Ex-Credit Suisse quants embrace machine learning

Founders of XAI Asset Management grapple with unsupervised learning and the problems of explainability

It’s been three years since Credit Suisse’s head of data science Aric Whitewood and chief global strategist, Jonathan Wilmot left to co-found XAI Asset Management, their artificial intelligence-infused shop. The hedge fund’s core machine learning system has been trading real money in a variety of asset allocation strategies for 18 months.

As well as a proprietary macro trading business, the firm creates and sells machine learning-based predictions and analysis to other asset management firms seduced by the allure of AI. One chief risk officer at a $500 billion US investment manager says XAI Asset Management has become his go-to resource for novel, yet explainable machine learning models.

Operating outside a big investment bank brings its own advantages. “It is really hard in a big company to do anything new,” Whitewood says. “You have particular constraints, and it takes a long time to get infrastructure set up in-house. Now we are pretty much unconstrained. Any particular computer processes we need are easily available, so it really frees us up to do fundamental research.”

XAI is one of a number of firms looking to benefit from a growing curiosity among investment houses to explore machine learning techniques to help super-charge returns by unearthing hidden patterns in data, or for more prosaic purposes such as trade automation, improved execution or risk management. To meet this need, established firms are staffing up new teams of data scientists, or leaning on specialist providers such as XAI.

Whitewood describes XAI’s investment approach as an AI variant of a global macro systematic fund that dynamically allocates risk between assets, optimising, for example, for Sharpe or information ratio. He likens it to relatively high-speed tactical asset allocation, where new decisions are potentially made on a daily basis.

On the analysis side, the firm applies machine learning models to financial data to try and determine two things. Firstly, what state the market is in and then, what is likely to happen next. Applications range from the simple: for example, in a regime where interest rates are rising, predicting which sectors are likely to outperform. However, XAI’s focus is on using machines to deal with more complex regimes with a greater number of variables and a mixture of shorter-term and longer-term trends.

“One of the benefits of these [machine learning] techniques is that you’re not just using a linear regression and trying to fit some model to what’s going on. You’re looking at what the data is telling you, but you’re not losing sight of economic fundamentals and macro information,” Whitewood says.

Unsupervised – which is closest to how most animal and human learning seems to work – is one of the most powerful sub-fields within machine learning, but also one of the hardest

Aric Whitewood, XAI Asset Management

XAI, which was known as WilmotML until a year ago, is working on a number of specific problems and projects. First, how to combine datasets that are different from each other; a market timing tool that looks for particular periods of risk; and an associated set of asset allocation strategies where machine learning helps identify complex, non-linear behaviours in the market.

Whitewood declines to divulge exact techniques being used as they are proprietary to XAI. However, in the above applications, he says: “We are taking a mixed approach, with a key component being an unsupervised learning stage which is where you don’t have labels in the data, you just learn what the relationships are. Unsupervised – which is closest to how most animal and human learning seems to work – is one of the most powerful sub-fields within machine learning, but also one of the hardest.”

The danger with complex AI is that it is difficult, if not impossible, to explain its inner workings to potential clients or regulators. Some fund managers, among them BlackRock, have had to shelve certain techniques on this basis, or limit their use. XAI says there are creative ways around the problem. The firm focuses on explainability for its predictions, including visualising the driving regimes in its models. Some of this is taken from other domains such as image recognition.

“Effectively you can create charts and visualisations where you can see what regimes have been learnt by the machine and how they relate to each other,” Whitewood says. “You can also look at the sub-features that make up the regimes, these could be spikes in particular time series or other interesting events.”

The process can be challenging, as variable sets driving the predictions can change quite often – although larger changes in the strategy portfolio allocations change less frequently. So on top of this, XAI has looked at taking the input datasets that are most important, and grouping them together for comparison; the goal is to collapse down all the input data into a smaller number of groups.

Whitewood, in keeping with his academic interest as an honorary lecturer at London’s UCL, has a wider eye on developments in the theoretical field. He is cautious of overselling the benefits of deep learning, the branch of machine learning that aims to mimic the workings of the human brain.

“It’s interesting to note that the conversation is now changing – where current deep learning architectures were once viewed as the answer to most problems, I think more recently researchers are re-evaluating that view. That’s not to say that current deep learning architectures are not useful, they certainly are, but probably have to evolve,” he says.

Editing by Alex Krohn


Stress buffer will not upend Citi’s capital plans

Citi gave away $22 billion to shareholders in 2019, depleting its Common Equity Tier 1 (CET1) capital to $137.9 billion, its lowest level since Q3 2013. But although the Federal Reserve intends to impose a new stress buffer soon, Citi managers do not expect to have to turn off the capital taps.

The New York-based lender reported a ratio of CET1 capital to standardised risk-weighted assets (RWAs) of 11.7% for Q4 2019, up from 11.6% in Q3 but down from 11.9% the same quarter a year ago. Citi is targeting a ratio of 11.5%.

Total CET1 capital is down 21% from a 10-year peak of $173.9 billion in Q4 2015.


Who said what

“We've obviously worked down over time much of the excess capital that we have … so there’ll be less excess that’s there. We’ll go through the CCAR [Comprehensive Capital Analysis and Review] process as we've done in the past, and try to responsibly come up with as much as we can return to shareholders. In terms of the stress capital buffer … we haven’t seen anything as of yet that would need to come out, I think, by middle of February. We are continuing with the normal planning of our CCAR submission” – Mark Mason, chief financial officer at Citi.

What is it?

