TriOptima adds MVA calculations to XVA service

PUBLISHED:
1 May 2017
6 AM

BY:
Louie Woodall

Three firms have signed up to use the tool, which calculates MVA across 100,000 scenarios.

Nex Group is aiming to kick-start the practice of pricing in the cost of funding initial margin on non-cleared derivatives trades with the launch of a margin valuation adjustment (MVA) calculator for dealers and buy-side firms.

Large dealers began working on methodologies for calculating the newest derivatives valuation adjustment (XVA) ahead of the roll-out of the non-cleared margin rules last year – although as yet no single approach has won consensus support. The adjustment has proven to be far more complex than other XVAs, such as the conceptually similar FVA, or funding valuation adjustment, a measure of the cost of hedging uncollateralised derivatives with collateralised ones, which dealers have been passing on to clients since at least 2012.

“MVA calculations are highly challenging since it is necessary to calculate all sensitivities for all of the time steps in the time series to arrive at the lifetime cost of margin,” says Mireille Dyrberg, chief operating officer for Nex Group’s post-trade services division, TriOptima. “This is a vast amount of data and calculations, and for that reason the few large tier one banks which do calculate MVA use approximations, rather than performing the actual calculations.”

Risk.net understands that the large banks currently subject to the initial margin rules are not charging one another MVA at present. However, market participants predict it will become common practice as more institutions are brought in-scope and the requisite stack of margin collateral grows bigger.
Speaking at last year’s Risk USA conference, Tomo Kodama, managing director in counterparty portfolio management at Bank of America Merrill Lynch, said MVA would have to flow to clients “at some point”. As a result, buy-side firms are also eager to understand how competing dealers calculate margin funding costs and incorporate that into their trading decisions.

TriOptima’s MVA tool is part of the firm’s triCalculate technology suite, which generates independent trade and netting set level XVAs and risk sensitivities. A group of 22 banks and corporates are currently testing the triCalculate service, and
three are slated to start using the MVA tool this week. “The triCalculate MVA service targets any organisation that is interested in having a precise view on the lifetime cost of margin across the whole marketplace from tier one through to the buy side,” says Dyrberg, who also heads the triCalculate business. “[Measuring] MVA, and the other XVAs, enables market participants to optimise their trading strategies and serves to minimise the lifetime cost of trading and holding over-the-counter derivatives.”

TriOptima claims its service cuts down the time taken to calculate XVAs across asset classes and business units from hours to minutes. It uses a two-step process; the first step employs a probability matrix to produce a large grid of all the possible future states of a chosen portfolio, generated using graphics processing units (GPUs). The second step uses central processing units (CPUs) to run Monte Carlo simulations using these projections to calculate MVA across 100,000 scenarios; the industry standard is 1,000 to 10,000 scenarios. A typical GPU is 50 times faster than a single CPU, allowing the first-stage grid to be produced faster than with other technologies.

Quants say they welcome the move by vendors to offer such tools, which can be an alternative to in-house development or a means to benchmark internal calculations.


“What third-party providers can offer are the analytical capabilities and good interfaces to create a coherent view of initial margin across counterparties and allow for the optimising of positions,” says Nicki Rasmussen, chief analyst in the counterparty credit and funding trading desk at Danske Bank.