To become a quant trader, which research areas should a machine learning PhD focus on?

Answer by Laurent Bernut:

42, for the meaning of life and Sideways markets and position sizing. I know, this looks unsexy and sounds counter-intuitive but… I am highly suspicious of delegating the thinking to a machine. What if You asked the wrong question and more importantly, what if You could not accept the answer…

Sideways markets

Trend following market participants make money in up or down markets, but give a lot of it back during sideways markets. Arbitrageurs compound a lot of small rapid gains during sideways markets but then get carried out as soon as trends emerge. So, rather than focusing on some “feel-good better sameness” smarter entries, focus on losing less. If You can know when You have entered a sideways market and when You are leaving it, the world will be yours

In simple terms, if You lose less than other people in sideways markets, You will be in a better position to benefit from emerging trends.

Position sizing

Trading edge = Win% *Avg Win% – Loss% *Avg Loss%

The Win% and Loss % are entries/exits. This has been ploughed ad nauseam. On the other hand, the field of position sizing is fairly fresh. There are some formulas coming from professional gambling, signal theory. Yet, they have not ignited the same level of passion as entries and exits.

The main reason why they have not sparked fire in the mind of such intelligent people is because most current platforms are signal processing engines. They test entry/exit on one lot/contract. Those one-dimensional engines never take into account the power of position sizing. They are designed to test traffic lights not horsepower.

To become a quant trader, which research areas should a machine learning PhD focus on?