the strategy of getting a PhD can be quite hit and miss - sometimes trading firms want very specific knowledge, e.g. specialist in solving certain kinds of PDEs, or DSP specialists, or bayesian/uncertainty statisticians. or sometimes they just want somebody who knows how to look at ambiguous problems and ambiguous term sheets and contract specifications and put it into solvable forms so that the already existing megabrains in the team can try to crack it. it really depends on the outfit. if one already has a phd and wants to get out of academia, then finance might be a good option. getting a phd purely to get into quant finance doesn't seem like a positive NPV investment (not just opp cost but also the soul crushing life of grad studies). as it turns out, quant funds also hire a lot of bachs and masters. not everyone needs to be churning out new models, 99% of the work is keeping all the systems and the analytics running and expanding existing knowledge to other markets.
Agreed - there is plenty of work for non-PhDs in both the buy and sell side. Quant developers are often possess undergraduate degrees, particularly in CompSci. Not much need for a PhD in the vast majority of systems development.
Yes but PhDs tend to be better on average than MFE graduates. If a fund has its choice of PhDs (which is the situation now) there is not much reason to hire someone with only a masters, except if she has some crazy skills.
Everything in this comment rings true from my experience (ex-science-software engineer, current equity derivative financial engineer at an investment bank, pining for the start-up fjords).
