The Gateway to Algorithmic and Automated Trading

Fixed income on E

Published in Automated Trader Magazine Issue 36 Spring 2015

Word on the street is that fixed income is lock-step behind FX and equities in terms of algorithmic trading. But there are some major hurdles along the way. Automated Trader takes a look.

Alex Struc, Portfolio Manager, PIMCO Source, photo courtesy of AFME

Electronic trading is maturing in fixed income, providing the kind of increased liquidity and transparency essential to developing algos. It's also given rise to a cottage industry of initiatives geared to attracting the buy side onto platforms, and perhaps more controversially, into giving up their trade information.

Alex Struc, portfolio manager at PIMCO, who also sits on the firm's best execution committee, said that what's needed is firm order posting, which is what kicked off the equity market.

"The data providers who have firm access to the orders from the buy side community, and with the defensive and secure trading protocol will eventually win," he said.

Struc was speaking on a panel at the European Market Liquidity Conference, hosted by AFME (Association for Financial Markets in Europe), and struck a cautionary tone to the burgeoning platform provider environment at the moment.

"I don't see a space for two or three or five different models. I see a space for one that is combining all the others," he said, adding that another differentiator is about getting access to orders and not just trading protocol.

"Trading protocol we can figure out among ourselves. But it is people who have the captive audience, and so far we have two strongest, and three in total, providers who have access to investors orders, who are posting actively and saying I am ready to do that. Both platforms have been gaining critical mass," he said. PIMCO declined to provide further details.

Enter algos

Some commentators note that there are a lot of legacy practices in fixed income that suit the status quo. "You just have to look at the P&L of an investment bank over the last 20 years and you will see how much money they have made out of fixed income. And why do you want to change practices of a very profitable business?" said a senior executive from a financial services firm.

For now, the growth of algorithmic trading is restricted to certain products within the asset class, namely the liquid rates products such as listed derivatives - Treasury futures, short-term interest rate futures - as well as Treasuries.

Sassan Danesh, co-chair of the FIX Trading Community Global Fixed Income sub-committee, and managing partner at Etrading Software, said that one area of interest to watch over the next few years is the potential growth of swap futures, which lends itself to algo trading. The others are vanilla benchmark swaps and CDS indices.

Aside from the increased maturity and transparency, other factors boosting the trend is that it's getting easier to acquire datasets for backtesting, and many firms are deploying cross-asset algo trading platforms that they are then customising for the fixed income markets.

FI traders are now using or considering the use of algos because low touch algo trading is getting more attractive in an environment where increased costs from regulation, such as Basel III, results in continued margin compression.

Additionally, algo traders from equities, and futures and options now see an opportunity to leverage their existing investments in FI.

But there are yet obstacles. "Much of the market is still OTC and opaque, with the lit markets not providing sufficient depth or liquidity to provide a meaningful representation of actual prices for institutional size trades," said Danesh.

Also, the characteristics of the FI market is itself a barrier.

"The much larger and diverse universe of FI instruments, many of them illiquid and rarely traded, is less suited to the exchange/central matching paradigm that is a 'best fit' for algo trading," he said.

Danesh has been involved in bringing greater transparency and efficiency into the fixed income OTC markets by facilitating collaboration between the broker-dealer community and execution venues to define industry best practices using the FIX Protocol for trading OTC bonds and derivatives.

He noted that post-trade transparency is lacking in Europe in comparison to the US. That's because Europe doesn't have TRACE (Trade Reporting and Compliance Engine), which was introduced in 2002 in the US to increase price transparency in the US corporate debt market.

Intersections and paths

As a provider of fixed income and futures algorithms, Quantitative Brokers has a front row seat to the growth demand, and has focused on the intersection of the crossover between pure fixed income and exchange-listed.

The take up of algos in FI has moved along a predictable path, said Christian Hauff, chief executive and co-founder of Quantitative Brokers.

"We are obviously on the agency execution side of this fence. And in that world, we typically see software solutions as the first primers of algorithmic execution, which are not algorithms at all, they are actually software solutions," he said. "Often, you will first see the presence of software tools which have some level of automation but are not algorithmic - they are not dynamic, they are not based on statistical inputs, they are really just software rule sets."

Such tools have been in the cash Treasury markets for a few years, and that is the primer for vendor demand to increase, as well as the consciousness of the user base, which wants to further sophistication and transparency.

"With an agent or player like ourselves, you start to then get algorithmic execution in an agency form, plus you get transactional cost analysis on the back-end, which not only drives further improvements of the users' transactions, but their experience as well," Hauff said. "And it starts to evolve even further from there."

He added that the tangential markets of cash Treasuries, in other words, government bonds, was an "amenable and logical product" for the firm to expand into.

In the last three months, Quantitative Brokers became a FINRA (Financial Industry Regulatory Authority) registered broker-dealer to participate as an algorithmic agent for government securities, in addition to the NFA (National Futures Association) registered futures market place.

Christian Hauff, CEO, Quantitative Brokers

Complex and correlated

The evolution now is in multi-leg, relative-value execution, which involves the trading of a number of instruments simultaneously. Previously the trader had to execute such orders as separate orders or via static software spreading tools.

"It is one thing to invent or innovate algorithmic execution, say on an agency basis in fixed income for outright fixed income transactions," Hauff said. "But what is unique in our mind about the fixed income market is that it's highly complex and it's also highly correlated, and the trading activities often are not just in one particular benchmark, as is the case with relative value global macro. It is simultaneous executions of one benchmark versus another and that could be very customised."

In the multi-leg, user-defined spread space, Quantitative Brokers developed an algorithm "Legger", which allows for trading of any structure of cash Treasury legs versus any other cash Treasury legs, and/or futures legs or equivalents.

"This is going to be a big space and algorithms are best suited to solve that complexity," he said.

There are a number of factors behind the growth of algo trading, from Hauff's perspective.

Algo trading allows traders to have more accuracy, more automation and more scale, where previously they would have been "babysitting" the transactions using various electronic tools or through their broker.

But it's also a matter of mitigating the risk of information leakage in terms of pre-disclosure of orders over the phone to supporting and traditional brokers.

Ultimately, it is also a matter of execution quality.

"There is an underlying, core acknowledgement from the buy side, the liquidity taking universe, that the market makers have heavily evolved to a high degree of automation and sophistication," Hauff said. "And using a mouse and a finger to click and navigate is surely not the appropriate solution for them to transact against that liquidity."

"It is an unlevel playing field unless you are embracing algorithmic execution for at least some of your business if not all," he added.

Credit quant

As with Danesh, Hauff sees future potential in the credit space, which has been much harder for algos to make headway in up to now.

"Interest rates are far more concentrated; you have these quasi exchanges, fixed income platforms, from providers like BrokerTec and eSpeed that effectively provide electronic market pools for these benchmark products. This is less the case in the credit sector," said Hauff.

In rates, there are "on-the-runs", which are the most actively traded benchmarks and consequently lend themselves to pools of liquidity. "They are a baseline of activity that is very significant in those benchmarks," he said.

In the credit space, certain names can vary widely in activity and are therefore less liquid. But technology is the solution to overcoming those problems.

"I would point to options on futures as being a case study of how illiquid complex instruments can ultimately move electronically over time but it is a passage that does take time because there is a lot of market dependencies from participants through to systems, through to regulations that will have to contribute to that evolution over time," Hauff added. "Those agents are at play."

Regulation matters

Aside from the Dodd-Frank exemptions for Treasuries, regulatory bodies in the US focused on derivatives and wanted to ensure that the funding ability of national and local entities was not impacted by reforms in response to the credit crisis.

The fact that US Treasuries had been processed centrally via FICC (Fixed Income Clearing Corporation) for a generation gave US regulators confidence, said Jeffrey Hogan, managing director of business development at BGC Partners, a global inter-dealer brokerage firm.

But in Europe, the regulatory mandate differed.

