I’m testing a signal study for trading. I am doing some autotrading and backtesting, and I have a great problem with data.
I am going to describe what has happened to me this morning. It was autotrade enabled. The system has detected a signal, and has sent the order to market. The order has been filled, and after a fwe minutes, has been closed in its target.
I have stopped autotrading, I have changed the settings of the chart to other minute resolution. I have been making some test, and get back to the original minute configuration. The graphic has changed: The bar which before, in real time, was red, now is a doji. So, the system not detects the signal to entry the market. The result in backtesting, therefore, is not real!
Also, I do backtesting to a period of time in, lets say, 15 minutes chart. I get a result (lets say, profit of 1000$). Close the program. Open again. Run the same test. Result: 950$ profit.
So, that way, the work is unusefull: if what I see in real time is not the same than what is after close and open the program again, the results of backtesting doesn’t correspond to the actual results in live trading, making impossible and a nonsense waste time in researching and testing.
Is it a bug of the program? Is there any solution to this?