It’s difficult to give a good answer without seeing the specific problem. There honestly could be numerous ways to solve this depending on the scenario.
One recommendation I have due to the amount of table that are being brought in is maybe look to summarize more of the information at the SQL level rather than in Power BI.
Try to imagine the model you want and then get your information straight from the database in that particular way.
If this isn’t possible that then you’ll have to imaging the end model you want and then try to work around it in Power BI.
19 fact tables does seem very extreme to me honestly. You shouldn’t really need that many to get the analysis you need, but as I mentioned it’s difficult to know the answer without seems the exact problem.