Machine learning, Deutsche auction and repo haircuts

The week on Risk.net, September 16–20, 2019

7 days montage 200919

Some quant shops doomed to ‘struggle’ – López de Prado

Theory-first firms must modernise their methods or wither, says machine learning expert

Deutsche opens bidding for interest rate derivatives

Fixed income assets on the block after equity derivatives sale closes

Watchdogs ask EC to delay repo haircut floors. Will it?

EBA says hedge funds will skirt the rules, but Basel and FSB want haircut minimums in place

 

COMMENTARY: Seeing the wood for the trees

It should come as no surprise that credit card companies supplement their revenues by selling real-time access to consumer transaction data – albeit aggregated and anonymised – and even less of a surprise that enterprising hedge funds have found a way to monetise it.

This week, Risk.net reported how scrutinising data from millions of credit card transactions allowed a quant team to infer whether a company’s sales are on the up or trending lower – without the need to wait for quarterly sales reports to be published.

The analysis was delivered through a machine learning implementation of the random forest technique in which multitudes of decision trees combine to produce predictions. In this case, the algorithm enabled the quant shop to get an early warning on the health of companies whose options it held. It meant Neuberger Berman Breton Hill could adjust its option strategies accordingly, for example to control gap risk on earnings announcement days.

“Quants in their traditional fundamental models will usually have a value tilt, and we do as well,” says Ray Carroll, chief investment officer of NB’s quant group. “And the value tilt would say buy Macy’s. But if you look at trends from transaction data, it would say sell Macy’s.”

The strategy partly illustrates points made this week by Marcos López de Prado, former head of machine learning at investment firm AQR, who has a new venture that aims to disrupt the traditional quant asset management business.

López de Prado believes tomorrow’s winners in the investment industry are those who can crunch vast quantities of data to gain what he calls “statistically significant” insight. Unlocking the insight will require a more mathematical approach than many existing quant investors currently employ. These theory-driven “econ-quants”, as López de Prado terms them, are unlikely to succeed (adherents of Fama and French, look away).

Lightning-fast quantum computers would also help, López de Prado says. Mastering the qubit promises to transform investor efforts to find gold nuggets in the ever-growing torrent of data.

 

STAT OF THE WEEK

Dealers in the UK accounted for $3.67 trillion of the average daily global turnover of interest rate derivatives in April 2019 – 50.2% of the total. This is up from $1.2 trillion in 2016, when UK sales desks accounted for 38.8% of turnover. US dealers accounted for 32.2% of turnover, down from 40.8% three years prior. Hong Kong followed with a 6% share, up from 3.6% in 2016.

QUOTE OF THE WEEK

“We have shown that we are able to put these decisions in place very, very quickly, in 48 hours… [and] Esma is ready to recognise any UK CCPs in the event of the UK leaving before March 31 next year” – Patrick Pearson, head of financial market infrastructure and derivatives at the European Commission, on whether the European Union can extend temporary permission for UK clearing houses beyond its planned expiry date of March 30, 2020.

 

  • LinkedIn  
  • Save this article
  • Print this page  

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact [email protected] or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact [email protected] to find out more.

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: