issue 68 .. resolved 0 variance

R/fitFfmDT.R

@@ -101,7 +101,17 @@ specFfm <- function(data, asset.var, ret.var, date.var, exposure.vars,
     obj$model.styleOnly = TRUE
   } else {
     obj$model.styleOnly = FALSE
+    # this would prevent the issue of having one company in a sector..
+    # this would produce 0 variance whcih causes the weight to blow up iin 
+    # WLS...  I check for the number of companies per date becuase we fit a model 
+    # for each day... 
+    if (min( data[ , .N, by = c(date.var, obj$exposures.char)]$N) == 1 )
+        stop("
+             There is at least one ", obj$exposures.char, " that has one observation which will cause a 
+             problem with computing residual variance.")
   }
+
+
   obj$rob.stats <- rob.stats
   obj$addIntercept <- addIntercept
   obj$lagged <- FALSE