Many providers of execution algorithms are looking at Asia as the next market for rapid growth of algorithmic trading. However, anyone who believes that an algorithm designed to trade in Europe or the US will work as well in Asia is in for a nasty shock. Asia has some unique features which make it a very distinct market. Startling differences between Europe and Asia can be seen by looking at factors such as tick size, market capitalisation, intraday volatility and market rules. This article shows that algorithms need to be specifically customised to trade effectively in Asian markets.
Tick size and spread
Many Asian stocks have very large spreads. NTT DoCoMo, the
Japanese mobile communication company, for example, has a median
spread of 60 basis points. The larger spreads are due to tick
sizes in Asia being greater than in Europe. Securities with large
tick sizes (in basis points) will generally have bigger bid-ask
spreads than stocks with smaller tick sizes. Figures 1a and 1b
show the tick-size distribution for the 1,100 most-traded stocks
in both Europe and Asia (Japan and Hong Kong). By comparing the
two tick-size distributions, it can be seen that Japan and Hong
Kong have significantly more stocks in the 10-60 basis-point
Similarly, Figures 2a and 2b show the median spread distributions for the 1,100 most-traded stocks in both regions. In this case, Japan and Hong Kong have significantly more stocks in the 20-60 basis-point range than Europe. If a stock has a smaller bid-ask spread, it will be easier to trade since the cost of execution will be smaller (in percentage terms relative to the stock price). It is harder to trade in Asia compared to Europe due to the fact that tick sizes and spreads are larger in Asia. The gap between spreads in the two regions will most likely increase in the foreseeable future as tick sizes in Europe appear to be decreasing faster than in Asia. The London Stock Exchange reduced tick sizes last year and Euronext is moving to further decrease tick sizes in 2008.
"It is harder to trade in Asia compared to Europe due to the fact
that tick sizes and spreads are larger …"
The general rule in Europe is
that the larger the market capitalisation the easier it is to
trade the stock using an execution algorithm. By looking at
Figures 3a and 3b, it can be seen that in Europe the highest
market capitalised stocks (ranked 1-100) have the smallest
spreads on average. In Europe, stocks have small bid-ask spreads
when there are many shares available to buy and sell. However, in
Japan the largest stocks by market cap have similar spreads to
the lowest market cap stocks (i.e. those ranked 1,001-1,100).
The price that needs to be paid in order to execute trades in Japan is relatively higher than in Europe, even when there are a large number of shares available to buy and sell. Consequently, using market capitalisation as an indicator to recognise easy-to-trade stocks is not possible (or, at least, advisable) in Asia.
Asian markets are generally more volatile than European markets.
From September 24 to October 15, 2007, intraday volatility
calculated from 20-minute returns for the European Stoxx 600 was
approximately ten basis points. Intraday volatility calculated
from 20-minute returns for the Japanese Topix 1500 was
approximately 19 basis points.
When comparing Asian and European markets, it is also worth bearing in mind that the continuous trading hours differ considerably between regions. In London and Paris, continuous trading lasts for eight and a half hours, while in Japan it is four and a half hours and four hours in Hong Kong (a.m. and p.m. sessions combined). Effectively, both these Asian markets have half the time to trade similar volumes to European exchanges. These shorter trading hours are partially responsible for the greater volatility of Asian markets.
Increased volatility will lead to larger deviations in the performance of execution algorithms (whether using VWAP or arrival as a performance benchmark). Since the intraday volatility in Japan is considerably higher than in Europe, it is harder to achieve good algorithmic performance in Asia than in Europe.
The unique market rules in Asia can lead to very different
behaviour in the trading day. Many Asian countries have two
trading periods, an a.m. session and p.m. session with a break in
between. In Japan, for example, a higher percentage of volume is
traded during auction periods and around the lunchtime break.
Anomalies such as the pre-lunch rush mean algorithms need to be
customised to market behaviour in order to be effective.
Also, algorithms need to be customised for the unique short-selling rules that occur on different Asian exchanges. Algorithms need to take into account the fact that short-sell orders cannot always immediately execute due to market rules and adjust their behaviour accordingly.
"Since the intraday volatility in Japan is considerably higher than
in Europe, it is harder to achieve good algorithmic performance
Customise and understand
Larger bid-ask spreads in Asia, high levels of volatility in Asia
and different market rules combine to make trading in Asia very
different from trading in Europe. Consequently, algorithms
deployed in Asian markets must be customised to individual market
behaviour. However, it is not just about changing the algorithms:
the end user needs to be aware of the unique market behaviour.
Even with an algorithm tailored for the Asian market, there are
some stocks that are not suitable to trade on analgorithm. Stocks
with very large bid-ask spreads and a very low number of daily
trades are not suited for algorithmic trading in any market.
The algorithmic user needs to be trained in how to get the best use out of execution algorithms. Essentially, algorithms will be deployed in Asia in the same manner as in Europe, i.e. as a productivity tool. Optimisation of algorithmic performance to the markets is not in itself enough, the end users also need to be aware of how the local structure can determine and effect algorithmic performance. For algorithmic trading to experience the growth as seen in Europe, there needs to be continuous dialogue between the providers and users to ensure education on the advantages and limitations of algorithms in Asia. Only by this two-way process will execution algorithms be seen in Asia as the valued productivity tools that they are in Europe and the US.