I don´t think there is a straight reply to you, as HCP metals might present twinning and slip on non compact planes depending on the T. I would be more worried about stacking fault energy than the crystal system. It is a bit tricky to consider changes in crystal structure and grain size and stacking fault energy at one. Better look for experimental results than over theorize using basic solid sphere modelling.

A paper by Schneibel & Heilmeier, Mater. Trans. 55 44-51, (2014) gives counterexamples to your expectation:

fcc metals:

Al    0.068Ag   0.068

Cu  0.14

Ni    0.158

bcc metals:

alpha Fe  0.6

Cr              0.9

Nb             0.04

Mo             1.77

Ta              0.556

W               0.788

hcp metals:

Zn    1.02

Mg   0.279

Ti     0.403

(Data collated from various sources, cited in the aforementioned paper.)

I agree with Prof. Cesar and Prof. Howie. The answer is not that straight forward. 

1) It depends whether Hall-Petch slope is for dislocation slip (Ks) or for dislocation twinning (KT). Please see the paper: "Grain size effect on deformation twinning and detwinning", Y. T. Zhu  ET AL.,J Mater Sci (2013) 48:4467–4475. 

2) Please see the paper which considers the dislocation slip mechanism:-  "Grain-boundary strengthening in nanocrystalline chromium and the Hall–Petch coefficient of body-centered cubic metals", D. Wu ET AL., Scripta Materialia 68 (2013) 118–121. This paper gives the opposite to your expectation. The Hall-Petch slope (KHP) and the resolved shear stress for source operation (Tc) are both higher for bcc than fcc and hcp metals. The reason is that the bcc metals seems to be more efficient in blocking the dislocation motion than that in fcc/hcp metals. 

3) Also please see the references: (a) R.W. Armstrong, Acta Metall. 16 (1968) 347–355 and (b) R.W. Armstrong, P. Rodriguez, Philos. Mag. 86 (2006) 5787–5796.