I don't see where a three-part limit is justified.
In just one trait, skin color, I discussed a cline (range of variation) from extremely dark to extremely light, to albino. This reflects a real-world condition.
There are many other traits involved in evolution.
These traits can be expected to vary, co-vary, and counter-vary in all sorts of ways. Some mutations will have both beneficial and detrimental results (sickle-cell anemia being one example).
I simply do not see a need to restrict the possibilities to only beneficial, detrimental, or benign. There will be "little bit beneficial-somewhat detrimental"; "mostly benign here, detrimental 2000 miles farther north"; "mostly benign here, advantageous 2000 miles farther south"; and so on and on and on.
That is why, when the climate changes, the range of variation within a population allows some to adapt better than others--they are already carrying the mutation from hundreds, thousands, or millions of years ago; it was benign but now (look at those darn glaciers!) its beneficial.
That's the way evolution works, and why simplistic mathematical models often come up with screwy results.