The flaw in this is that it assumes the economic conditions are always the universal determinant of a presidential outcome. While that has traditionally been the case, our current President has quite a track record of being a combo-breaker.
I would argue that their algorithm uses the economy as the primary determinant, rather than the sole determinant. True it doesn't account for other factors directly, but arguably it accounts for such factors indirectly, by integrating the states' previous voting history.
