Many charts fail because levels are drawn from the execution timeframe alone. Those lines look precise until the higher timeframe reveals they sit inside a wider zone or against a larger swing.
A practical method: mark the obvious swing on the higher timeframe first, then refine the zone on the middle timeframe without inventing new lines for every wick. If a lower-timeframe reaction occurs at the edge of that zone, you have confluence—not a new thesis.
Delete levels that never attract a reaction across at least two frames. Sparse charts are easier to trust when the session gets fast.