Capital requirements established by the Basel Committee oblige banks to hold a minimum ratio of capital to RWAs. Three binding ratios apply: the CET1; Tier 1; and total capital ratios. CET1 capital refers to the highest quality capital, namely shareholders’ equity.

Why it matters

Citi stockpiled capital in the post-crisis years to met regulatory requirements and assuage market concerns. With investors and supervisors alike assured of its stability, Citi started returning cash to shareholders in 2015 after watchdogs gave a green light to its capital plans. In 2015, its payout ratio was a lowly 55%. Last year, it hit 122%.

Even with a lower CET1 ratio, managers expect Citi to be well-capitalised once the Fed’s stress capital buffer (SCB) takes effect. But chief financial officer Mark Mason implied this was partly because regulators have said the changes to the post-crisis framework aim to be capital neutral in their overall effect.

If the SCB is more onerous than anticipated, it may cause Citi and its peers to revise their future capital distributions.

Get in touch

Sign up to the Risk Quantum daily newsletter to receive the latest data insights.

Let us know your thoughts on our latest analysis. You can drop us a line at or send a tweet to @RiskQuantum.

Tell me more

Fed’s rush to complete stress buffer likely to unnerve banks

CCAR: JP Morgan, Capital One adjust planned capital actions

View all bank stories


People moves: NatWest Markets senior shake-up, and more

Chris Marks: stepping down at NatWest Markets

Latest job changes across the industry

Chris Marks is stepping down as chief executive officer (CEO) of NatWest Markets and chair of the supervisory board at the bank’s Netherlands-based subsidiary, NatWest Markets NV.

Robert Begbie, treasurer of Royal Bank of Scotland, will become interim CEO, and join both the NatWest Markets board and NatWest Markets NV supervisory board. Marks will remain in his current position until June.

Meanwhile, Richard Place is leaving his position as chief financial officer at NatWest Markets and his post will be taken up in the interim by treasury finance director Robert Horrocks. Place will remain in his current role until March.

The appointments are subject to regulatory approval.

Elodie Blanc has been promoted to the newly created role of deputy head of investment banking capital markets (IBCM) at Credit Suisse. London-based Blanc will continue in her current role as IBCM chief operating officer (COO)for Europe, the Middle East and Africa (Emea), which she has held since 2018, reporting to Warren Young, global IBCM COO. She will also remain as co-lead on the European Union exit programme. Blanc joined Credit Suisse in London in 2002, and has held several IBCM posts in Emea and New York.

Brian Franz becomes State Street’s new London-based chief information officer (CIO), reporting to COO Lou Maiuri. Franz moves to the company from drinks firm Diageo, where he was chief productivity officer. He joined Diageo in 2008 as global CIO. Before that, Franz was CIO at PepsiCo International, and held positions at General Electric and AT&T. He has experience in artificial intelligence, machine learning and robotics.

Karolina Burmeister

Citi has appointed Karolina Burmeister as country head for Finland, replacing Brodie Donaldson. Burmeister will report to Alexander Stiris, country head for Sweden and head of Nordic corporate banking. She joined Citi in 2018, and previously worked at OP Financial Group in corporate and investment banking.

Deutsche Bank has named Christian Berendes as head of government and regulatory affairs. He reports to chief administrative officer and management board member Stefan Simon. Berendes succeeds Karin Dohm, who will take on new responsibilities. Berendes is also COO for Germany, a role he will retain, in which he reports to Karl von Rohr, president of Deutsche Bank. Berendes was previously chief of staff to CEO Christian Sewing.

Aengus Hallinan, group head of operational risk and business continuity at Credit Suisse in New York, has left the company to pursue other opportunities. At Credit Suisse, Hallinan helped spearhead the identification of key ‘survival processes’, such as treasury and funding, which has reduced the number of key controls needed to ensure continuity from the hundreds down to fewer than 10.

BNY Mellon Markets has appointed Paresh Shah as global COO for the foreign exchange business. Based in New York, Shah reports to Jason Vitale, global head of forex. Shah joins from HSBC, where he was the Americas COO for forex, commodities and corporate sales.

Markus Schmidt is the new head of Emea emerging markets foreign exchange and rates trading at Crédit Agricole. Schmidt moved to the French bank from Bankhaus Lampe in Dusseldorf, Germany, where he was co-head of financial markets and head of financial markets trading for one year. Previously, he spent seven years at Standard Chartered Bank in London.

Morgan Stanley has hired Michelle Wang as managing director in the fixed income capital group. Wang is currently on gardening leave after departing Credit Suisse, where she spent 12 years in the debt capital markets and derivatives solutions team. At Morgan Stanley, Wang will report to Damien Matthews, co-head of fixed income capital markets for the Americas.

Rahul Mani has joined Barclays as a macro trader for South-east Asia in Singapore. Mani moved to the bank from Nomura, where he worked in macro products and liquid markets proprietary trading from 2011. Prior to that, he worked in rates and forex trading at Morgan Stanley for almost two years. Mani started his career as an associate at Credit Suisse in 2006.

Bill Grant has joined Pimco as a vice-president in the global equity team, reporting to Eden Simmer in New York. Grant was most recently director of synthetic equities trading at Deutsche Bank, which he joined in 2009. He started his career at UBS Securities in Stamford in 2001.

Pedro Gustavo Teixeira: role change at ECB

Two executives at the European Central Bank are to swap roles from February. Petra Senkovic becomes director-general of the ECB secretariat and Pedro Gustavo Teixeira becomes director-general of the secretariat to the supervisory board. Before taking up her current role,

Senkovic was deputy director-general of the ECB’s legal services department. She holds a PhD in law from Sorbonne University. Teixeira started work at the ECB in 1999, initially in prudential supervision and financial stability. He holds a PhD in law from the European University Institute.

