If I have a series of ten numbers (.22, .58, .61, .75, .77, .79, .91, .92, .95, .99) and rank them 1-10/decile, then give those rankings a weight relative to four other series of 10 numbers that I have also ranked, would you be confident in the value of my results which, yet again, combine those weighted rankings?
Numbers don't need to lie when they tell you nothing of value.
If x is not predicative of wins, player points, where the player will rank next season, prior season's stats/wins/whatever, what do you actually have? You have x, which along with $2 will buy you a lottery ticket.
Hockey advanced statistics are in their infancy, much like basketball. New approaches hint at interesting potential methods going forward, but with the puck and the player unchipped and many of the numbers collated by interns up to 300 ft. away from the ice surface without being cross checked, there's very little predictive value to be had.