The visualizations make simple scenarios clear, for example if you block a sweep point blank (or whatever a highly punishable tekken equivalent would be, a snake edge or a hellsweep or whatever), then you are at advantage and can land a strong punish.
But many moves that are highly disadvantageous on block are not punishable when used at their furthest effective range, because the opponent does not have a far reaching & fast enough move to punish it.
At higher levels of play, this kind of knowledge of move properties is actually used against you. There are characters who have uninterruptable block strings that ends in a disadvantageous move, but the string has significant pushback on block.
Players will use these strings to push the opponent to a deliberate spacing where it looks like they can land a long range normal to punish the last move of the blockstring. But the string is designed to place the opponent just outside of the range of that normal, where it looks like it will connect, but it it will not. In this situation, the person who did the "unsafe" blockstring can watch their opponent's character model, visually confirm the startup animation of the attempted punish, and punish _that_ move's recovery on reaction. Tricky tricky!