Not all combinations are beneficial. Some can have disastrous results on an extreme scale due to feedback loops. For example, the 2008 credit crisis resulted from an unfortunate combination of low interest rates, perverse incentives, easy loan underwriting standards, poorly understood mortgage-backed securities, and lax credit rating agencies.
As described in the Black Swan, we live in a world characterized by non-linear outcomes. Arguably, predictions are often wrong due to a failure to understand the combination and interplay of factors that can lead to unforseen results.
People adapt to dealing with such circumstances. The principles of value investing are built around the idea that you should not have to rely on predictions and should do as little predicting as possible. Margin of safety involves purchasing shares at a price well below intrinsic value to reduce potential losses from an incorrect prediction. In some cases, investors focus on what a company would be worth in liquidation (the worst case scenario), which removes reliance on uncertain future growth predictions and potential losses.
Perhaps prediction minimization, particularly on speculative macro factors, should be the general rule. But is it still possible, in some degree, to improve anticipation of combination effects, and in doing so predict the future?