ZiadFrancis/Reinforcement_Trading_Part_2 — reverse-engineered prompt

Build me a research pipeline for trading XAUUSD from MT4 or MT5 style 1 minute CSV data. I want to drop in a file with columns like time, open, high, low, close, volume, have the app safely handle broker time, resample the minute data into higher decision bars, and create clean causal features that are normalized in a sensible way for modeling.

It should let me run one main command that keeps the final test period hidden, tune and compare a simple baseline strategy on train and validation data, and save charts and reports for candles, indicators, volatility, sessions, feature relationships, trades, equity, and drawdown. I also want an optional reinforcement learning training path for bracket trades with stop loss and take profit choices, using walk forward validation so I can see out of sample performance across rolling windows instead of just one split.

Please include a one time final holdout evaluation step and a notebook demo so I can inspect everything easily. Look up current docs online if you need to.