Experience alone produces very little improvement in noisy domains. What produces improvement is isolating one component, working at the edge of your ability, and getting feedback fast enough to correct.
In trading the components can be separated. Replaying historical sessions bar by bar trains entry recognition. Marking up charts before the open and scoring the calls trains reading. Rehearsing exits in a simulator trains the part that costs the most real money. Each has faster feedback than live trading, which delivers a noisy signal weeks late.
An hour of focused replay usually beats a day of watching. The measure of practice is not time at the screen but repetitions with feedback attached.
Related: paper-trading, post-trade-review, behaviour-journal, flow-state