"The European Commission and ESMA were tasked not to reform certain market segments, but rather to expand MiFID I, from a narrow focus on equity products only, to encompass all financial instruments," Hogan said.

This is beneficial for the HFT community, where specific conditions and privileges for algo trading are enshrined in the Level 2 language.

"Clear and consistent open access terms, explicit and consistent market maker requirements, and universal pre- and post-trade transparency provisions across regulated venues combine to pave the way for increased participation for these firms across the European financial marketplace," he said.

HFT outlook

BGC's chief information officer, Eric Hirschhorn, said that the large majority of the bond market was in illiquid, corporate bonds, and this was a highly unfavourable area for HFT operation.

But it is favourable to a technological solution for the problem of finding bonds for trade and matching them with dealers.

Eric Hirschhorn, CIO, BGC Partners

"If you think about the ultimate high frequency markets in futures and equities, it translated pretty well into foreign exchange. In some cases it translated reasonably well into on-the- run Treasuries," he said. "When people try to put that to the credit market, it was a complete failure because that market is not deep enough or fast enough."

Hogan believes that it is more likely that HFT will direct flow to swaps rather than search for new income opportunities.

"What is probably more interesting over time is the algo companies looking to expand their repertoire beyond fixed income into derivatives markets," said Hirshhorn.

Looking over the coming year, Hogan expects several developments; more fully electronic trading by non-banks; implementation of capital controls will induce banks to hedge risk electronically earlier in the trade cycle, and regulatory moves outside the US based on substitutive compliance and regulatory equivalence will assist US- based firms to execute more heavily in overseas markets.

Bids and offers

There are more than a few companies bidding to bring an electronic dimension to the fixed income market, and seeking a technology solution to the problem of opening up the vast acres of illiquid bonds.

Bondcube, backed by Deutsche Boerse, is one of them.

Earlier this year, Bondcube became a FINRA broker-dealer member. This follows its approval as a crossing network throughout Europe. It's now set to apply to the SEC to become an ATS.

The idea is to take indications of investor interest and then match buyers with sellers, with participants loading up IOIs (indications of interest) along with security, direction and minimum and maximum trading size.

These IOIs remain dark until a match is found, and only participants who have the offsetting interest in that security are notified and eligible to begin a negotiation for a trade.

Paul Reynolds, CEO, Bondcube

Paul Reynolds, CEO of Bondcube, said that the market "desperately" needs new functionality to catch up with other asset classes.

Reynolds has an extensive history in the evolution of FI electronic trading systems, and his career highlights include setting up Deutsche Bank's Autobahn platform and introducing Bloomberg firm pricing at UBS. Most recently he introduced a fixed income trading platform at Amias Berman that allowed the buy side to put firm bids and offers on Bloomberg's ALLQ fixed income trading platform.

"Our belief is that the old infrastructure and the paradigm of investors asking banks for prices is functional but it is not going to grow this market," Reynolds said. "Taking all the liquidity of all the investors in the market and saying: network and make it matchable and tradable. That brings massive additional volume."

Initiatives like that make it easier for people who are dissuaded from trying to automate trading in fixed income, because the liquidity finds you, he added.

That's not to say it won't be a challenge. And algo trading would be required to access currently hard-to-access bonds and instruments and make them more liquid.

"I would not put any limit on what you can do with algo trading. Algos are written by very clever people to exploit arbitrage, and the nice thing about fixed income is that there is a lot of arbitrage to be done," he said.

That includes four sub-sectors with a lot of arbitrage potential: CDS (credit default swaps), indices, ETFs and underlying cash bonds. Knowing where the liquidity and traded volume is, and how to access it, is an environment where algorithms can thrive.

"Because no one else is doing it and the infrastructure is not quite there yet, it is one of those markets where the alpha has not been discovered and exploited," Reynolds said.

Notes from AFME's European Liquidity Conference

Alex Struc, Portfolio Manager, PIMCO Source, photo courtesy of AFME

Don't complain about liquidity in the fixed income markets around Alex Struc, portfolio manager for PIMCO. First of all, he'll tell you that "liquidity will always disappoint". The reality is, there's never enough of a good trade, and there is always too much of a bad trade.

