I am finding it challenging to build a successful automated trading system. The back test results are very compelling as I have built a system to get as much [INSIDE THE BARS] as possible buy designing the system in such a way that it can deliver positive backtests using Renko 3 brick sizes. However when running the system on replay, the same conditions that the back test reports poor results, the replay appears to be working. I would like to get a better understanding of back test results and the use and benefit of "walk forward". Thank you in advance for your feedback and insight. Have Backtests proven useful in the making of automated trading systems. FYI: I am using NinjaTrader Wizard Feature and have not unlocked code because I have not learned how to write code.
Look at a Renko-chart and a Range-chart side by side (using 3-Renko, look at 2-Range). At the extreme of each leg, you'll see on the Range-chart that price did actually exceed the Renko brick. If you backtest using a Renko chart, without directing your order to a 1-sec. timeframe, you are doomed for a failure. It seems you are just starting trying to automate ... there is NO way you'll get anywhere profitable by using Ninja wizard (or any similar tool) ... take this back, if you are working on 1/2h or higher timeframe, there is some (limited) hope. Anyway, you are embarking on a journey where 99% fail. I would suggest you quit right now, that would be your statistically best possible outcome. If you don't, do yourself a favor, learn C# & read Aronson's Evidence-Based Technical-Analysis. Cheers D.
Good Advice all around. LOL on last paragraph. Btw, I have developed automated systems using easylanguage on MC - with holding time measured in hours. If you are trying to hold for minutes, you really have no hope with TS/MC/NT kind of software.
I would suggest that you unlock the code and learn it that will give you a few a ton more flexibility.
Even with the best idea (strategy) if you have bad, incomplete or sub-par data then your "backtest" results will look decent but perform poorly in real-life. It's all about the data. Get good, realistic data and then try to backtest.