The three derived roads don't record wins at all. They answer one question about the big road's shape: "is this shoe behaving repetitively or chaotically?" Red means the pattern repeated, blue means it broke — regardless of who won the hand.
Each new big-road cell asks: "does the depth pattern match what that earlier column did?" Match → red. Break → blue. It's a fixed mechanical rule — two people drawing the same shoe always get identical derived roads.
Players watch whether all three roads lean the same color. Three reds = the shoe has been strongly repetitive at three different depths; players who "follow the pattern" bet accordingly. Three blues = chaos; pattern-followers stand aside. This is the read behind our roads engine's "group agreement" signal.
Derived roads are deterministic bookkeeping of the past, not prediction. Their practical use is discipline: they make "the shoe is streaky/choppy" a defined, measurable statement instead of a hunch. Our terminal computes all three live and backtests how often each lean actually continued in that room — so you see the real number, not folklore. Also read: do betting systems work?