In our in-house developed automated trading strategy, we make a once / day calculation to create and send orders for the following day looking at daily charts only for each instrument. Even then I have found out that there is a tremendous difference in live order vs backtest order levels if we keep CalculateOnBarUpdate false, so we have changed it to true to have the same exact calculations in real life as the ones in backtesting.
Now the only difference remaining is the "Period Type of Non Day Series" which is set to "minute" in backtesting by default and is set to "tick" for live trading. My question is, if, to get a better representation of backtest in real life, we put the period type to minute for live trading as well, would this cause some weird effects like the profit target order being sent only after the minute bar closes, or the stop loss only activating at the close of a bar (instead of as soon as the tick level is reached).
What would you recommend to make sure backtest calculations of our algo are most accurately reflected in real life algo calculations as well ? Please note that I don't need tick granularity for this automated strategy, so I don't want to make the algo more cumbersome by introducing secondary time series etc... I just would like to make sure that our calculations on live data will be 100 % equivalent to the logic of the backtest.
I would be totally satisfied, if for example the initial calculations at the beginning of the day are made on minute granularity for real life, but then the profit target and stop loss behave on a tick basis (as we use tick based parameters for them in the strategy).
Now the only difference remaining is the "Period Type of Non Day Series" which is set to "minute" in backtesting by default and is set to "tick" for live trading. My question is, if, to get a better representation of backtest in real life, we put the period type to minute for live trading as well, would this cause some weird effects like the profit target order being sent only after the minute bar closes, or the stop loss only activating at the close of a bar (instead of as soon as the tick level is reached).
What would you recommend to make sure backtest calculations of our algo are most accurately reflected in real life algo calculations as well ? Please note that I don't need tick granularity for this automated strategy, so I don't want to make the algo more cumbersome by introducing secondary time series etc... I just would like to make sure that our calculations on live data will be 100 % equivalent to the logic of the backtest.
I would be totally satisfied, if for example the initial calculations at the beginning of the day are made on minute granularity for real life, but then the profit target and stop loss behave on a tick basis (as we use tick based parameters for them in the strategy).
Comment