NHC has as of 2PM elevated their probability to 50% and I feel rightfully so. It is important not to be hung up on climatology; climate does not dictate what any given event will do, but is cumulative result of many seasons, then normalized.

SSTs
As others have gone great strides to point out and is common knowledge at this point, the SSTs in the region between the Puerto Rico archipelago and Africa are in +AMO.

SAL:
There does not appear to be any SAL toxicity in the vicinity of the ingest regions for the zygote cyclone. There is some SAL related material observable here: http://cimss.ssec.wi...vetrak/sal.html ...but this is consistent with normal background contaminant - it would appear - and not directly associated with the burgeoning circulation of the system, either.

Deep layer shear:
Nearly non-existent at this time when studying the upper most elevation wind overlay at NHC's satellite website: http://www.ssd.noaa..../flash-avn.html The mlv cut is difficult to ascertain because cloud tags are used to determine vectors and they are obscured by a fairly healthy U/A anvil debris field; therefor, have to rely on modeling (as far as I am aware) to determine if there is a pesky intervening layer of shear.

Madden-Julian Oscillation related tropospheric tendency :
Currently the Atlantic Basin et al is in a neutral-positive 200mb anomaly state, http://www.cpc.ncep....m_monthly.shtml, which lay-term is usually conducive to sustaining, not compensating, for upward vertical motion in ambient field. Any system on the rail service between African and the western Basin would avail of that circumstance.

Integrated energy:
The circulation envelopes a huge area. Larger systems tend to spin up slower. However, given to their immensity they are somewhat more resistant to negative impacts. It all really comes down to momentum and larger systems having more of it, but there are also papers that discuss why in more technical terms.

I have not spent much time looking at models, but unless they are parameterized deliberately I can't imagine they would to exceptionally well until such time as the system is initialized in the input more sufficiently.

John