A set of $n$ points in the plane in generic position (no alignement of three points) has at least $2.012^n$ different triangulations of its convex hull involving only the set of given points.We call such a triangulation a checkerboard-triangulation if every interior vertex belongs to an even number of triangles (ie. triangles can be colored black-white with adjacent triangles being of different colours, equivalently, vertices of a checkerboard-triangulation can be coloured using only three different colours). A "random" set of $n$ points is certainly in generic position and the parity of the degree of a fixed vertex in such a triangulation should be "random", ie. even for about half of all possible triangulations. Assuming independence of parities of degrees of different vertices, one should thus expect the existence of checkerboard-triangulations for large random configurations of points.
Otherwise stated, let $A_n$ be the number of combinatorially different sets of $n$ points in generic position (also called "order-types") and let $B_n$ be the number of such sets (up to combinatorial equivalence) which have a checkerboard-triangulation. Does $B_n/A_n$ tend to $1$? (Two configurations are combinatorially equivalent if they are isotopic in the set of generic configurations, endowed with the obvious topology).
Warning: There are arbitrarily large generic sets of points with no checkerboard triangulation: Take $(x_i,y_i)$ with $x_1=-1<x_2<\dots<x_n=1$ and $y_i=x_i^2-1$ and add $(x_{n+1},y_{n+1})=(0,-2)$. Every triangulation is obtained by joining the first $n$ points $(x_1,y_1),\dots,(x_n,y_n)$ to the last point $(0,-2)$ and by triangulation the convex hull of the first $n$ points. This leads always to the existence of a point of odd degree among $(x_2,y_2),\dots,(x_{n-1},y_{n-1})$.