Opportunities don’t happen, you create them.
A good learning algorithm should be able to generalize from a limited set of examples to unseen data.
Economic theory provides the tools, but real-world testing refines them.