The NHC 5-day track is largely with the model guidance...the model guidance, that is, that actually captures the storm. The BAM-series models and the LBAR (and other models) are simplistic, older models and not given much weight in making a forecast.

It's also worth noting that the NHC 5-day track errors generally are lower than all of the model guidance except for the consensus (multi-model average) models and the FSU Superensemble. I'd put more trust in the NHC over the long haul than I would almost any model.