Forecasting hurricanes can be difficult. You have to get both where the storm is going (the track) and how strong it will be when it gets there (the intensity). Forecasting the damage a storm will cause is even harder. The reason is that damage is proportional to the cube of the wind speed (if the wind speed is v, damage is related to v^3, or v*v*v). So slight differences in the track or intensity can make a huge difference in damage. At 70 mph sustained winds, we would expect a large number of average (not hurricane resistant) wood frame houses with shingle roofs to have damage costing 3.7% of total value (mostly roof damage). At 75 mph, that goes up to 5.8%. So small changes in winds make a big difference in damage.
Compare these simulations of Tropical Storm Arthur as of this morning. Here is the official NHC forecast track and wind swath from my TARU model:
On this track and intensity, my ISTANU model says the impact should be about $30 Million dollars. Now compare that to the forecast using the HWRF objective track/intensity model:
On this track and intensity, impact top out at almost $400 Million dollars. By contrast, the GFDL model is only $25 Million,mostly due to disruption, not physical damage: