speaking as an engineer
The fundamental problem with physics is the results are only as good as the model. Theoretical results are valid under theoretical conditions; practical conditions always contain a long list of qualifiers and modifiers that can completely overwhelm the theoretical underpinnings of any problem. Example: in theory, carbon fiber is an obvious aerospace material. In practice, many of the carbon fiber components of the Boeing 777 were swapped out for aluminum because the non-destructive testing options for composite materials are severely curtailed compared to traditional aerospace alloys.
Any stochastic system has this problem. Statistically, you can predict that five out of 100 assemblies will fail in the first year. In practice, you have no idea that 50 of them will fail because your distributor's warehouse is marginally more humid than your permeable packing materials can handle. It comes down to the model: if you have all the inputs modeled correctly, all the outputs will be correct. However, statistics in general (and economics in particular) relies on rejecting spurious inputs.
Physicists are drawn to economics because both trades boast complex systems with remarkable reducibility. Wolf populations, stock market cycles, harmonic oscillators and spring dampers can all be successfully modeled with first and second order differential equations therefore there are mathematical underpinnings to the whole of the universe. Staring into that shit is like staring into The Matrix. But in practical terms, the dependencies that influence the wolf population are not the primary dependencies that influence the survival of one wolf. The equities markets can be modeled, the behavior of individual stocks (and investors) cannot. A spring is an easy equation. A worn spring under varying temperature conditions less so.
Economics equations are stupid simple compared to physics equations. Even Newton's Four have some squares'n'shit in them. The Phillips Curve, which the Federal Reserve believes governs employment, has like one log in it. Algebra. That's where the most prestigious economists in the world think the secrets of one of the most important functions of economics lives. Everything we do, the assembled mass of the US economy, and it all comes down to
(wage growth) = (wage growth)^(trend rate) - f(unemployment rate)
Economists predicted the likelihood of a global recession in 2007 at two in a billion - it was a "six sigma" event. This despite the fact that economics is essentially a history of recessions and depressions. It's not that everyone is stupid - it's that they come to worship the model more than what they're modeling. I would argue that this weakness can also afflict physicists.
over here in the real world, we think you're all arrogant. Any profession that steadfastly insists on the relevance of asymptotes over margins of error has no place in the real world.