The role of data in gaining valuable financial insights

Deep with­in the world of finance, primeval data-sniff­ing crea­tures are grow­ing and evolv­ing. Great wafts of data are drift­ing from your mobile phone. Every dig­i­tal action sends a cloud of infor­ma­tion out across the net­work. You are leak­ing data and, in finance, these sig­nals can form a use­ful pic­ture of mar­ket activ­i­ty.

These crea­tures’ anten­nae are tipped by a vast array of sen­so­ry equip­ment; drones film­ing crop growth, vehi­cle activ­i­ty and shop­pers on high streets; vast elec­tron­ic “ears” lis­ten­ing to the clam­our of social media; a fin­ger on pul­sat­ing stock mar­ket price move­ments.

“Two lap­top com­put­ers are occu­py­ing an entire floor which used to be full of traders,” says Bartt Charles Keller­mann, chief exec­u­tive of hedge fund con­sult­ing firm Glob­al Cap­i­tal Acqui­si­tion. “That trend is accel­er­at­ing, the sophis­ti­ca­tion of these machines is increas­ing and investors see the writ­ing on the wall.

“They know that at some point they are going to con­vert most of their allo­ca­tions to strate­gies that are being run by machines, because they don’t have emo­tion, they don’t get out of the wrong side of the bed. They sell when they are sup­posed to sell, they buy when they are sup­posed to buy.”

This does not mean that peo­ple are out of the pic­ture. Behind every good trad­ing sys­tem is a good data sci­en­tist. Being data lit­er­ate, data sci­en­tists under­stand what par­tic­u­lar data does and does not rep­re­sent. They have to find the right datasets, check it for qual­i­ty and then build a mod­el that reflects the real­i­ty of the mar­ket.

If data is a con­stant flow of rich infor­ma­tion, it can be the equiv­a­lent of a sat­nav in nego­ti­at­ing the mar­ket

In its recent paper, Big Data and AI Strate­gies: Machine-Learn­ing and Alter­na­tive Data Approach to Invest­ing, invest­ment bank J.P. Mor­gan not­ed that new datasets are often larg­er in vol­ume, with greater veloc­i­ty and vari­abil­i­ty com­pared with tra­di­tion­al datasets such as dai­ly stock prices.

Among the alter­na­tive datasets it not­ed were being used to guide invest­ments, it includ­ed data gen­er­at­ed by indi­vid­u­als in social media posts, prod­uct reviews, search trends and so on; data gen­er­at­ed by busi­ness process­es; and data gen­er­at­ed by sen­sors.

If data is a con­stant flow of rich infor­ma­tion, it can be the equiv­a­lent of a sat­nav in nego­ti­at­ing the mar­ket. Yet in many cas­es, that data is more akin to the cat’s eyes in the road, with a sparse and unre­li­able dis­tri­b­u­tion, not com­plete enough to base a deci­sion on.

In those instances, firms may need to go out and search for data sources that can be used to piece a more com­plete pic­ture togeth­er. AXA Invest­ment Man­agers has done just that, devel­op­ing its own tools in-house to sup­port traders and clients, by aggre­gat­ing fixed-income mar­ket data.

Paul Squires, glob­al head of trad­ing and secu­ri­ties financ­ing at AXA Invest­ment Man­agers, says: “We want­ed to cre­ate an envi­ron­ment where the data that our trad­ing coun­ter­parts pro­vid­ed us was made mean­ing­ful beyond the nor­mal mar­ket inter­ac­tion. We want­ed to incen­tivise them to give us data that we would then respond to, to cul­ti­vate a bet­ter dia­logue.”

Cur­ren­cies and many stocks trade fre­quent­ly, but bonds trade more infre­quent­ly, mak­ing it hard­er to find accu­rate price infor­ma­tion, and where they can be sourced or sold. Com­bin­ing the use of mul­ti­ple data sources with human trad­ing skills can pro­vide a real advan­tage in trad­ing at the right price and exe­cut­ing a deal opti­mal­ly.

“Data is a key dif­fer­en­tia­tor in your abil­i­ty to trade a bond at the right lev­el, but if you were only look­ing at the data, you would hit a lot of prob­lems,” says Mr Squires. “Frankly there is no price dis­cov­ery until you ask a coun­ter­part to make you a price, so hav­ing a pre-trade pic­ture of where a price should be is only part of the price dis­cov­ery process. Pick­ing up the phone and get­ting a firm price to trade at is anoth­er.”

There are also lim­its to exist­ing datasets as Mr Squires notes that any mar­ket stress sit­u­a­tion can make data mean­ing­less. “For bonds, in par­tic­u­lar, if you have, for exam­ple, five mil­lion to sell and no one wants to buy them, there is no price – it’s as sim­ple as that,” he says.

Data is not only a tech­nol­o­gy play, although invest­ing in the tech­nol­o­gy to analyse it is fun­da­men­tal. Whether a retail bank, an invest­ment bank or a fund man­ag­er, firms that excel are able to embed the idea of data as an asset in their busi­ness. This ensures data is con­stant­ly sought and gath­ered in such a way that it retains its val­ue.

“It is less a tech­ni­cal issue and def­i­nite­ly more a cul­tur­al issue,” says Matthias Krön­er, co-founder and chief exec­u­tive of Ger­man dig­i­tal bank Fidor. “It’s a mat­ter of atti­tude what you want to do it with it. I can use data in the sense of man­ag­ing peo­ple on the one hand, but on the oth­er, I can use it so my cus­tomers can access data eas­i­ly and get the out­come of data algo­rithms eas­i­ly in a way that is actu­al­ly help­ful for their finan­cial life.”

As tech­nol­o­gy evolves, what can be con­sid­ered use­ful data is chang­ing. Analy­sis of graph­ics and writ­ten word as unstruc­tured data, as opposed to the struc­tured rows and columns of fig­ures held in data­bas­es, is cre­at­ing new oppor­tu­ni­ties. Satel­lite pho­tog­ra­phy and Twit­ter feeds are all sup­port­ing trad­ing ideas.

“It’s all about unstruc­tured data; find­ing data, clean­ing it, putting it into some kind of ana­lyt­i­cal frame­work from which the PhDs can then make heads or tails of,” says Mr Keller­mann. “There is still some scep­ti­cism whether or not that real­ly adds val­ue; a lot of the social media data, peo­ple have two views on. Sen­ti­ment has an impact, but clos­er to an event. So this sort of data is a very sophis­ti­cat­ed ani­mal and you have to under­stand mul­ti­ple dimen­sions of it to eval­u­ate its true val­ue.”