Pre-trade information is the key. Execution itself is a secondary consideration.
“The noise around market structure change is becoming much more concentrated now around quality of information and point-to-point connectivity,” says Dominic Holland, director of credit e-commerce at Deutsche Bank. “The old model was about passing as much information as possible to as many people as possible in the hope of discovering the other side of the trade. But market participants now see the need to become much more focused.”
Deutsche is far from the only sell-side firm encouraging buy-side solutions to the liquidity conundrum. “Ultimately, the winning model in all this will be the one that gets the buy-side’s inventory moving,” says Niall Cameron, head of EMEA markets at HSBC. “Enticing that to happen just by offering to show more dealers’ axes has a limit. It's been estimated that the dealers collectively hold approximately 1% of the bond market now. What’s needed is something that further coaxes out the buy-side’s cares and interests and makes those big holders of the 99% of the inventory more likely to move some of it. This is now the cutting edge.”
Algomi is right on that cutting edge. Set up in 2012 by three founders of the UBS Price Improvement Network (PIN), Algomi, as Euromoney reported in May 2014, casts itself as the anti e-trading alternative for bond dealers and investors.
Its first product, Synchronicity, was for dealer banks and comprised technology to capture all the stray data flowing across a dealing room – on previous filled and unfilled investor queries in certain bonds, on completed trades, on portfolio positions after a new issue. New enquiries and orders might then be properly collated and channeled to the right salespeople, and give them a better chance of matching up customers pinging in requests to the bank from anywhere in the world.
When Euromoney sits down with Algomi chief executive Stu Taylor, the firm is just two months into full launch of its second offering, Honeycomb, which this time is targeted squarely at the buy side.
|You have to recognize that for clients wanting to do large size in today’s illiquid markets, even merely enquiring about a bond can move the market price against the investor|
Honeycomb does something quite remarkable. It allows an investor to set up an encrypted listening station inside a banks’ bond dealing room – with the bank’s permission of course. Without the bank even knowing, the investor can interrogate that bank’s systems about certain bonds, as it seeks to identify the right bank to approach to get an order filled.
“We are an information network concentrating on illiquid, non-benchmark bonds in which investors want to do larger ticket sizes than typically flow across e-channels,” says Taylor.
|Algomi chief executive|
“You have to recognize that for clients wanting to do large size in today’s illiquid markets, even merely enquiring about a bond can move the market price against the investor. Investors used to check prices with five banks in the days when the banks all used to show off about their risk-taking capacity. But today those five banks, instead of taking the order down on their books, will immediately start contacting selected other clients to try and set up the opposite side of the trade, and so the information leakage is both widespread and immediate. That’s damaging in thin markets even for benchmark bonds and potentially very damaging for the overwhelming majority of infrequently traded securities. If it has a large trade to do, the client would be much better off taking it to just one bank and maybe splitting it between principal and agency. But the question becomes: which one bank?”
That’s the question Honeycomb seeks to answer.
Taylor runs Euromoney through a quick demo. The buy-side user of Honeycomb – it could be any large institutional asset manager – wants to buy $27.6 million of Intel bonds maturing in 2022. The Honeycomb system shows up a heat map for all the dealer banks that allow the asset manager to access their systems. It may show that BNP Paribas has traded these same Intel bonds five days ago. It may show that Credit Suisse has a current axe on the bonds to trade for its own account. But it shows that Deutsche Bank also has a current axe to trade on its own account, has traded the bonds in the past three days, has some portfolio knowledge of which accounts may hold the bonds, has recently traded in similar bonds, perhaps of the same issuer but a different maturity, has a new research call on the name and has a hot live investor query to deal right now.
|Honeycomb shows an investor buying Intel bonds banks that have traded recently and know where to look|
The system homes in on Deutsche Bank. Up to this point, no one aside from the asset manager’s own portfolio managers and dealing desk knows that it is even looking to do anything in Intel at all. Now the asset manager hits “reveal” and opens up a chat with Deutsche Bank – and only Deutsche Bank – about the Intel 22s. Almost as a reward for showing its hand to a single bank, the asset manager now gets back more detailed information about what Deutsche has done or seen recently in these bonds. It shows that it did indeed cross a buy order for a client for $4 million of bonds three days ago. It has an axe to sell $3 million itself and a live sell order from another client. But the mother load is that it has portfolio knowledge of where $478 million of the bonds now sit, perhaps from having put buyers into them in primary or filling other secondary trades.
The asset manager starts to chat with the Deutsche Bank sales people, perhaps initially revealing that it wants to buy $5 million, but holding back information about the whole order to see what benchmark price Deutsche can set on this first portion. “Algomi works really well with voice coverage,” says Taylor. “At this point the Deutsche salesperson, having done $5 million, will ask if the client wants to do anymore. The client might then say: ‘I’ve another $22.6 million to do and I’ll leave you the order to work, within these price limits, for the next hour'.”
“And by the way,” adds Taylor, “if the salesperson is worth their salt, this is the time to ask the client: ‘While I have you on the phone, do you still own those GE 17s we sold you three weeks ago because we currently have strong investor interest on the other side?’”
It’s clever stuff. Of course banks don’t want to reveal this kind of information to investors, which they don’t even know might be interrogating their systems. But at least the banks can entirely control which clients have this access. And maybe granting it is the cost of being in the bond business now, as it transforms into a largely agency one.
Banks all now claim to have superior information on where bonds sit, and which holders might be motivated to buy or sell at what price. Algomi’s Honeycomb elevates all that from a salesperson boasting to the client over lunch – and waving the bank’s position in the new issue league tables – to a far more formalized version. Algomi validates all the data on previous trading history, current axes, portfolio knowledge and the rest against the banks’ own internal systems.
Algomi is just one example of smart people who understand the intricacies of the bond markets looking to bring liquidity where it has ceased to exist, or perhaps never existed. In January, MTS, the bond trading platform that is part of the London Stock Exchange Group, announced a new tie up with B2Scan, an aggregator and search engine of bank bond inventory, runs and axes that aims to help fixed-income investors identify the bank most likely to help them to buy or sell a particular instrument by sourcing the other side of the trade.
It’s a tie up that encapsulates how the secondary bond-market structure is now fast evolving in response to vastly diminished dealer willingness and ability to warehouse risk. Bond investors realize that the days of dealer banks happily taking down as principal the other side of their $100 million order have passed, never to return. It’s not just because regulators have insisted on banks holding much higher capital against market risk and counterparty credit exposure on the kind of disguised proprietary carry trading banks used to love. Also the single-name credit default swap market that dealers once used to hedge out credit risk is now a mere shadow of its pre-crisis size.
Some investors had hoped that new exchange-like trading venues for bonds, with central limit order books, might restore liquidity. Regulators have quietly encouraged these as vehicles for greater post-trade price transparency in the over-the-counter bond markets and, hopefully, for controlled risk through central counterparties.
Also in January, MarketAxess launched its all-to-all Open Trading platform in Europe. This enables bond investors to source liquidity from all other system participants in a single market place and brings MarketAxess’s alliance with BlackRock’s ground-breaking Aladdin community, which crosses trades between asset managers, to Europe.
Rick McVey, chairman and chief executive of MarketAxess, pointedly linked the potential of all-to-all trading to improve bond market liquidity to recent heightened volatility and rising investor fears over price gapping. McVey notes that over 7% of US high-grade trades on MarketAxess now take place via Open Trading protocols, compared to just 2% a year ago, an acceleration of take-up that has coincided with rising volatility.
Frédéric Semour, managing director and founder of B2Scan, is a bond market veteran who learned the business during 20 years on the sell-side at Salomon Brothers and Credit Suisse before moving to the buy-side at Natixis Asset Management, where he learned the difficulties of bond investors at first hand.
“Liquidity is the big issue for fixed-income investors and it’s difficult. Say that your portfolio managers want to put on a position in bonds of a big German corporate. The instinct of the dealing desk might be to go first to the German banks that are closest to the company and may have been on its new issue syndicates. But by now, those German banks might well be short. It could be the Spanish or Italian banks that are long and so might take the other side, but how can you possibly know that?”
At Natixis Asset Management Semour started to ask relationship banks to send in lists of all their inventory, axes and indications of interest so that his dealers could look through for where they wanted to trade. He says: “It helped the banks because we weren’t calling them up over bonds they didn’t own or had no insight where to find.”
Semour saw the idea for a new business back in 2011 and left Natixis to build B2SCAN as an extension of this approach. Of course, banks are wary of simply sending inventory to any but favoured clients, and certainly don’t want to broadcast it openly to third parties.
“It took time to get the banks on board, but HSBC and Bank of America Merrill Lynch were supportive. Then Citi came on board,” Semour says. “Banks have realized the need to communicate their axes and inventory better, and we do two things the banks like: we give them control over which buy-side clients get access to the data and we undertake not to re-sell it".
B2SCAN has built its buy-side client list among real money asset managers, notably in France and Scandinavia, not among hedge funds or interdealer brokers. Brett Chappell, global head of trading at Nordea Investment Managers, welcomes the latest step, saying: "The partnership between MTS and B2SCAN will further increase efficiency for buy-side traders by dramatically cutting down the time it takes them to identify a counterparty for the credit products they want to trade.”
| The beauty of this tie-up is that buy-side traders are usually agnostic on where they trade. So once they’ve done all the hard work of actually finding the other side of the trade through our search engine, the execution part is easy|
It becomes apparent from talking to banks and investors that use B2SCAN that those buy-side clients have in turn asked more sell-side dealers to post their axes and inventory on the system. Power is shifting in the bond markets. Investors dominate the inventory as never before, and so when large asset managers ask dealers to join such a platform, the banks tend to listen. From just two to start with, B2SCAN now has a dozen banks on board, also including Rabobank, Santander, Commerzbank, UBS, Société Générale, Danske Bank, UniCredit, Natixis and BBVA. Some of the big dealers, notably among American banks, are still missing. If B2SCAN can sign up more big asset managers then the other big dealers may be more likely to join too.
Semour knows Simon Linwood, head of credit at MTS, from their days together at Credit Suisse, and when the two looked at each other’s offerings last year, the fit seemed obvious. B2SCAN offers MTS an intriguing piece of content at the cutting edge of innovation in the crucial pre-trade area for the buy side. Linwood says: “B2SCAN has the dealers, it has the data, and in a bond market with lots of new start-ups, it is market tested and has been up and running for three years.”
“Trading platforms are generally good on execution but limited on premium content,” observes one sell-side source. “But the buy side’s biggest challenge is accessing the quality information that might lead to the other side of the trade. We need to give the buy side tools to engage in conversations with relevant parties before moving to risk transfer. That’s why MTS and B2SCAN makes great sense.”
Once an investor has worked out the best dealer to approach and has firmed an order up, execution is the simply the final step, and putting B2SCAN on MTS’s BondVision platform provides an automatic link to a regulated market place, where trades can be struck in full compliance with best execution and disclosure rules. The execution component may be a bit of a commodity, but MTS is one of the big platforms along with MarketAxess, Bloomberg and Tradeweb. Its heft also offers the chance to bring more buy-side clients onto B2SCAN from elsewhere in Europe where it is not so well known, such as among UK asset managers.
Semour says: “The beauty of this tie-up is that buy-side traders are usually agnostic on where they trade. So once they’ve done all the hard work of actually finding the other side of the trade through our search engine, the execution part is easy. We have said to buy-side and sell-side clients: ‘We can bundle this service with trading execution via BondVision thanks to the agreement with MTS’".
The bigger question now becomes: can aggregators of axes and inventory and search engines such as B2SCAN encourage the buy side to bring more of their own inventory out into the shrunken secondary bond market and so improve turnover and liquidity?
There are encouraging noises. Laurent Albert, head of global trading at Natixis Asset Management Finance, comments: “As credit liquidity has grown scarce in recent years, the market has looked to technology providers such as MTS and B2SCAN to deliver the innovation required to tackle this very complex challenge. Enabling the buy-side community to search sell-side axes will open up a whole new pool of trading opportunities, significantly increasing our credit trading activity across the board.”
|Niall Cameron, head of EMEA markets at HSBC. says: 'It's been estimated that the dealers collectively hold approximately 1% of the bond market now. What’s needed is something that makes those big holders of the 99% of the inventory more likely to move some of it'|
The landscape is shifting fast.
Cameron at HSBC was one of the early backers of B2SCAN and approves the tie-up with MTS. He says: “Frédéric Semour was one of the early thinkers who spotted that the solution to lack of liquidity is not so much about trading. It’s about getting the data sets for bonds that might be available to trade to connect to each other more efficiently. This is another move in the right direction. Several dealers are likely to put their axes on it and I’m sure the buy side will look at them.”
As head of markets, Cameron, like his peers at many of the big bond dealing banks, is moving towards an agency model, which is already openly declared at many of the second-tier banks, though some bigger bond dealers aren’t so ready to call it what it is.
“I call what we do now risk-assisted crossing,” says Cameron. “If a client wants to sell $30 million of a bond for example, and we can develop bids on the other side from clients wanting to buy $20 million, then we’ll do the last $10 million as principal.”
That’s a big change from a few years ago. “There was a time when the Street would've taken down the whole $30 million itself. But instead of competing on risk-taking capacity we have to prove our value in information about bonds, knowledge of where the other side of the trade might be, and reach to that other side.”
New trading niches
Systems such as B2Scan, Algomi and Neptune potentially allow for banks to build expertise in new trading niches and be rewarded for it.
Taylor says: “Trading heads change at banks, business strategies change and banks may seek to carve positions in sales and trading in new niches of the bond market. Your bank may not have lead many of the tier-2 bank deals in sterling over the past 10 years, but if you’re actually building up knowledge and expertise and doing more secondary market turnover now, this system will show that in real time to clients that you approve to have access to it.”
Let’s go back to the Intel 22s. Say the full $22.6 million order goes ahead. It’s possible that other buy-side clients have quietly been listening, observing, waiting for action in those bonds. Now, when it all suddenly heats up, may be the moment they declare themselves. After a period when there was minimal activity in the Intel bonds with no one on the buy or sell sides wanting to show their hand, the action suddenly visible at Deutsche Bank may tease others out and induce them to move some of their inventory.
Algomi depends on smart and expensive technology, which is time-consuming to hook new customers up to. It is in a multi-year build out. It certainly looks well funded. Its key disclosed equity investor is Lakestar, the technology venture capital fund that backed Spotify, Skype and Facebook and which no doubt sees the troubled secondary bond market as a potentially large one for network technology to apply across.
Algomi has nine banks taking its service, with Deutsche Bank and Credit Suisse the two previously disclosed but other leading firms known to be on board. Algomi’s London offices in One America Square are buzzing. The firm is expanding in the US and has just hired a new head for Asia.
As Euromoney sits down with Taylor on the soft furnishings of a central breakout area to take in the demo on three large screens excited staff interrupt to announce two knew buy-side firms have signed up. Is this all stage-managed and pre-arranged? Euromoney wouldn’t be so cynical as even to suggest it. This is not a bank we’re visiting, after all.
Taylor says 30 large buy-side clients have contracted and installed the system and advanced contract discussions are underway with another 30 or 40 more.
If Algomi makes serious headway it will challenge many of the providers of new technology solutions to the problem of bond illiquidity. Encrypted areas inside banks’ trading systems, allowing discreet order formation with minimal market disclosure, are the very opposite of open all-to-all markets with fully visible – even if anonymous – pre-trade order formation and post-trade price transparency and transaction cost analysis.
|Oasis: Definitely, maybe, says Deutsche Bank|
|Dominic Holland, director of credit e-commerce at Deutsche Bank, is approaching his 30th anniversary of working in the European credit markets, and what he sees today reminds him of the market when he started. |
“In the 1980s, rather like today, banks had much smaller amounts of capital to put to work in bonds,” he says. “You might take a client order to sell $4 million of bonds, try and cross $3 million with other customers and take down just the last $1 million yourself.”
After two decades of expansion in size and risk taking before the shock of the financial crisis, it is tough for market participants to adjust back to that.
