This is a good question, which hadn't received enough researchers' attention. People regularly print different objects, some of them with strength requirements and the need for a method of strength estimation is high.
Good experimental way to estimate the change would be to find a COTS cast plastic object, be it ABS or PLA or whatever, buy 3-5 pieces, then reverse engineer it, copy the design and print it 3-5 times in different orientations. Then you need to start destroying your objects in a manner that coincides with your needs. If your parts would be experiencing compression - crush them, if stretching - tear them apart, and measure the required strength. Then compare and get relative strength, that you can further use in your calculations. You'll notice that adhesion between FDM layers is much weaker than strength of bonds in the layer plane, so you'll have two coefficients - one for Z axis, one for XY (note, that printer settings can heavily affect the result, so every coefficient will be a function of printing parameters). Can't tell 100% reliably, but SLA may have just one coefficient - I hadn't noticed any difference between parts' strength in Z and XY directions.
If you're not a fan of thorough scientific approach, then you can just print your part and test it under your target conditions as much times as necessary. Or make a casting mold, then a solid object.