Separately, the European Council has appointed Fabio Panetta and Isabel Schnabel to the ECB’s executive board. Panetta is deputy governor of the Bank of Italy, having joined the central bank in 1985. Schnabel is a professor at the University of Bonn and a member of the German Council of Economic Experts.

Foreign exchange company Oanda has appointed David Grant as COO for Asia-Pacific. Grant will be based in Singapore, and takes on responsibility for overseeing the firm’s administrative and operational functions, as well as improving the efficiency of the business. Prior to Oanda, Grant was Asia-Pacific COO at forex trading firm Gain Capital.

Oanda has also hired Mark Chesterman as head of trading and Lucian Lauerman as head of business solutions. Both report to Kurt vom Scheidt, who has taken on a wider role as COO.

ASX, the Australian securities exchange, has appointed Robert Woods as a non-executive director. Woods joins from Australian investment firm Challenger, where he was head of strategy and, prior to that, chief executive of the fund management and asset management units. Stephen Knight will replace Woods as chairman of ASX Clear and ASX Settlement. Knight was previously a director of ASX’s clearing and settlement boards. Earlier, he headed NSW Treasury Corporation.

Carolyn Colley has been named non-executive director of ASX’s clearing and settlement boards. Colley is currently a non-executive director of OnePath Custodians, Oasis Fund Management and Smartgroup Corporation, and COO and co-founder of Faethm, an analytics software company. She has held senior executive roles at Macquarie Bank, St George Bank and BT Financial Group.

Julie Becker has been appointed as deputy CEO of the Luxembourg Stock Exchange. Becker joined the exchange in July 2013 and was appointed to its executive committee in 2015. She has spent two decades working in Luxembourg’s financial sector, including positions at the Central Bank of Luxembourg and Dexia.

Lucas Bruggeman

Lucas Bruggeman will become CEO of Swiss stock exchange BX Swiss on February 1, succeeding Harald Schnabel. Bruggeman moves from Sentifi, a fintech company and alternative data provider, which he joined in 2013. He is currently a partner, responsible for sales and marketing. Before Sentifi, Bruggeman was a member of the board of management at Liechtensteinische Landesbank in Zurich.

Bakkt, a digital trading and payments company, has promoted Mike Blandina to CEO. This follows the departure of founding CEO Kelly Loeffler, who moves into politics to become US senator for the state of Georgia. Blandina joined Bakkt in 2019, as chief product officer, after working at OneMarket, Westfield, PayPal, Google and Blackhawk Network.

Adam White becomes president of Bakkt, having joined as COO after the firm’s launch in 2018. Prior to that, he worked at Coinbase. White began his career as an officer in the US Air Force.

Andres Cifuentes has taken on the new role of executive director for institutional sales at cryptocurrency firm BitGo, has learned. Cifuentes moved to the company from Cboe Global Markets, where he worked in foreign exchange sales from 2016. He started his career as an analyst at Goldman Sachs in 2004, moving to JP Morgan where he worked in prime brokerage for four years. In 2009, Cifuentes joined Morgan Stanley’s forex prime brokerage business unit and worked in forex sales for three years.

The Securities Industry and Financial Markets Association has appointed Justin Sok as managing director of the federal advocacy team. Based in Washington, DC, he reports to Jamie Wall, executive vice-president of advocacy. Sok joins from the US Department of the Treasury, where he developed and implemented policy positions, including tax and trade, as a senior adviser in the Office of Legislative Affairs. Prior to that, he served as a policy adviser for several members of the US House of Representatives.

Su Yen Chia has been named head of Asia-Pacific public policy at the International Swaps and Derivatives Association. Chia moves from Euroclear Bank, where she was head of strategy and government affairs for Asia-Pacific, having joined in 2017.

The Securities and Exchange Commission has chosen David Bottom as its new CIO. Bottom oversees the agency’s information technology systems and security. Previously, he was CIO and chief data officer at the Office of Intelligence and Analysis at the Department of Homeland Security. Bottom also worked at the National Geospatial-Intelligence Agency as head of its information technology directorate and deputy director of enterprise operations for 10 years.

Law firm Seward & Kissel has promoted Kevin Neubauer to partner. Neubauer joined the firm in 2009. In 2012, he became in-house counsel at Providence Equity Partners before returning to Seward & Kissel three years later.


Podcast: Andrew Dickinson on CCPs’ defence mechanisms

Trades’ size limits, membership rules and more transparency key to avoid another CCP default

For this Quantcast, spoke with Andrew Dickinson, who leads the CCP analytics group at Bank of America.

With Leif Andersen, global head of the quantitative strategies group, Dickinson developed a model to assess exposure to a central counterparty. Crucially, the model has the ability to capture wrong-way risk stemming from the presence of clearing members with outsized positions. Such positions can trigger the member’s default in the case of large, adverse market moves.

If the defaulting member has an unhedged position and is unable to meet the margin calls, the CCP’s default fund could suffer significant losses. The case of power trader Einar Aas’ default at Nasdaq Clearing in September 2018, and the default fund’s subsequent loss of $119.7 million, set alarm bells ringing for banks, CCPs and regulators.

“In our view, initial margin on its own is not a sufficient risk mitigant in isolation. It needs to be complemented by suitable controls in order to prevent members clearing outsized, unhedged positions,” says Dickinson.

The concerns of broker-dealers have been debated in meetings organised by the Futures Industry Association and the International Swaps and Derivatives Association. The message from the industry seems to be unanimous.

“We and our peers are suggesting there needs to be a strengthening of the regulation to either impose limits or stronger membership criteria, or at the very least have greater transparency,” says Dickinson, lamenting the scarcity of transparent information released at the time of Nasdaq’s default.

He explains how their previous model was extended to include a probability distribution, a Student-t, which allows for arbitrarily fat tails to model rare events. Its output is one easy-to-interpret figure that can provide precious information to all of the parties involved and help them decide how to deal with such circumstances.


00:00 Intro

02:02 Central clearing vs bilateral clearing

05:05 One bad apple

07:10 What are the results in the paper?

09:40 Key components of the model

12:30 Wrong-way risk

17:00 How this model helps understanding of Einar Aas’ default at Nasdaq

19:15 What info does the model give you and how can you use it?

20:50 Who should decide to adopt this model?

23:15 Feedback from CCPs and regulators

26:20 Model’s further developments

To hear the full interview, listen in the player above, or download. Future podcasts in our Quantcast series will be uploaded to You can also visit the main page here to access all tracks, or go to the iTunes store or Google Podcasts to listen and subscribe.


Judgement day looms for dealers in swap shift to Sonia

Regulator pushes Q1 deadline for users to adopt risk-free rate as norm for interdealer trades

Grief is said to exist in five phases: denial, anger, bargaining, depression and acceptance. Derivatives users are experiencing a similar range of emotions as they switch from Libor to alternative risk-free rates. Now, regulators have added a feeling of urgency after giving banks a deadline to shift interdealer swap activity to the alternative risk-free rate for sterling markets, Sonia.

The move intensifies the pressure on dealers to ditch the mistrusted benchmark and adopt Sonia as the standard way of pricing new trades by the end of March. Despite a nagging sense that not all participants have made the necessary upgrades to systems and infrastructure to handle the new fixing, dealers accept the rationale for the change.

“To be able to move the balance to Sonia within the interdealer market, making Sonia the market convention is exactly the right thing to do,” says a senior rates trader at a US bank.

“We always knew this would happen,” adds Phil Lloyd, head of market structure and regulatory customer engagement at NatWest Markets. “It’s just a question of when it would happen and what the catalyst for it would be.”

The Financial Conduct Authority’s Edwin Schooling Latter told an audience at’s Libor Summit on November 21 that while Sonia is already the norm in new issuance of floating rate sterling bonds and securitisations, significant volumes of new Libor swaps maturing after the end of 2021 – the date which Libor is expected to cease – are still being struck.

Year-to-date volumes of Sonia-based overnight index swaps stood at £3.55 trillion ($4.62 trillion) versus volumes of Libor instruments, including interest rate swaps and forward rate agreements, at £2.62 trillion, according to clearing house LCH’s latest data.

“In sterling interest rate swap markets, we will be encouraging market-makers to make Sonia the market convention from Q1 2020,” said Schooling Latter, the FCA’s director of markets and wholesale policy. “That does not, at this stage, mean no more sterling Libor swap transactions for those who have a particular reason to prefer Libor, but it does mean making it standard to quote and offer swaps based on Sonia rather than Libor.”

Sonia liquidity has traditionally been concentrated at the short end of the curve, with Libor remaining the dominant fixing for longer-dated swaps. But, anecdotally, Sonia volumes are starting to grow at the long end due to greater activity by insurers and liability-driven investment funds, particularly in the final two months of 2019. LCH was unable to provide a breakdown of sterling swap volumes by tenor.

“The short end of the sterling market is principally Sonia-based flow,” says Alistair Sharp, Credit Suisse’s head of interest rate trading for Europe, the Middle East and Africa. “90% of the over-the-counter business that I do has some element of Sonia to it – whether that’s Sonia only or Sonia versus Libor – so to a large degree that side of the market has already transitioned. The goal is now for the medium and long end of the market to transition to Sonia now too.”

The US bank’s senior rates trader says long-dated volumes are now evenly split between Sonia and sterling Libor, so dealers have little excuse not to use Sonia as the norm.

“The reality is that there is nothing stopping dealers from trading Sonia with each other, other than the fact that the market infrastructure is set up for Libor and has been for a while,” says the trader. “But that infrastructure needs to change and I think Q1 2020 is as good a point as any other.”

Full stream ahead

To hasten the shift to Sonia, interdealer central limit order books (Clobs) would need to start streaming Sonia prices, traders say. Currently, trades occur via request-for-quote protocols.

“The more people can see prices then the more people can deal on them and mitigate risk on them, which in turn will result in more Sonia liquidity,” says Credit Suisse’s Sharp. “Streaming Sonia prices is part of the process that will make Sonia more dominant.”

To this end, the FCA is working with dealers, platforms and other infrastructure providers to move Sonia swap trading on to lit markets. Speaking on a webinar in October, Dan Marcus, chief executive of Trad-X, confirmed his platform already has Sonia swaps in its testing environment ahead of a planned launch of the full service in the first quarter, with a transition period due to start in February.

Members of the working group on sterling risk-free rates, convened by the Bank of England, have committed to stream executable quotes for one-, three- and six-month Sonia overnight index swaps from February, according to the minutes of the November meeting.

It’s a case of changing the mindset of the interdealer market and getting them to accept the fact that in the near feature they’ll be delivered basis risk if they do a trade versus Libor as opposed to a trade versus Sonia

Senior rates trader at a US bank

The US bank’s senior trader believes Sonia streaming might pose a challenge for dealers who don’t currently have the set-up or capability to do so, but those parties should be in the minority.

“A lot of people are already streaming and providing pricing in Sonia swaps to clients so this isn’t that big an ask in terms of the costs or technology involved,” the trader says.

“It’s more a case of changing the mindset of the interdealer market and getting them to accept the fact that in the near feature they’ll be delivered basis risk if they do a trade versus Libor as opposed to a trade versus Sonia. That’s a big change for the market to get its head around,” the trader adds.

Basis risk would affect counterparties who are left holding Libor exposure in a market that is increasingly moving to Sonia.

While Credit Suisse’s Sharp agrees that a reasonably liquid electronic marketplace already exists for Sonia at the short end of the curve due to a lot of activity from pension funds, bank treasuries and corporates, this is less the case for the medium end of the Sonia curve due to a lack of demand from end-users.

“Unfortunately, to get a Sonia Clob you need more than just a bunch of market-makers willing to provide prices, you also need end-user demand. That’s the part which is more challenging but we’re all doing our bit to make sure we can facilitate this,” he says.

Swaption step

As well as helping the swaps market in particular to shift away from Libor, the streaming of firm Sonia swap prices through Clobs is also deemed a crucial step to transition other parts of the derivatives market to the risk-free rate.

For example, swaptions and many rate-linked structured products rely on the Ice swap rate – a measure of term swap rates out to 30 years – for settlement. Ice Benchmark Administration, which manages and publishes the Ibor-based rate for sterling, US dollar and euro, currently takes live swap pricing from three multilateral trading facilities, BGC Partners’ BGC Trader, Icap’s i-Swap and Tradition’s Trad-X.

Following an initial consultation on its planned overhaul for the Ice swap rate – partly fuelled by recent publishing failures of the rate in the US – IBA is expected to finalise its approach for an introduction of a Sonia-based variant of the Ice swap rate to run alongside the existing Libor version. The final methodology consultation is planned for early 2020.

The development of Sonia trading on Clobs would also help along the production of forward-looking term rates based on the overnight benchmark. These rates, which the FCA wants reserved for niche areas of the market only, will be based on tradeable futures and short-dated Sonia swap quotes.

In sterling interest rate swap markets, we will be encouraging market-makers to make Sonia the market convention from Q1 2020

Edwin Schooling Latter, FCA

Swaptions have made a slower start to the switchover. In August, NatWest Markets and HSBC became the first banks to trade Sonia-linked swaptions, but little progress beyond a few token trades has been made so far.

While traders expect that a Sonia alternative to the Ice swap rate will help nudge along the uptake of Sonia swaptions, the products haven’t gained much traction. This is because liquidity in swaps – which banks use as a hedge – is still concentrated at the shorter end of the curve, whereas swaptions are often used by long-dated investors like asset managers and pension funds. A lack of observable Sonia swap levels at those maturities also makes pricing the swaptions difficult, with limited liquidity meaning that Sonia swaptions are likely to be priced higher than Libor swaptions.

“Libor is still the hedging instrument of choice within the interdealer market and so the cost of a [swaptions] trade against Libor is the cheapest. That cost argument wins over the risk-free rate argument and so volume still ends up sitting in Libor,” says the US bank’s senior rates trader.

For the swaptions remaining linked to Libor, banks also aren’t keen to create basis risk. For example, the trader says users will likely still want to hedge their Libor swaptions portfolio with Libor swaps.

However, a move to a world where Sonia is the interdealer standard would hugely simplify the current way banks hedge longer-dated swaps. Currently, a trader would have to hedge a fixed versus Sonia swap with a six-month Libor swap first, as it is the most liquid product. It would then need to conduct a basis swap between six-month Libor and three-month Libor, which allows it to then enter into three-month Libor versus three-month Sonia basis swaps – the most liquid Libor to Sonia instrument – as a final step.

“Essentially there are three hedge trades you have to do whenever you’ve got a fixed trade versus Sonia with an end-user,” says the US bank’s rates trader. “That’s what this Q1 2020 goal is trying to move the needle on.”

Carrot or stick

While helpful, the trader believes that regulators could also do more to help market participants in their transition to Sonia than simply giving the market a Q1 2020 goal, suggesting regulators give firmer guidance or explicit targets for how they expect to see Sonia volumes develop.

“Something in that vein would be helpful and the working group obviously has a role to play here as well in terms of disseminating that message,” says the trader.

NatWest’s Lloyd agrees that additional regulatory guidance would be helpful, pointing to the success of the FCA’s Dear CEO letter in particular.

“The Dear CEO letter was a big point where the market realised that this transition was actually happening and because of that there was a general move more towards Sonia than Libor,” he says.

“The regulatory tone in the UK market has increased a lot over the past 12 months and this most recent speech by Schooling Latter [on November 21] has continued to increase the pressure on market participants to move off Libor. The more that pressure and momentum builds – and the more things in the industry become clearer – then the more the end-user view that Libor is actually ending will harden, and more trading will be done in Sonia as a result,” he adds.

Failing additional regulatory guidance, Credit Suisse’s Sharp believes that regulation itself would pave the way for a greater Sonia boost – such as if leverage ratio add-on exemptions or reduced risk-weighted asset weightings for dealers trading Sonia were offered.

“I think that would be the single biggest boost to Sonia trading within the interdealer market and a way to incentivise early adoption,” says Sharp.

However, Lloyd disagrees that such an extreme measure from regulators is necessary given that the Sonia market is liquid and that the majority of clients are now looking to trade it. Rather than regulatory intervention in the form of capital relief, Lloyd expects regulatory scrutiny of market participants who continue to trade Libor to significantly increase instead.

As Libor nears its predicted death, users may have to reach the acceptance stage sooner rather than later.

Sonia and invoice spread trading

Invoice swap spread trading – a popular interdealer trade that typically involves buying a gilt future and paying fixed on a related interest rate swap with a similar risk profile, and vice versa – could also facilitate the Sonia shift.

As the majority of such interest rate swaps are currently pegged to Libor, traders expect that a switch to Sonia in the swap leg could be a key driver of Sonia becoming the market convention.

“Invoice spread trading is quite popular in the interdealer market and again there’s no reason why you’d choose to trade Libor over Sonia except for the fact that you want to manage your risk with whatever the more liquid instrument is, and right now that’s Libor,” says a rates trader at a US bank.

“There’s a bit of a chicken-and-egg problem here as liquidity doesn’t move from Libor because it’s the more liquid instrument to do your hedging, and because people use Libor for their hedging it stays as the most liquid instrument. That’s why this Q1 2020 goal is a step in the right direction as it will effectively focus minds to proactively transition to Sonia,” the trader continues.

Editing by Alex Krohn


Legal woes dent Wells Fargo’s earnings

Wells Fargo set aside $1.5 billion of legal reserves relating to its ‘ghost account’ scandal and other dubious sales practices in the fourth quarter of 2019. Operating losses over the previous two quarters, including these legal costs, have been the highest going back to Q4 2017.

Total operating losses for Q4 2019 were $1.9 billion, the same as in Q3. Legal provisions made up $1.6 billion of these losses in Q3 and $1.5 billion in Q4.

Prior to these last two quarters, the highest operating loss, of $3.5 billion, was disclosed for Q4 2017, which included litigation costs concerning mortgage-related regulatory investigations, shoddy sales practices and other consumer-facing scandals. 

The Q4 hit to earnings pushed net income down to $2.9 billion, a 38% decrease on the quarter and 53% on the year.


What is it?

Wells Fargo is alleged to have opened about two million deposit and credit card accounts without customers’ permission. In 2016, US regulators found the bank’s aggressive sales targets were to blame, which pushed employees to commit fraud. 

In February 2018, the US Federal Reserve imposed a cease-and-desist order on the bank in response to consumer abuse and risk management failings that led to the scandals. This prevents the bank growing its balance sheet above its end-2017 amount of $1.952 trillion. The bank's managers anticipate operating under the cap through 2020.

Why it matters

Wells Fargo is still paying the price for a series of mis-selling scandals, which have already cost the firm billions in losses and probably contributed to its operational risk capital requirement spiralling higher.

The bank expects the most troublesome of its punishments – the asset cap imposed by the Fed – to be lifted only when the regulator, and a third-party review, concludes a plan to enhance governance and operational risk management has been implemented correctly.

Until it has cleared this hurdle, the beleaguered lender cannot increase its assets, and therefore is at a disadvantage to its rivals in terms of gaining market share and generating the profits it needs to make up for the amount put aside as legal reserves.

Get in touch

Sign up to the Risk Quantum daily newsletter to receive the latest data insights.

Let us know your thoughts on our latest analysis. You can drop us a line at or send a tweet to @RiskQuantum.

Tell me more

Top 10 operational risk losses of 2019

Constrained by Fed cap, RWA density rises at Wells Fargo

View all bank stories


Giant £174bn Sonia swaps trading day may be biggest ever

Swaps trades referencing the Sterling Overnight Index Average (Sonia) soared to at least four year highs last week, according to data from the Depository Trust & Clearing Corporation’s (DTCC) trade repository.

The £174 billion ($226 billion) and £154 billion in notional traded on Thursday (January 9) and Friday (January 10), respectively, made the days both the largest single and largest consecutive trading days for Sonia-referencing swaps since at least the start of 2016, and possibly all-time for trades reported to the DTCC.

The swaps centred on upcoming announcements from the Bank of England’s Monetary Policy Committee with tenor structures consistent with trading around previous MPC announcement dates.


For instance, the two days saw £93 billion notional on swaps set to start on the MPC’s January 30 announcement and end on March 26, the MPC’s second announcement date of the year. Another £90 billion in notional starts on the March announcement and ends on the committee’s third 2020 announcement, on May 7.

The two days’ trading last week represents roughly one-sixth and one-fourth of notional traded for swaps, with tenors dated between the first and second and second and third announcement dates, respectively.


What is it?

Post-crisis regulation requires some swap trades to be reported to trade repositories. The DTCC maintains one such repository, which includes trades involving US-regulated counterparties.

The data therefore excludes firms not mandated to report to the DTCC and, in some cases, trade notional amounts are capped, further underreporting the size of total trading.

Still, the repository reflects one of the best publicly-available insights into the swaps market.

Each leg of the trades referenced above are sterling-denominated, meaning the data does not include notionals for cross-currency swaps.

Why it matters

Sonia swaps with tenors extending between MPC announcement dates have historically drawn trader interest. Such trades are likely put on by institutions betting on the direction of sterling interest rates.

The jumbo notionals reported last week may indicate increased trader comfort with Sonia, the risk-free rate chosen to replace the discredited Libor benchmark. This will please policy-makers, such as members of the sterling risk-free rate working group, who tabled a series of proposals to speed up adoption of Sonia before Libor is set to cease printing at the end of 2021, when the Financial Conduct Authority will no longer compel panel banks to submit quotes.

Get in touch

Sign up to the Risk Quantum daily newsletter to receive the latest data insights.

Let us know your thoughts on our latest analysis. You can drop us a line at, or send a tweet to @RiskQuantum.

Tell me more

Swaps data: have SOFR and Sonia swaps and futures lived up to expectations?

Sonia-Libor basis narrows after fallback verdict

FCA urges dealers to quote Sonia swaps on Clobs

View all derivatives stories


JP Morgan takes $2.7bn capital hit from CECL

Adoption of the Current Expected Credit Losses (CECL) accounting standard at JP Morgan caused loan-loss allowances to surge $4.3 billion (30%) on January 1, and will lop $2.7 billion off retained earnings over the next three years.

Allowances for credit card portfolios leapt the highest following the switch to the new regime, by $5.5 billion (97%) to $11.2 billion. Those for home lending and other consumer loan portfolios edged up $100 million each, to $2 billion and $1.4 billion, respectively.

In contrast, allowances for wholesale loans dropped $1.4 billion (26%) to $4 billion.


The net $4.3 billion hike to allowances has to be supplied out of Common Equity Tier 1 (CET1) capital, which is mostly made up of retained earnings at JP Morgan. But the bank reclassified a set of contra loans upon adoption of CECL, subtracting $800 million from the allowance increase. Tax effects reduced these additional allowances by a further $800 million, making the total hit to retained earnings $2.7 billion.

This reduction to CET1 capital will be phased in by 25% a year to 2023, in line with transitional arrangements put in place by US prudential regulators.

What is it?

The US Financial Accounting Standards Board introduced the CECL methodology in June 2016, to replace the incurred loss model. It replaced the existing standard as of January 1, 2020.

CECL obliges banks to record lifetime expected losses for all loans, instead of just those that have incurred actual losses as of the reporting date. These expected losses are estimated with reference to “reasonable and supportable” forecasts, as well as current and historical credit conditions.

The updated methodology also introduces a new accounting item, allowance for credit losses, to replace the current allowance for loans and leases. The latter covers credit losses only on loans and lease-financing receivables, whereas its replacement encompasses allowances for held-to-maturity debt and those related to off-balance-sheet exposures, among others.

Why it matters

JP Morgan’s chief risk officer, Ashley Bacon, raised concerns about the application of CECL to credit card loans in an interview with in 2017. He said that because these loans don’t have a contractual maturity, applying the concept of an expected lifetime loss, as mandated by the accounting standard, is difficult and could lead to a big build-up of reserves. 

His prediction was right on the money. It looks as though JP Morgan has erred on the conservative side when applying CECL, considering the 96% jump in allowances for card portfolios. Although the capital pain is limited and will be spread out over four years, the dynamic provisioning model introduced by the accounting standard means allowances will fluctuate from quarter to quarter, injecting a new element of volatility to banks’ capital planning.

Bacon said in 2017 that in light of this, capital requirements should be reviewed for loans particularly affected by CECL and presumably lowered.

If changes are not forthcoming, banks will probably be encouraged to hold back more retained earnings to furnish both required loan-loss allowances and CET1 capital, leaving less to be handed out to shareholders.

Get in touch

Sign up to the Risk Quantum daily newsletter to receive the latest data insights.

Let us know your thoughts on our latest analysis. You can drop us a line at or send a tweet to @RiskQuantum.

Tell me more

JP Morgan’s CRO on the bank’s six buckets of risk

CECL prompts loan sales, hunt for insurance

Citi raises CECL reserves estimate

View all bank stories


When a lapse in concentration is no bad thing

Fortifying too-big-to-fail firms against future crises could make entire system more vulnerable

Is safety found in size, or numbers? It’s a question worth asking in light of new datasets pointing to one inescapable conclusion: the stability of the global financial system depends on the health and vitality of just a handful of massive, interconnected firms.

Take the US repo market. In September, it plunged into chaos and order was only restored after the Federal Reserve pledged to lend out tens of billions of dollars for weeks on end. One reason for the debacle was the extreme concentration of excess reserves in a few bulge-bracket dealers: Citi, Bank of America, JP Morgan and Wells Fargo.

Then there’s the derivatives markets. Data published by a European watchdog found that just five financial institutions accounted for almost half the notional outstanding in credit and currency derivatives in the European Union in 2018. In the US, it’s even worse. The Office of the Comptroller of the Currency reports that JP Morgan, Citi and Goldman Sachs alone account for 57% of the $270 trillion of derivatives notionals held by the top 25 bank holding companies.

The concentration of essential activities in a tiny cadre of super-banks renders the entire financial system vulnerable if one of their number should fail. There is no firm, or even group of firms, that could step in to take the place of JP Morgan if it suddenly imploded. Yet the benefits of scale in wholesale banking and capital markets inevitably leads to greater concentration.

The Basel Committee’s systemic risk framework imposes penalties on colossal lenders to curb their growth, but not to shrink down. Research by the Bank for International Settlements found that global systemically important banks have simply expanded more slowly than other banks.

There is no firm, or even group of firms, that could step in to take the place of JP Morgan if it suddenly imploded

Instead of breaking the oligopoly at the summit of the financial system, regulatory reforms have forced these financial titans to ‘self-insure’ for their own destruction, by loading on capital requirements, bail-in debt quotas, stress tests, living-will requirements and more. The hope is that these banks, so crucial to the smooth running of the financial system, will be so well fortified that they’d be untouched by another global financial crisis.

Some worry this enshrines ‘too big to fail’ rather than ending it. Regulatory policy is increasingly formed around specific, systemic entities instead of in the interests of the system as a whole.

For example, because of its importance in US money markets, JP Morgan is de facto the world’s lender of next-to-last-resort (just ahead of the Fed). Following the repo crisis in September, the central bank took steps to replenish banks’ excess reserves. This does nothing to address the underlying, systemic problems in the money markets. Instead, by cementing the uneven distribution of reserves, it simply shunts responsibility for greasing money markets back onto JP Morgan.

