The common approach to dark matter is to obtain a candidate particle in a framework designed to solve some other problem of the standard model. The most studied example is the lightest neutralino in the MSSM. In this case, the dark matter particle is a by-product of a theory whose main motivation is to solve the hierarchy problem. This kind of attitiude is perfectly understandable from the psychological point of view. By the same mechanism, a mobile phone sells better if it also plays mp3s, makes photographs and sings lullabies.
But after all, the only solid evidence for the existence of physics beyond the standard model is the observation of dark matter itself. Therefore it seems perfectly justified to construct extensions of the standard model with the sole objective of accommodating dark matter. Such an extension explains all current observations while avoiding the excess baggage of full-fledged theoretical frameworks like supersymmetry. This is the logic behind the model presented by Marco.
The model is not really minimal (adding just a scalar singlet would be more minimal), but it is simple enough and cute. Marco adds one scalar or one Dirac fermion to the standard model, and assigns it a charge under SU(2)_L x U(1)_Y. The only new continuous parameter is the mass M of the new particle. In addition, there is a discrete set of choices of the representation. The obvious requirement is that the representation should contain an electrically neutral particle, which could play the role of the dark matter particle. According to the formula Q = T3 + Y, we can have an SU(2) doublet with the hypercharge Y= 1/2, or a triplet with Y = 0 or Y = 1, or larger multiplets.
Having chosen the representation, one can proceed to calculating the dark matter abundance. In the early universe, the dark matter particles thermalize due to their gauge interactions with W and Z gauge bosons. The final abundance depends on the annihilation cross section, which in turn depends on the unknown mass M and the well known standard model gauge couplings. Thus, by comparing the calculated abundance with the observed one, we can fix the mass of the dark matter particle. Each representation requires a different mass to match the observations. For example, a fermion doublet requires M = 1 TeV, while for a fermion quintuplet with Y = 0 we need M = 10 TeV.
After matching to observations, the model has no free parameters and yields quite definite predictions. For example, here is the prediction for the direct detection cross section:

The model was originally introduced in a 2005 paper. The recent paper corrects the previous computation of dark matter abundance by including the Sommerfeld corrections.