We’ve added a new parameter to FRank that allows you to exclude N/A’s. It’s the third parameter (after the scope and the sort) and is simply #ExclNA. You can also use #InclNA, which is the default, but it’s not necessary. The parameter has also been added to FOrder.
For example, let’s say you want to create a universe that includes the top 40% of stocks in terms of next year’s estimated PE and EBITDA/EV (the formulas are ProjPENextFY and EBITDAEstNY/EV). You have a problem here. Because more than 60% of All Fundamentals are N/A, if you use #all, you wouldn’t exclude any stocks, since N/A’s would all get ranked around 65. You could exclude N/As in one rule and then use #previous, but with the second FRank you’d be applying #previous to the first one. The only way to do this used to be to use #GroupVar.
But now you can avoid using #GroupVar and simply add #ExclNA to your formula: FRank(“ProjPENextFY”,#all,#asc,#ExclNA) and FRank(“EBITDAEstNY/EV”,#all,#desc,#ExclNA).
Here’s another convenient use of the #ExclNA parameter. Let’s say you want to measure the performance of the top 50% and bottom 50% of various factors using a rolling screen backtest. If you include N/A’s and just change #asc to #desc, you will be including many of the same stocks in both tests. (If 10% of stocks are N/A, then stocks ranked 50-60 with #asc will be ranked 50-60 with #desc too.) Only by excluding N/A’s will you be testing altogether different stocks each time.
The #ExclNA parameter is also very useful for making aggregate series, especially as you go back to the early part of the century, when N/A’s for estimates and institutional data were more prevalent.