>>/18248/
Human mind is very good at piecing together information from sparse and scarce sources, if you git gud enough it won't be much worse than using complete information on every facet because we're too dumb to compute all there would be possible to compute.
That's why all those machine learning algorithms are so successful despite being very simple approximators, just stacked together many many times.
Even the HARD SOIENCE is mostly very good approximations based on clever and elegant formulas like all the Fourier, Laplace and Taylor series.

But besides that: this is a well established mine on which many stubborn Berts will step on - overpaying for Merlier or Jonny. Philipsen was a better idea but not that great. Nys is a very good idea that solves most of your problems AND skors sam fakin assists if Jonny does not die on hills in a TdF uptempo.