Start by collecting data from NinjaTrader. Make sure you have a CSV exporter and a dedicated folder where every trade is automatically saved and organized.
That folder should capture everything you may need for future analysis, not just the basic trade information. The more complete your dataset is, the more you can learn from it later.
Run your strategy in auto execution and let it collect between 1,000 and 2,000 trades. The larger the sample size, the more reliable your analysis will be.
Once enough trades have been collected, stop the strategy and close NinjaTrader.
Open the folder containing your CSV files. Review the data to make sure everything was recorded correctly. Check for missing values, duplicate trades, incorrect timestamps, or anything else that could affect your results.
You can also create a scanner to validate the CSV files automatically instead of reviewing everything manually.
Once your data is clean, run your analysis through your terminal so you can get your information , At this point, you are not trying to predict the future. You are simply learning what happened and identifying patterns in your trading.
Once you can successfully collect, organize, clean, and analyze your data, you have completed the foundation.
Now build your first analytical model.
Its job is not to predict the market. Its purpose is to discover relationships and identify which market conditions produce your best and worst trades.
Think of this model as your research assistant.
After your first model consistently identifies , use those insights to build a second model whose purpose is prediction.((((Quant))))
once u build your models your models the model should give all the adjustments u need for each one of his ,
Beginners will always choose nodes ,
[quote=“icarus, post:1, topic:7119”]
Which part would you want automated first: data prep, back testing, dry-run, monitoring, or execution logs?
[/quote] Data prep I always first, or else it would be like I want to drive a car without gas