Perfect expectations just aren't part of the plan.

Next, he'll let you know that fixed income investors may complain about liquidity and yet, when it comes down to using platforms, "people are trying to game the system".

"The only choice is to trade and to trade at a good price and for me it is a good social experiment. If we fail at that, I don't think we have a right to complain about liquidity," he said.

In terms of platforms, PIMCO is in favour of all-to-all. At the same time, Struc noted that the firm would be in favour of a dark pool where only price matters.

"Size and reading into what people are trying to do, and try to follow, is not really what we are interested in," he said.

Best Execution

He's equally straightforward about the concept of best execution. The debate in terms of fair price versus best price, said Struc, is a "conundrum" that the sell side sells to the buy side.

"There is only best price. There is no such thing as a price for the size, there is only best price and if the prices moves, so will the expectations and so the views on valuation," he said. "Best execution is leaving no better price than you traded behind."

PIMCO's approach to trading, he added, is to not dwell on whether orders are big or small, but to find the cheapest to deliver.

"On a stack of 10 bonds we will never go for the second cheapest because 200 million is available. We would always test the first bond, even if it's 500,000 available," he said. "There is no such thing as little money for us. The reason for that is if you go for second cheapest you may have missed the big elephant sitting in a number one that you haven't observed."

On the flip side of that, it doesn't do to slavishly fill an entire order either.

"If we have a view of buying something of 100 million and we are 60 million through that trade, and the price of trade has moved to where it no longer becomes interesting, we are not obsessed about buying 100 million because there so many bonds outstanding, there are so many opportunities outstanding," he said.

Because of this mentality, Struc added, PIMCO has become something of a balance sheet, in stark contrast to the situation for banks since the financial crisis. And PIMCO can fulfil that role for both buy and sell.

"It will be foolish for us not to monetise on this by providing valuation and price, so we do. In exchange for that, we demand price transparency," he said. "We are not prepared to tell the rest of the world for free what it is exactly that we are doing. So if you want to be part of that club, and I think it is not unique to us, it is unique to asset managers, then as dealers we invite people for price transparency."

Pre-trade transparency is about integrity of pricing, and Struc said that PIMCO tries to promote the notion of price guidance. After all, price signals from market makers turn into valuations, sometimes of entire economies.

"With great power comes great responsibility and I think there is very little regulation and acknowledgement of the fact that this price guidance by the people who have the banking licence and market making licence is a bit of a free-for-all," Struc said. "The home run over the next two years will be putting some rules in place with respect to what price guidance should be."

Dark pools

Mark Pumfrey, head of Liquidnet EMEA

Automated Trader talks to Mark Pumfrey, head of Liquidnet EMEA. After silencing critics over 10 years ago with the success of its equities dark pool, Liquidnet is setting its sights on fixed income. The firm purchased bond trading platform Vega-Chi and is working on how to solve the technological and cultural hurdles before launching.

AT: Why did you go with Vega-Chi?

Mark Pumfrey: Our members are the big institutions globally, 75% of the world's bonds are owned by our members. Post-financial crisis, we have been asked again and again by our members: "You understand the solutions in equities, can you offer a similar model for fixed income?"

We didn't have to build something from scratch as Vega-Chi became available. What it didn't have is a dark book, which is a key ingredient to success - because to build a successful liquidity platform for fixed income you need to aggregate the liquidity with no information leakage.

AT: You are fully in the corporate bond space?

MP: We will start in the corporate bond space but there are opportunities in sovereign bonds.

The issue really is, can you aggregate liquidity on the buy side to give crossing opportunities? That means a very significant change in behaviour. If we were sitting here in 1996/97 we would have been having the same conversation about equities, and there would have been the same skepticism about the ability ever to aggregate liquidity the way we have in equities because it would never work, nobody is ever going to share their order book.

What I think will happen is there will be a liquidity crisis at some point and when that happens the problems will be so apparent that everyone will realise they have to change their behaviour. In the meantime I think we can make some decent progress.