FRTB spurs data mining push at StanChart

Bank building “single golden source” of trade data in a bid to lower NMRF burden

data-mining

Standard Chartered is turning to data mining techniques in a bid to bolster its ability to model illiquid products, ahead of new market risk capital rules that punish markets suffering patchy liquidity and gaps in trade data.

The lender is building a central repository to host all of its trade data, which Albert Chung, the bank’s head of market risk analytics, Asia, for group market risk models, hopes will allow its risk and finance teams to share a “single golden source of data”. Such efforts could help the bank maintain an internal models approach (IMA) for more products under the Fundamental Review of the Trading Book (FRTB), Chung hopes.

Where the bank doesn’t possess all the required data sets to easily model a given product that has suffered a gap in liquidity, its data team has been trying to develop algorithms that can spin out synthetic time series based on historical data, he said.

“Data mining is now the sexy word,” Chung said. “On the market risk side, we’ve been looking at advance data proxy models: we use the data going to [trade] valuations, [and] we also use historical data. Our model is a simulation model. We try to use historical data and try to predict what could potentially happen tomorrow on the trading book.”

Chung was speaking during a panel discussion at Asia Pacific Financial Information Conference in Hong Kong on November 13.

Standard Chartered depends on Asia-Pacific for two-thirds of its profits – but many of the region’s markets suffer from patchy or seasonal liquidity, even in vanilla instruments like government bonds. Under FRTB, banks using their own models to set market risk capital requirements must slap on punitive add-ons for risk factors for which fail the rules’ threshold for stable and observable pricing – so-called non-modellable risk factors (NMRFs).

To qualify as modellable, a risk factor must be supported by 24 price observations during the course of a year, which cannot be more than a month apart. Many banks in local markets argue the rules were predominantly written by US and European standard setters, and do not make sense for developing markets. Many have called on regulators in their own jurisdiction to soften the regime’s NMRF framework when implementing the rules locally.

Despite the challenges – and despite reports that enthusiasm for the own-models approach is waning at some large banks – Standard Chartered intends to pursue IMA, Chung says: “If we go back to the standardised approach, it’s easily going to double down capital requirements.”

Instead, the bank has focused its efforts on bolstering its market risk modeling ability through technological innovations such as machine learning and data mining – monetising data already held in house by turning it into useable, surfaceable information, for instance by harvesting the quotes it sends to clients on retail and institutional trading platforms.

“The fact that some of the markets that we’re trading, we simply are not able to observe those prices because of illiquidity,” Chung says. “That’s when we really need to think about how we tackle those technical challenges. We’ve been talking internally. We’ve been trying to build internal IT solutions to allow us to directly access the quote platform that the bank uses to [price to] retail and institutional clients. That’s probably where we could leverage a bigger database based on transactable quote data for FRTB purposes.”

The bank’s data harvesting efforts run parallel to the banking industry’s attempts to team up with technology vendors such as DTCC, Bloomberg and IHS Markit to pool data. In some jurisdictions, banks are also attempting to pool their trade data among themselves and shun global vendors’ solutions.

The banks still have time as the Basel Committee on Banking Supervision in December 2017 announced a three-year postponement to the implementation of FRTB, which had been due for January 2019.

  • LinkedIn  
  • Save this article
  • Print this page  

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 indvidual account here: