There are no "combination types'--every type has its own unique logic.

If you're used to other systems for identifying types--Ayurveda, Chinese medicine, the system of somatotypes developed by William Sheldon, you may be used to thinking that there are broad categories and then many combinations. Ayurveda, for example, has three major types of people--Air, Fire, and Earth--and then four combinations (Air-Fire, Fire-Earth, Air-Earth, and a triple type that partakes of all three).

The GenoTypes don't work like that (though they do overlap with some of the types from other systems). Instead, they represent coherent wholes, six self-sustaining efforts to solve the problems of survival that our ancestors encountered.
I think of the GenoTypes as six different types of tractors, each designed to solve the challenge of a particular type of terrain. One tractor is built very high off the ground so that it rolls right over any stones or stumps it encounters. Its weakness, of course, is that it's not always so stable. Another tractor is build low and squat. You couldn't knock it over if you tried—but when it runs up against even a small stone, it's stuck. There's no one model you can invent that will encounter all difficulties equally well.

Likewise, you don't have an infinite number of solutions. Rather, there's a natural limit to the number of solutions and combinations you could come up with. Once you've given yourself a choice between big, small, and medium wheels, and between wide, narrow, and medium tread, you've pretty much exhausted the possibilities of wheel size—after that, the differences aren't so significant. And since you can't put giant wide-tread wheels on a small, agile tractor that's built for easy steering, or tiny little wheels on a huge, wide, bulldozer-like tractor, you've got a kind of natural limit to the ways the combinations can come out as well.

'The GenoType Diet" page 111

I think some readers may have difficulty with this idea, since it is almost impossible to have ALL the characteristics of one GenoType and NONE of the characteristics for any of the others, though perhaps our minds would like things to work that way, since we most often think in a bivariate (low number of variables) world, but inhabit a holistic, multivariate world.

The GenoTypes are actually clusters of data characteristics: Genes (like blood type), anthropometrics (like leg length or somatotype) and other traits, such as dermatoglyphics (fingerprints). These traits themselves overlap in their significance; blood types influence somatotype (some, like the A antigen, more than others); fingerprints, a measure of fetal-placental nutrient balance, are also a good indicator of asymmetry, another marker of developmental instability.

Clusters come in two flavors, ones where each datum is limited to one cluster (and one cluster only) and 'fuzzy' clusters in which the data can belong to more than one cluster. The key concept behind the latter is the idea of membership i.e what gets you into a particular cluster (or GenoType). Fuzzy logic allows partial membership in a cluster.

This is usually determined by distances between the data points and variations between the clusters with regard to membership criteria; the closer that two data points are the more likely it is that they will share the same cluster. Distances here can be measured in the normal (Euclidean) way; i.e "it's ten miles to town." or in something called the "Manhattan or Taxicab" method, which, if you've every lived in New York City, DOES qualify as its own unit of measurement.

You keep adding clusters until you hit what is euphemistically called the The Elbow Criterion: a point where adding another cluster doesn't add sufficient information or encompass any more variation. At that point you're done. Initially it looked like there were going to be a LOT of GenoTypes (like, 32!). But then it kept falling: to 16, then 8 and finally 6.

I was hoping for 7 (for good luck) but the elbow kicked in and that was that.

So, if despite the last last five minutes you are still left wondering why are there no combination types, let me give you the simple answer:

They already overlap each other!