get_pricing with machine learning on the backtest

I'm trying to implement my notebook research code on the backtest. My algorithm uses past price data to learn and predict, past data(price) can be of 6 months old, I was able to get that data using get_pricing on the research platform. Is there any way to get that efficiently in the backtest ?
I prefer not to use data.history(ex: data.history(context.security_list,'high',1000,'1d') as it'll gonna take a lot of resources.

Have you taken a look at Thomas Wiecki's posts on machine learning? Part 2 specifically addresses moving from the research environment into implementation within a factor. Part 3 brings it all together with then using that factor in an algo. You are right in preferring not to use data.history. The same data can be retrieved with a custom factor and pipeline is better optimized to fetch data. Additionally, pipeline is allotted more compute time than scheduled functions.