers.