The Gateway to Algorithmic and Automated Trading

Collateral advantage

Published in Automated Trader Magazine Issue 37 Summer 2015

Regulatory changes and shifts in technology requirements have delivered collateral management to the front office.

After cracks developed in global financial markets post-crisis, regulators embarked on sweeping reforms to the OTC derivatives world. The $630 trillion notional OTC derivatives market has been impacted by regulations such as Dodd-Frank in the US, EMIR in Europe and Basel III globally. Implementation means a major shake-up in how derivatives are settled, collateralised and reported.

Grigorios Papamanousakis, Quantitative Strategist, Aberdeen Asset Management

Consequently, the once obscure world of collateral management is being thrust into the spotlight.

For portfolio managers, there are two main concerns: the amount of liquidity required in the portfolio to post collateral to counterparties; and the amount of cash needed in order to face a client's liquidation.

This may sound simple. But when the marching orders are to optimise capital as efficiently as possible, the amount of computation required is actually quite complex.

As a result, more and more quants are coming into the space.

Moreover, standard computers can't be expected to keep with the billions of calculations that need to get made."The more collateral management transforms to a front office procedure, the more front office quants will be required," said Grigorios Papamanousakis, a quantitative strategist at Aberdeen Asset Management.

Making the right decision can have a big impact on the bottom line in a capital constrained world. Instead of posting cash directly, for example, it might make sense to get bonds via repo markets, even with the additional haircuts.

Presenting at the Alphascope conference earlier this year, Papamanousakis demonstrated using a hypothetical example how making different decisions on the order of posting collateral among varying tiers of haircuts (cash, UK government bonds, AAA bonds, high yield bonds) could mean a difference of many millions of pounds.

Source: Aberdeen Asset Management

"From the moment you have to add bonds into your portfolio as collateral, you have to monitor the credit risk of the bond - the credit quality - because if it downgrades or upgrades, that means you have to take it back or post additional haircut," he said.

Then there's decisions to be made about more complex collateral products, like credit default swaps or MBSs (mortgage-backed securities).

"The daily valuation is more complicated and more difficult to monitor," said Papamanousakis. "Computational impact increases exponentially when you add non-cash collateral, because we have to monitor and optimise the portfolio based on the credit quality of the bonds."

The moment that the first non-cash collateral has been posted from either party, he added, there's a need to start simulating the credit migration risk associated with that bond. And the same applies to equities.

The computational intensive exercise involved in modelling the best scenarios is "ideal" for parallelisation and acceleration using high performance computing, added Papamanousakis.

That includes: multicore processors, FPGAs, or cloud environments. A case study using GPUs shows acceleration of between 24 and 40 times depending on the number of scenarios. Instead of taking two hours to complete calculations, it takes four minutes.

The remainder of this article is only available to Paid Subscribers

Click here to purchase a subscription to Automated Trader

  • Copyright © Automated Trader Ltd 2018 - Strategies | Compliance | Technology

click here to return to the top of the page