False results with Selected Timeframe setting and work-around?
1件の返信
When i use Selected Timeframe setting for ES-1hour data, I get many profitable strategies in Build, which also backtest fine in Ninja at 1-hour bars. However, when i run Market Replay (tick-by-tick), they are unprofitable, as most of my trades enter and exit through Limit (and a few Stop) orders. SelectedTimeframe gives false optimistic results because the orders are placed intra-bar without any consideration for the direction of the bar (which can only be determined by fine-grained minute/tick data).
1) The Help doc suggests using 1-minute precision, but i get an error message when i select 1-minute Test Precision with ES-1Hour Symbol-Period. How do i retest ES-1Hour at 1-minute precision?
Retesting with ES-1minute data will not work, because the strategy was intended to evaluate at each hour but is coded to evaluate at each bar, so it will trigger signals each minute.
2) Is there some other work-around in the Build phase to trigger an order only by approximating the correct direction of the Bar (e.g. exit above entry only on green candles, exit below entry on red candles)? Not having this check generates a lot of falsely optimistic strategies in Build, which fail on finer granularity
3) Trade on Bar Open setting – re-test doesn’t work with this option, as “strategy isn’t compatible with On Bar Open”. Well, how do i force SQ to trigger Limit/Stop orders at the next bar open?
ありがとうございます
キラン
0
WIth StrategyQuant 3 the NinjaTrader and Tradestation backtester can work only with Selected Timeframe precision
It is wise to use Selected Timeframe mode for Market orders and Tick simulation mode for Stop/Limit orders.
You should still be able to use limit orders with “selected timeframe” precision. This would require though to set SQ to avoid trading intra-bar using bigger SL and PT
0
1件の返信を表示中 - 1 - 1件目 (全1件中)