Any ideas on how to implement Tensorflow model predictions into a backtest

Hello my fellow quants

I’ve been trying to run a backtest on some USD/JPY data based on the predictions of the AI model and I could really use some help on where to start, I have been fiddling with zipline for a while but as someone who has never used Quantopian It’s almost a different language for me.

Any help, ideas, sample code :heart_eyes::heart_eyes:, would be much appreciated.

Thanks

Edit

It’s really important for the dates of the Fx data that I will run my backtest to be in sync with the dates of the machine learning Dataset as I am not making predictions based on any price action.

As it’s been 7 months since you posted I am guessing you may have found your solution.

I am not an expert, but this helped me understand TensorFlow and the breakdown.

Siraj Raval - TensorFlow in 5 Minutes (tutorial)