Attempts to make sense of the complexities of our benighted world are necessarily reductionist. “Alabama,” for instance, did not redeem America (as if such were possible) and save itself from further embarrassment last night. Non-white Alabama did (again, to the extent that such was done at all). There’s really no such thing as a unified body of bitcoin marks: Crypto-enthusiasts have some radically different ideas about what fake currencies ought to be (to the extent that things that don’t exist can really “be” anything).
And so it goes with the hottest of hot categories, quants. It certainly seems that this group is unified in its ceaseless pursuits of more data and more artificial intelligence, but this is in fact not the case. Take, for instance, superhero enthusiast Cliff Asness. Certainly, the AQR chief is a quant. But Asness is of the “factor investing” flavor of quant, which basically means that the only numbers he believes in are market numbers, and not “alternative data” like satellite photos or rising global temperatures.
“In the factor investing world we worry a lot about finding spurious patterns by data mining. In big data combined with machine learning this is even more dangerous because the data sets are so big and machine learning is so good at finding patterns,” Mr Asness said….
“You have to fight the next war all the time,” he said. “People doing this are probably occasionally finding good signals, but in a matter of months they are competed away.”
That said, what’s the harm in giving it a whirl?
That is now shifting, with AQR running several experiments to see whether it can find profitable patterns in markets using machine learning to parse through novel data sets such as satellite pictures of shadows cast by oil wells and tankers, according to Mr Asness.
“We’re feeling our way,” Mr Asness said. “If our first few experiments bear fruit, we’ll do more of them. If we find out we’re good at this, it will become a bigger part of AQR.”