An Associational Examination of the CaptialCube Effect Context for the MPV over the Linguistic Partitions: Testing Sensitivity & Specificity

Abstract

In this third examination of the CapitalCubeÔ Market Navigation Platform [CCMNP] we have selected the previously vetted set of embedded variables: Market Performance Variables [MPV] for their Linguistic Qualifiers [LQ] considering their directional market effects or MPV[LQ[{Neutral: Unfavorable: Favorable}]]. In the testing, we are interested in the Sensitivity and the Specificity of these vetted variables over the annual S&P500 Panel from 2005 to 2013. The inference framework employed a Median Split: High or Low for each of the 13 MPV tested and a random selection to avoid the FPE-jeopardy that is part of the Chi2 testing model. We used the Tamhane & Dunlop cut-off to identify Chi2 cells effects of interest and used these to develop the Sensitivity and the Specificity tests. Results: We were able to reject the a priori Nulls proffered for the testing protocols indicating that one may reject the supposition that the labeling of the LQ is formed by random processes in the CCMNP.