Buy-side Expertise: Best Execution in Asia
Schroders’ Head of Asian Trading, Jacqueline Loh, shares her thoughts on trading in Asia, offering comments on which markets are primed for change, how to find value in dark pools and whether unbundling is as useful as people say it is.
Fragmentation arising from multiple sources of liquidity is a necessary step in the evolution of best execution and in the long term, fragmentation will increase the quality of trade executions in Asia. What it means for the buy-side is investment in infrastructure spending to develop new order routers and the like, so we can electronically seek out and have exposure to multiple liquidity sources. For the sell-side, it means acceptance that there will be more competition for the same block of business in the marketplace. It means different things for different buy-side firms as well.
When I think about the investor ID markets in Asia, I am not sure any model is particularly productive because ID markets make it administratively more difficult to trade. IDs can make best execution very difficult to implement, especially if cash and stock checking is the primary consideration. Some of the ID markets, namely Taiwan and Korea, allow trading through omnibus accounts and that seems to be the way it is evolving. The ID markets are slowly going away, but having said that, the most productive example is probably China because the brokers seem to have a handle on exactly how much cash and stock you have in your account, and therefore how much you can sell and buy. You cannot overspend or oversell, and it is relatively easy to take part in IPOs.
Trade allocation used to be a problem with investor IDs; for example, explaining to compliance and regulators why the prices are not exactly the same between accounts. In these cases the use of omnibus accounts really help. Executing through omnibus ID means you know exactly what is in an account and do not experience many of the issues associated with overselling or settlement. It is a lot cleaner.
With retail-heavy markets, anonymity is the primary consideration for us. We tend to trade more using electronic means and make use of dark pools in retail-heavy markets. In addition to that, the algos we use will be more price-specific, rather than volume-participation models, which are more price impacting.
Best Execution, in the Dark?
You would think that dark pools would have more success in markets where spreads are currently wide and there is a need to be anonymous, which would imply ASEAN markets. In practice, however, it has had more success in Hong Kong, and that is because there are more users of electronic trading there. Perhaps the users are a little more sophisticated as well insofar as they are willing to take accountability for their executions. Which is, in fact, what defines electronic trading.
In our experience, dark pools make a difference in terms of liquidity, however, the question is what creates that difference? Is it the electronic trading system feeding through the dark pool that provides the benefit or is it the dark pool, itself? I would say it is the former, but that may depend on each user. routers. I hope the Securities and Exchange Board of India will consider further change including allowing stock crossings and clarifying the rules regarding P-Notes.
The biggest change this year, and the bane of everyone’s life, is the low market turnover. In this context, the better brokers have differentiated themselves by being able to find liquidity outside the market and being able to source and place stock at a price, instead of trading everything in the market. That is taking place on the sales trading desk but equally, in electronic trading, there have been more and more algos which are geared toward that very purpose.
In a period of low volume, technology providers and sales trading desks each have their advantages. To succeed, the sales traders have to change the way in which they do their business, which is to look for blocks. On the electronic trading front, developers have to work harder to find ways of hiding flow, analyzing the order book and posting the right order sizes. For a buy-side desk, the inclination is to trade more stocks electronically, by virtue of the fact that it is anonymous and implies minimal price impact. For sales traders to be able to take that flow away, they have to differentiate themselves from what the algos can do.
The other things some brokers are doing better is that they are working on developing better algos. The algos I use most is actually one which buys the maximum you can at a price without moving the price. The good algos will analyze the size of the order book and decide the size and price of the bids and the offers to post. That is where different algos can differentiate themselves, even with VWAP.
Are all VWAP algos the same? They are and they are not. The principles are the same, but I do a lot of work to analyze algos through the results of the trades we do. Although conceptually, all VWAP algos are supposed to be the same, the results they offer are not. The reason for that is because probably some people fine tune their volume curves more often than others. VWAP algos can probably be as simple or sophisticated as you like. Obviously, the linkage to internal dark pools and how much of it crossed at midpoint also affect the success of a VWAP algo.
If you look at VWAP algos, every broker has one, but they will all give different results, within a band. Some will consistently do better than others and this is a function of how much work the brokers put into fine-tuning it. But having said that, I do not use VWAP algos, except over short periods.
For us, unbundling is a huge step forward in helping us deliver our policy of best execution to our clients. It allows us to trade more with the brokers who offer us flow at a price we want and to trade more with the brokers who offer the necessary expertise. It has worked well in other markets in which we have implemented it, which is all of Asia-Pacific apart from Japan and Australia. We are very happy we implemented unbundling and CSAs. The other thing I would say is it probably could be better understood by brokers. Sometimes brokers will attribute not being paid to a bad call by the research department.
For unbundling and CSAs, the most important thing that people need to know is that it was implemented for the purposes of best execution, that it is policed properly and it works. Apart from that, it probably could be automated better at the trade reconciliation level because it is really time-consuming. Ideally, it should be automated, but it requires the OMS system to be talking to the broker systems but it is not happening yet. To say whether this trade should be unbundled and then if it is, it should go through one pipe; the broker receives it and they match it up at the end of the voting period.