Continuing  our study of TBT, today’s post looks at how the XLF/TBT pair performs using the Z-Rank algorithm in lieu of the pure Z-Score.  The Z- Rank tends to have a slightly shorter excursion than the Z-Score and, as expected, “N” days drops to 13 from the original 14.  The Z-Rank also produces a 5 % improvement in the equity curve over the Z-Score. Truth be told, MR users who have spent time backtesting various pairs using both the Z- Score and Z-Rank know that these 2 algorithms frequently do NOT produce similar performance metrics and sometimes display wildly disparate results.  It is therefore very encouraging to see the alignment of the 2 Z tests and lends further credence to the technical stability of the XLF/TBT pair.  Tomorrow we’ll compare the trigger dates of each Z test to see how the 5% edge is realized.