Long time lurker (at least three years) and finally have something to add to the discussion.

Those of us wondering if the GFDL is out to lunch this week may not have long to wait. Until now the GFDL has been an outlier beginning at 48-72 hours but has been within the model consensus during at least the first 24-36 hours of each run.

As of the 00Z run the GFDL is predicting significant northward movement Real Soon Now and forecasts the COC will pass over extreme southwestern Hispanola early Sunday morning. If it sees something (maybe up around 300mb given the power of the storm) that all the other models are missing it should start to verify during the day Saturday.