But what of Oasis, the project Deutsche Bank has supposedly been working on for a number of years now to develop a multi-dealer platform to be managed and operated by a third-party vendor as a matching engine, with buy-side clients going through sponsoring banks to post indications of interest to trade on an anonymous basis?
Euromoney reported on this in September 2013 when it seemed close to launch, but since then – nothing. “The concept of Oasis is still something we’re very committed to,” says Holland. “Algomi’s Honeycomb works well and will be the story of the next six months but it still involves a lot of verbal communication in a very disclosed manner between bank and client. The next evolution may be towards a more discreet matching engine where clients leave resting orders anonymously and will only be alerted when the algorithm finds the exact or closest match to the opposite side of their order. At that point it will be settled through the two clients’ introducing brokers.”
Oasis has been a concept for so long that doubts about whether it will ever see the light of day are now widespread. Project Neptune was a big distraction last year and Honeycomb is occupying people this year. Some of the early movers behind Oasis are working on dealer-to-dealer matching engines of their own.
Taylor says: “It’s important to note that Algomi is not a hub. Our view is that neither banks nor investors want to load indications into the cloud, or onto some third-party-run trading venue, because the danger of information leakage is so great in this market. You might get your trade done at an apparently tight spread or commission. But that’s not much use if you’ve already moved the price several full points against yourself. This only works if it is not on a hub, and I would argue that new solutions built as trading hubs are being built on the wrong architecture.”
And the validation for this assertion is the size of some of the interest and orders already coming across the Honeycomb system. Taylor says: “We had expected investors to use it for large size orders, but they have been larger even than we thought, averaging over $20 million, with a few of $100 million. We are serving a very different market from e-trading that focuses on small lots across the well known channels.”
If asset managers are to put more big, portfolio rebalancing trades through venues such as Honeycomb, while automating smaller orders over e-trading venues, that leaves open the question of how they prove best execution for clients. It may seem obvious that the benefit for an asset manager’s clients in its keeping a big trade quiet and so not moving the market price far outweighs any savings on spread or commission. But the discreet concentration of large volume into few hands with minimal open price discovery raises questions at least of appearance, if not substance.
Taylor takes a crack at this. “So in our example, the client doesn’t go to five different banks it goes to one. How does the asset manager’s dealing desk justify this to compliance? I can tell you that often buy-side dealers do this kind of thing with a fairly subjective two or three sentence written report. ‘That bank has always been the best for Intel bonds.’ Honeycomb lends some actual substance and justification. ‘We went to that bank because we could see that it had recently traded the bonds in size, had substantial portfolio knowledge, a current axe of its own and incoming client orders.’”
Is that enough, though?
Taylor has another go. “In today’s market, best execution is a bit of a misnomer. The current rule of thumb is that you need to have checked prices with at least three banks. But out of the 80 odd banks dealing bonds, that’s a small fraction of the market. We have had lots of conversations with trade bodies and regulators about this, and what they all want to see is a systematic approach to best execution. That’s what the three-source price check stands for. I would say that Honeycomb is thoroughly systematic. The whole point of it is to drive the trade not just to any three banks that might have an historic market franchise and be showing a price but rather to the banks that are truly most active in and knowledgeable about the bonds in question.”
The buy side seems to get this. Banks like it because they are still at the middle of the action with Honeycomb. Decisively winning the argument over best execution with regulators requires convincing them that large-ticket trading in non-benchmark bonds is about as far away from trading equity as it is possible to go in the securities markets.
Most banks' vision for the future of the business rests on building networks of connections to many venues, including their own proprietary ones. But banks’ senior managements are tight on costs, while shrinking revenues have taken the shine off FICC as a driver of profits. Dealers have to question the costs of supporting many market initiatives. Does it replicate what is already being developed elsewhere? Does it, perhaps, only solve what is fast becoming yesterday’s problem?
Holland at Deutsche Bank says: “Post-financial crisis the request-for-quote model is challenged because dealers are not taking down as much risk as principal. Any negative impact on indicative pricing that results is not malicious. Banks are genuinely trying to locate the other side of the trade in their client’s interest but they need some sharper institutional memory of where to go to find it.”