Backtest
Replay strategies offline
Download Dzengi candle history into a local cache, then run the same EMA/RSI/ATR evaluator the engine uses — with next-bar fills, re-anchored stops, and a simple fee model.
Candle cache
Download Dzengi history into a local cache (used by the simulator). Deep 1m history can take a few minutes.
No cached datasets yet.
Run backtest
Replay the EMA/RSI/ATR evaluator on cached candles (next-bar open fills, ATR stops, fee model). Split into train / holdout.
Find better params
Grid-search the system template (same fee / spread / holdout as above). Ranks by holdout green-day rate, then expectancy. Needs cached candles for the template symbol — can take a few minutes.