Putting price discovery first

A proponent of introducing anonymity to exchanges, Mark Northwood, head of trading at Fidelity International in Hong Kong, spoke to The TRADE Asia about the impact of crossing engines in the region and whether this is likely to result in more efficient price discovery or merely serve to fragment liquidity.

Given the disparate nature of the markets in Asia and the heightened challenges associated with front running and information leakage, are the PMs understanding of some of the more endemic obstacles that traders can encounter?
There are some known hot spots in Asia. You often find that PMs have varying levels of experience. Many have come to the job having previously been analysts. If you are new to the game a key lesson you need to learn is that there is price impact associated with trading decisions and the role of the trading desk is to try and minimise that. We work closely with the PMs to determine what the best strategy and timing might be for what they want to achieve.

Do you ever dissuade a PM from entering a particular line of stock because, for example, the likelihood of price movement is high? It has happened. We have kept PMs out of a stock for a number of reasons, the likelihood of big price impact on their alpha being one of them. There will also be situations where the PMs conviction is high and their view is multi-year. If an investment is deemed likely to quadruple over two years, a 10% impact cost on entry appears largely irrelevant. These factors may not always be obvious upfront. What may happen, then, is that we will try to buy something and it quickly becomes apparent that it will be difficult so we stop. Positions are often easier to enter than get out of, because there might be a liquidity event, which allows the position to be secured as a result of a company raising capital via a placement.

Are you particularly concerned about buying into a line of stock that you know is going to be difficult to get out of?
Anything that impacts the free flow of liquidity or the dynamics of how a stock will trade has to be included in the decision of whether or not to invest in that stock. Institutions should have internal processes to monitor liquidity risk, particularly where the free float is tight or in names with substantial holdings.

Broker internal crossing engines and external networks such as Liquidnet are either already available or are in the process of setting up in a growing roster of Asian markets. Is this helping you in the search for liquidity?
In my view, some of the tools and venues that are being touted as offering new liquidity, do not. They fragment the existing liquidity. A crossing network doesn’t necessarily add liquidity. It may make some investors feel more comfortable to show more of their hand at the margin, but I would argue that the best way to promote liquidity is to centralise it and introduce market structures that protect investors who do show their hand. I’m a big fan of anonymity on exchanges to help with this, and it has been gradually spreading through the region but is certainly not universal. The delayed reporting offered by the ASX completes my wish-list by enabling scrutiny of brokers’ activity after the fact.

Asian exchanges are evolving in a very dynamic period and there are situations where the technology and processes available on the exchange may not be as good as those available through a third-party venue, particularly one that is bringing the technology from the US or Europe.

The interesting question is whether the buy-side is going to be able to keep up with these developments and, indeed, other investors who help generate the liquidity we all live on.

Why would the buy-side not be able to keep up with the pace of development?
There is an investment in technology required on our side just to gain access to fragmenting markets. Typically, that means taking advantage of a product or service that is being offered by another firm. That’s certainly an easier route to go down than to try and build something yourself, but it has other implications. We are all highly sensitive to the issue of information security and any captive conduit of order information is something we are very careful about using. There are also workflow factors which impede a large buy-side’s ability to just plug in new tools.

Internal crossing is being presented as a means to reduce market impact and achieve price improvement. A less charitable view however, is that crossing mechanisms are designed solely to capture liquidity rather than aid price discovery. What’s your take on it?
Exactly that. Seeing flow has a number of benefits; everything from your league table ranking, to visibility of what certain clients might be up to, through to the ability to cross stock and generate commission on the other side. If you think back to where exchanges came from, they used to be owned by brokers. Many exchanges have since become listed companies and firms like Fidelity can now invest in them. With that step, brokers effectively lost control over the process. What I think is happening now, is that brokers are trying to re- establish ownership of the process, and with it the flow, and creating their own mini-exchanges to do that.

Brokers have different ways to attract flow to their crossing engines. Obviously, first mover advantage is critical. What we are seeing in this part of the world is that the brokers who were early with algorithms are the ones who stand to benefit from rolling out crossing engines. The concept is a sound one of course. If a crossing opportunity doesn’t exist, the client flow that sits in a broker’s crossing engine is going to make its way to the market. But this impedes the primary function of the centralised order book, by removing some buyers and sellers from the price formation process. So, in my view, internal crossing makes the overall market a little less efficient. It might benefit specific investors on individual trades, but ultimately I don’t think it serves the central aim of regulated markets – achieving the most efficient price (and volume) discovery.

If the exchanges recognise what the competition is up to and pre-empt them, they could probably steal a march on this. But they clearly have other considerations. The exchanges in this part of the world have historically been more concerned with the needs and the protection of retail investors, who, if you go back 10 years, constituted the bulk of the market turnover. But the institutionalisation of savings in Asia is a huge phenomenon. It’s the reason why firms such as ours are here. And that has changed the nature of trading significantly over the last decade.

Two internal liquidity models seem to be emerging; one based on exclusivity, the other offering links to other dark books. Do you have a preference?
If the exchanges don’t respond, then we’re happy to have competing products available. The open model is the one we favour; otherwise we would have to find a way to check every broker’s exclusive liquidity pool. That’s clearly not a proposition we’d choose. Firstly, there’s the issue of signalling trade intentions. Secondly, you will potentially have very small trades with multiple counterparties; the cost of doing that is too high. So, to use an aggregator linking pools is a sensible option and one that we will likely favour.

Brokers’ internal crossing engines tend to be fuelled by institutional order flow. Would you like to see retail flow brought into play?
Retail flow, particularly in small cap stocks, can be useful to larger institutions. But I would certainly differentiate good flow from bad flow. There is a whole part of the market in Asia that exists by feeding off other people’s business and anything that facilitates that process is not good for us.

The type of liquidity we like is in blocks, and likes to trade at a price; generally, that’s not retail. The typical small trader will tend to be much more reactive. If they are selling, and believe it’s a buyers market, they may just pull their offers and turn buyer. In Hong Kong, where we can see the broker identities and can see how orders are being managed, this can be challenging when it comes to trading less liquid names against retail brokers.

Notwithstanding your reservations, will you test the waters with internal crossing?
As professional investors, we have an obligation to do so. As long as we are comfortable with the level of risk, which takes account of everything from counterparty and operational considerations through to the stability and robustness of the system, then an institution has to investigate whether these new systems will provide better execution outcomes.

Most markets in this part of the world have historically been suspicious of mechanisms to allow arranged crosses. Again, it comes back to the whole retail focus – typically, the retail segment of the market doesn’t do many big crosses. The result being that internal crossing is not easy in some markets for now. My starting position, though, is that I am sceptical about who will benefit from some of these new trading venues. It will certainly benefit the sponsor of the network. In most cases they are designed to benefit liquidity providers - the stat-arb fund or some other prop-traders – as they are typically incentivised to make prices in those venues. It remains to be seen whether the wider buy-side community stands to significantly benefit as well when executing larger orders. If the matching process within a crossing engine is fundamentally better than that offered by the exchange, perhaps offering new order types or tighter spreads, I can see where we may benefit however.

There seems to be a number of liquidity solutions being imported from Europe and the US. Liquidnet, for example, now offers buy-side only crossing in five Asian markets. Chi-X, the pan- European trading facility, plans to start operating in the Australian market in the third quarter of this year. Do you believe the relative success of execution platforms such as these can be replicated in Asia or are market conditions so different that the glove simply won’t fit?
Asia is not different in terms of what traders and institutions want to achieve. Historically, the critical difference has been the size of the investor base. So, to make alternate pools liquid you have to attract a lot of participants. Until three years ago there were really very few buy-side trading desks of scale in the region. What you’re seeing now is an influx of a lot more players and different types of investors and that gives these new arrivals a much better chance of success.

What impact will liquidity fragmentation have on the use of algorithms in the region?
The sophistication within the algorithms will obviously have to increase. The smart routing technology will have to be built in as standard. I think the trader sitting on a buy-side desk may still wish to choose the algorithm with the same label as before but it will just function in a different way. Traders may still be looking to achieve a VWAP-type execution or, indeed, some more aggressive strategy but the model itself will have to be checking multiple liquidity pools.

Are VWAP-calibrated algorithms still dominant in Asia given the impact of overnight desks?
The VWAP influence from overnight desks is still prevalent. But it’s important to remember that a benchmark is not necessarily a target; it’s what you’re looking to beat as a trader. If a price is expected to be volatile on a particular day, VWAP effectively removes the risk of being wrong – it doesn’t take a view and sets out to get the average. This may be a prudent way to trade if there is no news expected, which may lead to unpredictable volume patterns, which upsets these algorithms. If the price is trending away from you, this is not a good choice, however. A more proactive strategy will be calibrated to prioritise price or volume with different benchmarks like arrival price or completion rates. A trader trying to get best execution will need to make informed choices. Algorithms should be used tactically within an execution strategy. That’s where they become valuable over and above where they are at the moment.

Yet, there are many buyside traders in Asia who remain sceptical of algorithms. Over half the traders we polled in a recent survey admitted they didn’t use them. One head trader even said they were “dangerous”.
Everyone has different experiences of them. It’s still early days. We don’t yet have a universe of data points large enough to make definitive judgements. You need to see how algorithms work under all market conditions. They’re just tools and you can use tools well or badly. If you’re buying with a volume participation algorithm, for example, and the stock starts moving sharply higher on increasing volume, then you will probably be unhappy with the outcome. The thing is to understand what you are using rather than blame the tool. It’s nice to think that an attentive trader can do better than an algorithm, but I suspect that in the future, given the arms race that is taking place, there will be many situations where the trader is slower to react to a change in the market than an algorithm. Such a situation is where there is a big bid/offer size on a reasonably wide spread, when that offer starts getting lifted then how quickly will a human operator react versus a tactical algorithm? An algorithm will beat the operator to the punch every time. Good traders, like good drivers, can supplement their skills with sophisticated electronics to make them more effective.

In view of the challenges you face – not least the disparate nature of markets in Asia – how do you go about defining best execution in the absence of regulatory guidelines?
We think of best execution in terms of consistency within our trading process. The idea being that an individual trade is always going to be sensitive to the conditions that prevail on the day when the order arrives on the trading desk. Then there is the trade off between price and size, which can differ on each trade. My standard response to the question ‘what is the implementation cost of this portfolio trade going to be’ is that it mainly depends on who else is doing what on the day you do it.

 A final quick comment on TCA: there is a limited amount of information that you can realistically get from analysing an individual trade. A better approach is to look at aggregate numbers and monitor the way the traders are going about their job: what tools are they using? What sort of information are they acting on? How do they interact with the market? And to your earlier point: are they using all the available venues? In the future that may well be a critical answer to the question of whether they are getting best execution.