Business Times - 12 Aug 2009
Hock Lock Siew
By R SIVANITHY
THE first thing to note about high-speed computerised trading, whether it applies to trading in stocks or derivatives, is that it's been around for more than 20 years, so it's not a new phenomenon. The second is that its development probably parallels the evolution of financial markets and so its pervasiveness in today's markets is inevitable and unavoidable.
The biggest thing however, is that high-speed trading, program trading or 'algo' (short for algorithmic) trading as it's sometimes known, has been getting a bit of a bad rap lately, though this is probably due more to misinformation than anything else. This doesn't mean the activity is entirely kosher or doesn't warrant attention from regulators or policymakers; however, there is evidence to suggest that present fears may be a tad overblown.
When the US stock market crashed 20 per cent on Oct 19, 1987, a lot of fingers were pointed at program trading as one possible culprit. However, extensive studies at the time failed to establish a convincing connection between algo trading and the crash, especially when it was found that the largest falls occurred when programs were not operating.
In addition, several other markets that did not have program trading actually crashed by more than the US. As a result, other than introducing 'circuit breakers' to slow or halt large moves in the major indices, regulators did not act to rein in high-speed trading.
Fast forward 22 years and the debate over whether algo trading should be regulated was revived a couple of weeks ago when a US senator raised the need for legislation to prohibit 'flash' trading. It's important to note that the call was made specifically for one type of algo trading, but somehow or other subsequent media reporting roped in the entire gamut of high-speed techniques as also possibly requiring regulatory scrutiny.
Flash trading (which incidentally, is a technique not found in the local market and is expressly prohibited here) is simply high-speed front-running. It involves delaying order routing to the best quoted prices by a few milliseconds to let some parties view those orders first. These parties can then buy (or sell) ahead of the actual order execution. Since this is blatantly wrong, few would object to the need for some form of official control to stop this practice.
However, as noted earlier, all other types of algo trading have now been flagged as 'bad' or in need of some form of official intervention. For example, algos usually slice up large orders into smaller blocks, mainly to preserve anonymity and prevent detection by other algos. In the US, for example, although high-speed computerised trading has sparked exponential volume growth in the past few years - as it probably has in the Singapore market - average trade size has dropped sharply, and according to a recent paper, the average trade size at the end of 2008 was only 300 shares.
This in turn has led institutions who need to move large blocks quickly to search out alternative venues, known popularly as 'dark pools' where prices may be better and transactions quicker. Since retail players don't have access to these venues, it is sometimes argued that the playing field is thus not level, especially since dark pools are off-exchange and so present an opaque face to the outside world.
The dark pool story is complicated and the issues numerous. However, if we were to keep the discussion as simple as possible, it's worth noting that if algo trading leads to the proliferation of dark pools and this in turn robs some liquidity from established exchanges while offering benefits to large players, it is then incumbent on the affected exchanges to compete by improving efficiency and offering customers better service.
Over time, this should lead to everyone benefiting - spreads should narrow, transaction costs should fall and price discovery must logically become more efficient.
Of course this presupposes that regulators and exchanges can find a way to live in harmony or complement the dark pool movement.
In Singapore, it is clear that the exchange is very aware of this and is closely watching developments in the high-speed trading and dark pool arenas. The starting point, however, appears correct - except for flash trading, most other forms of algo/program trading are probably okay since they enhance liquidity, market sophistication and market efficiency. As such, only a light regulatory touch would probably suffice.
Hock Lock Siew
By R SIVANITHY
THE first thing to note about high-speed computerised trading, whether it applies to trading in stocks or derivatives, is that it's been around for more than 20 years, so it's not a new phenomenon. The second is that its development probably parallels the evolution of financial markets and so its pervasiveness in today's markets is inevitable and unavoidable.
The biggest thing however, is that high-speed trading, program trading or 'algo' (short for algorithmic) trading as it's sometimes known, has been getting a bit of a bad rap lately, though this is probably due more to misinformation than anything else. This doesn't mean the activity is entirely kosher or doesn't warrant attention from regulators or policymakers; however, there is evidence to suggest that present fears may be a tad overblown.
When the US stock market crashed 20 per cent on Oct 19, 1987, a lot of fingers were pointed at program trading as one possible culprit. However, extensive studies at the time failed to establish a convincing connection between algo trading and the crash, especially when it was found that the largest falls occurred when programs were not operating.
In addition, several other markets that did not have program trading actually crashed by more than the US. As a result, other than introducing 'circuit breakers' to slow or halt large moves in the major indices, regulators did not act to rein in high-speed trading.
Fast forward 22 years and the debate over whether algo trading should be regulated was revived a couple of weeks ago when a US senator raised the need for legislation to prohibit 'flash' trading. It's important to note that the call was made specifically for one type of algo trading, but somehow or other subsequent media reporting roped in the entire gamut of high-speed techniques as also possibly requiring regulatory scrutiny.
Flash trading (which incidentally, is a technique not found in the local market and is expressly prohibited here) is simply high-speed front-running. It involves delaying order routing to the best quoted prices by a few milliseconds to let some parties view those orders first. These parties can then buy (or sell) ahead of the actual order execution. Since this is blatantly wrong, few would object to the need for some form of official control to stop this practice.
However, as noted earlier, all other types of algo trading have now been flagged as 'bad' or in need of some form of official intervention. For example, algos usually slice up large orders into smaller blocks, mainly to preserve anonymity and prevent detection by other algos. In the US, for example, although high-speed computerised trading has sparked exponential volume growth in the past few years - as it probably has in the Singapore market - average trade size has dropped sharply, and according to a recent paper, the average trade size at the end of 2008 was only 300 shares.
This in turn has led institutions who need to move large blocks quickly to search out alternative venues, known popularly as 'dark pools' where prices may be better and transactions quicker. Since retail players don't have access to these venues, it is sometimes argued that the playing field is thus not level, especially since dark pools are off-exchange and so present an opaque face to the outside world.
The dark pool story is complicated and the issues numerous. However, if we were to keep the discussion as simple as possible, it's worth noting that if algo trading leads to the proliferation of dark pools and this in turn robs some liquidity from established exchanges while offering benefits to large players, it is then incumbent on the affected exchanges to compete by improving efficiency and offering customers better service.
Over time, this should lead to everyone benefiting - spreads should narrow, transaction costs should fall and price discovery must logically become more efficient.
Of course this presupposes that regulators and exchanges can find a way to live in harmony or complement the dark pool movement.
In Singapore, it is clear that the exchange is very aware of this and is closely watching developments in the high-speed trading and dark pool arenas. The starting point, however, appears correct - except for flash trading, most other forms of algo/program trading are probably okay since they enhance liquidity, market sophistication and market efficiency. As such, only a light regulatory touch would probably suffice.
No comments:
Post a Comment