Quote from Antisyzygy:
Academics basically stopped releasing seminal works in the 90's on mathematical finance. The stuff you see today is lower quality and/or sparse. It could be most of the ones doing the work went into trading themselves, or maybe quantitative finance since the rewards are larger. Either of those or the funding dried up.
Most of what you see today is hacks and con artists pushing some new set of indicators on you. This has always been a problem in finance fields, just look at all the "Educators" out there.
A quant is basically doing what an academic would do, but they keep all the information they gather to themselves or are obligated to by contract. Obviously they do something right otherwise we wouldn't see them making 100-200k a year plus bonuses and even more once they get more senior positions. I've looked into entry level quant positions and they start at 125k + bonuses.
Of course, this is your "feature" selection. Think of all market data as one large multidimensional space. You can select which dimensions you want to use for your model using a variety of methods. One for example would be to measure correlation in some regression (not necessarily linear).
That doesn't mean a PhD in mathematics or finance wouldn't have a better way to quantitatively look at information or develop models, or maybe they would look at it different enough it would open whole new doors to you. It just seems there seems to be no interest in that field these days, or the people who are interested get snatched up to be quants. You also have many traders who seem to push this totally anti-intellectual agenda and really don't even know the variety of mathematical tools they have available to them.
PnL obviously is the only metric we care about, but not necessarily the best to use. It could be your method will tank under certain circumstances another metric would expose. I am not saying you have a bad system, just throwing out there something to think about.
