@Cyracuz,
...if they "fit" even if partially selecting relevant data for us then there is correlation in place, if they don't fit in the fundamentals, naturally they wont work...if we did in fact create the correlation on top of a reality, if we "invented" it without true correlations we would catastrophically fail in predicting behaviour...
...if I see a square where there is a circle I am seriously in trouble...it maybe that the circle is a sphere, but if I live in 2 dimensions seeing just a circle suffices and its fairly true...meaning there is a
true circular function coming up from the sphere...I am not creating reality I am seeing what is relevant in reality for my needs...