I've been following the recent Knight Capital algorithm debacle for some time, and am a bit puzzled: what's the advantage of allowing all these algorithms to run amok in the capital markets? In particular, those algorithms that make hundreds of trades per second.
Brokerages and exchanges will want these algorithms in the market, since algos will allow them to earn commissions. With a global trend of declining commissions, most brokerages and exchanges need to increase the volumes to make up for the declining commission rates. Prop firms (including hedge funds and proprietary desks of investment banks) will want these "algos", since the algos supposedly give an "edge" (i.e. competitive advantage over the rest of the market). As my previous trading boss used to say, "no edge, no profit".
However, if you look at the capital markets on the whole, it's hard to see how the advantages outweigh the risks involved. A rogue algorithm running amok in the market can react much more quickly than any human, causing markets to crash without any underlying economic basis: besides this Knight Capital incident, the Flash Crash of 2010 also comes to mind. This can spread fear through a market faster than you can swear "F----"... and fear (as an emotion) is a dangerous emotion to have spreading throughout a market. Worse, once a "situation" occurs, it's next to impossible to stop it without disrupting the entire market's function. High frequency trading basically raises the possibility of more negative black swans, without any real economic benefit.
It's also arguable if allowing milli- and micro-second trades helps in the price discovery process: is it really meaningful price discovery between a willing buyer and seller who can cognize the price and economic benefit/cost, or is it just "noise" between one algo and another? Can it really be meaningful price discovery if a human watching the exchange (even an experienced floor trader) won't be able to cognize the price, as it just happens far too quickly?
Perhaps as a policy, exchanges globally should consider limiting the speed of transactions, s.t. an algorithm cannot be faster than, say, the typical human reaction time on an electronic trading desk. Otherwise, we run a real risk that the Machines will overtake the markets, and leave us wondering what happened in the aftermath.
The Kay review of equity markets provides some useful ideas: trust & relationships are key to financial intermediation, where the 'goal' of equity markets should be to allow savers to participate in the growth of corporate economic activity.
ReplyDeleteSo the increase in HFT is part of a trend that favours trading vs trust & relationships, and unlikely to contribute to a positive environment.