It appears that the calculation takes the 'frequency' (quarterly) and used that number of historical payments to sum the annual dividend.  With specialty payers, this vastly distorts the entire calculation and analysis. (see TSLX) I've seen this in many other instances as well.