What could a physics PhD do to find a finance job?

Other than brushing up on your C++ and doing some side reading on mathematical finance, your first priority should be to do good physics research.  The important thing about Wall Street firms is that they are looking specifically for *physicists*.  Your research has no obvious connection to finance, but as a physics researcher, you will quickly find that there are non-obvious connections. If you try to set up your research based on what you *think* the flavor of the day is in finance, you will guess wrong, because the papers are outdated, and no one has a clue what math and computational techniques will be useful in X years.  But if you have a decent physics background, you can take whatever your learned and apply it to whatever the problem is in X years. It's not what you know, it's how fast you can learn.

My intuition is that the days of physicists just jumping into Fenance are mostly over. However, finance firms like D.E. Shaw, etc., and fixed income shops are always looking for smart guys who can express themselves well. My impression is that having some computational chops is useful but not necessary.

You might be better to consider switching to a Quantitative Analytics course. The big question an interviewer might ask is: "Why did you study Physics, if you wanted to work in Finance?"I've conducted many interviews for financial institutions,  some where this very question was raised - and often the perceived "ease" of CS you mentioned above is a pit-fall. Once inside a large finance organisation you may need a broader portfolio of operational skills from server-back end to various GUI / UX skills, and everything in-between, in various languages; using various developmental paradigms. It's clear that based on your direction of study you have the intellectual capacity and academic commitment to cross into the Finance arena. Also, it's good that you know a language like C as this is the grandfather for many other languages currently in use. This might be a better course of study to pursue.. These type of courses will also guide you in the right direction technically. Knowledge of C is useful, but understanding the syntax of a procedural language is very different to understanding the architectural structure of modern multi-discipline distributed applications. After all,  there are qualifications in more "Finance-relevant" fields. Is it the financial subject-matter you're interested in, or did you watch "Margin Call"  (great film by the way) and decide the money was more attractive and easier to obtain than a career in Faculty? Computer science should be no more a fall-back career for a physicist, than physics would be for a chemist. The issue here is that you're trying to enter Finance via Computer Science, and (by the looks of it) at heart; you're a Physicist.Footnoteshttp://www.mastersindatascience.org/careers/quantitative-analyst/

If you want a finance job, you need to take as much numerical analysis as you can.  Numerical analysis is the main thing they are looking for because you can model financial systems.   Hidden state models and monte carlo methods as well as modeling unkown functions using splines or polynomials is useful.   In physics, you usually know the applicable non-linear functions or differential equations.  In finance, the functions are usually unknown, but presumed to exist.   Linear algebra and group theory are also important.    It could even be possible to model the behaviors of individual traders using a type of Quantum Mechanics, where individual traitors are assumed to be linear combinations of various Eigen states of trader.