Regulatory capture is the logical endgame of a financial market philosophy that hallows too-big-to-fail firms. To keep the financial markets humming, the very largest companies need to be taken care of. But the more they are, the more important they become to keeping the show on the road.

It will take brave policy-makers, regulators and politicians to break the cycle.


Swaps data: have SOFR/Sonia swaps/futures met the hype?

Progress on volumes of SOFR and Sonia swaps and futures

In just two years’ time, Libor could cease to exist, casting uncertainty over the fate of financial contracts worth hundreds of trillions of dollars. Alternative risk-free rates (RFRs) have been selected in core markets and have the backing of the official sector, but it’s not so clear whether they have industry backing.

Given the huge significance and reliance on Libor, there are always going to be voices casting doubt on whether these new RFRs will succeed and the start of a new year is a good point in time for such sceptical views to be aired.

Hard data on open interest, outstanding notional and volumes traded for swaps and futures referencing Libor successors such as SOFR, Sonia and €STR provide some perspective on progress to date. So far, it seems to be a mixed bag.

There’s no doubt RFR derivatives volumes have been increasing over the past year, as an eight-fold jump in SOFR futures open interest attests. But in some of the most established RFR markets – for example Sonia swaps – growth has stalled, particularly at the long end. And there’s little sign of a slowdown in Libor-linked swaps activity in tenors past January 2022 – the benchmark’s assumed end date.

As the clock counts down, the switch from Libor to RFR contracts may need to accelerate to embed those rates in the new landscape. The year ahead will be critical, with some pivotal events to watch out for – not least in October, when CME and LCH change their discounting curve from Fed funds to SOFR for all US dollar swaps.

SOFR futures

Let’s start with the largest currency, US dollars and futures on the SOFR index, and chart the growth in open interest over the past year for the CME and Ice futures.

Figure 1 shows:

To put this $2.1 trillion open interest into perspective, we need to compare with the CME Fed funds futures and CME Eurodollar to see whether the ratio of SOFR open interest to each of these is increasing.

The answer to that is an unequivocal yes.

Between January 31, 2019 and December 31, 2019, SOFR futures open interest has increased from 2% to 22% of CME Fed fund futures and from 2% to 19% of CME Eurodollar. Those are significant jumps indeed and augur well for the success of SOFR trading. We wait to see if CME’s launch of options on SOFR futures this month will add another fillip to volumes and open interest.

Sonia futures

Next, let’s look at Sonia futures open interest in 2019.

Figure 2 shows:

We can get a perspective on the £160 billion of Sonia futures open interest by comparing with Ice short sterling and checking if we see a similar increase as evidenced in SOFR.

Between January 31, 2019 and December 31, 2019, Sonia futures open interest has increased from 4.5% to 8.3% of Ice short sterling – so far less than the 20% we saw in SOFR, but evidence of a gain none the less.

SOFR swaps

Next, let’s look at progress in SOFR swaps trading.

Figure 3 shows:

If we follow our futures analogy, we should look to see if the ratio of SOFR swaps outstanding is increasing rapidly compared with Fed funds and Libor swaps.

In this case the jury is still out. As at December 31, 2019, the $345 billion of SOFR swaps notional represents just 3% of Fed fund swaps and 1% of US dollar Libor swaps. The latter ratio of 1% would look even worse if we tenor risk-adjusted the figures as SOFR volume remains dominated by short maturity trades.

It may be that we need to wait until October 2020, the month when CME and LCH change their discounting curve from Fed funds to SOFR, for any real pick-up in volumes.

Saron swaps 

Next let’s look at swaps in Swiss francs where Saron is the new risk-free rate.

Figure 4 shows:

And what if we compare Saron OIS swaps to Swiss franc Libor swaps and include volume at CME and Eurex?

On December 31, 2019, that gives us Sfr100 billion Saron to compare with Sfr1 trillion of Libor – a ratio of 10%, and one that has increased from 3.4%, so good progress there, but we need to see more long-term trading in Saron to make a material dent in a risk-adjusted measure.

€STR swaps

€STR is the new risk-free rate in euro markets and while Eonia is now changed to be determined from €STR plus a fixed spread, Eonia swaps continue to trade, but we are now starting to see €STR swaps as well.

I will skip the chart as the data is sparse but LCH SwapClear now has €66 billion ($73.1 billion) of single-sided notional outstanding in €STR swaps from a start in October 2019, while Eonia swaps remain massive with €8 trillion of outstanding notional and Euribor even larger at €25 trillion.

Sterling Libor and Sonia swaps

Finally, let’s turn to sterling, where the existing Sonia rate – after reform – has been selected as the new risk-free rate to replace Libor. The two questions we are most interested in are: first, is swaps volume in Libor decreasing, while Sonia swaps are increasing, and second, are longer tenors trading in Sonia?

Figure 5 shows:

So no evidence of a reduction in sterling Libor swaps trading and not even for those with maturity of greater than two years, which would continue past January 2022, the assumed end of Libor.

Let’s now chart Sonia swaps volume and, because the below two-years tenor dominates the figures to such an extent, to make the chart useful we will exclude this tenor, otherwise the monthly average of £1.9 trillion in below two years will dominate the total of £100 billion from all other tenors.

Figure 6 shows:

To summarise, 2019 saw excellent progress in SOFR futures volumes, some progress in SOFR swaps, Sonia futures, Saron and €STR swaps, but little in Sonia swaps. We need to wait and see what 2020 brings.

Amir Khwaja is chief executive of Clarus Financial Technology.