The common financial ratios are Net Profit/Net Sales, Net Profit/Net Working Capital. There are ratios suggested by Beaver, Altman etc. Better if you take standard deviations of the ratios. You start with a number of ratios and then you can use F-test to segregate the ratios and finally use discriminant analysis to find how the ratios change.
Read 'Financial Statement Analysis - A New Approach' by B. Lev. You will also find a new technique called information analysis (entropy) developed by Lev.
Michal, the basic question is whether the ratios give more accurate prediction years before failure. Researchers' claim varied percentage of accuracy. Lev's book is important as it would provide you a direction of your future research which you may not find in Ravis paper. Pertinently, can we build up a theory or even an acceptable explanation on the basis of ratios for failure mechanism. Lacking a mechanism, as Lev put it, it is like a fishing expedition where a researcher chooses a large number of ratios and then reduces them to a relatively few by various methodologies for model's enhanced prediction ability. But none of the ratios reported could be taken for sure as the best representative ratios - the lineage of research in this area (identifying many different ratios for prediction- as a result we cannot evaluate comparative efficiency of the models) is the proof of this comment. Lev's book, therefore, is still important in this area of study.
Study of financial distress is very diverse as it is very difficult to generalize the things. here it is important to define the stages of financial distress. Moreover, definition of sample and context is also important to identify changes in financial ratio.