All models and agorithms have some weighting factors built in (a good example would be coriolis force). This is what accounts for many models to be so persistant. If you noticed with Irma, tracks consistently called for recurvature over Florida. Early on, it was up the east coast. As time went on, the same track was constantly adjusted... same orientation, but from a more western perspective. All these tracks are based upon a storms current position. I have spent many years in the Western Pacific and know that any model is to be used as a tool, not as gospel.