When develloping scalping strategies which work on (fast) Renko charts, the current backtest is useless, because it misses intrabar triggering of stoploss. To test a strategy the only valid option now is to use market replay. Because market replay speed is limited to 500x it can be a very time consuming procedure to test a strategy over 3 months.
Therefore I suggest adding tick-based backtest.
At thispoint you'll probably come up with the argument that tick-based backtesting will take considerable more computing/processing power and resources. But so what ??
Right now I am running a strategy on market replay at 500x speed, and NinjaTrader takes 0% of CPU power in the task manager and takes a whopping 67Mb of RAM. Why ?? To me it seems the replay can run much faster. The highest % CPU power I've ever seen during market replay at 500x speed was 3% ....need more RAM, be my guest, even my browser takes more RAM then NinjaTrader...
In all seriousness, why does NinjaTrader choose to only use 5% of the computational resources of a modern computer, in favour of correct tick-based backtest results ?
And frankly, I don't care if the backtest would take longer to finish. Far more important is that the results are closer to reallity.
So the following suggestions:
1) give the user an option for tick-based backtesting, besides the current backtest.
2) higher speed (multi-threaded) market replay if no1. isn't possible. I'm talking 2000x - 5000x or better a speed that always maxes-out one CPU-core.
Marco
Comment