Daddi et al .
have recently reported strong clustering of a population of red galaxies at z \sim 3 in the Hubble Deep Field–South .
Fitting the observed angular clustering with a power law of index -0.8 , they infer a comoving correlation length r _ { 0 } \sim 8 h ^ { -1 } { Mpc } ; for a standard cosmology , this r _ { 0 } would imply that the red galaxies reside in rare , M \geq 10 ^ { 13 } h ^ { -1 } M _ { \odot } halos , with each halo hosting \sim 100 
galaxies to match the number density of the population .
Using the framework of the halo occupation distribution ( HOD ) in a \Lambda CDM universe , we show that the Daddi et al .
data can be adequately reproduced by less surprising models , e.g. , models with galaxies residing in halos of mass M > M _ { min } = 6.3 \times 10 ^ { 11 } h ^ { -1 } M _ { \odot } and a mean occupation N _ { avg } ( M ) = 1.4 ( M / M _ { min } ) ^ { 0.45 } above this cutoff .
The resultant correlation functions do not follow a strict power law , showing instead a clear transition from the one-halo–dominated regime , where the two galaxies of each pair reside in the same dark matter halo , to the two-halo–dominated regime , where the two galaxies of each pair are from different halos .
The observed high-amplitude data points lie in the one-halo–dominated regime , so these HOD models are able to explain the observations despite having smaller correlation lengths , r _ { 0 } \sim 5 h ^ { -1 } { Mpc } .
HOD parameters are only loosely constrained by the current data because of large sample variance and the lack of clustering information on scales that probe the two-halo regime .
If our explanation of the data is correct , then future observations covering a larger area should show that the large scale correlations lie below a \theta ^ { -1.8 } extrapolation of the small scale points .
Our models of the current data suggest that the red galaxies are somewhat more strongly clustered than UV-selected Lyman-break galaxies and have a greater tendency to reside in small groups .