Much has been written recently on the millisecond trading and its relation to the recent flash crash. I said not long ago that I thought that the legitimate purposes of the stock market could be met by clearing once a day. This was based on the idea that a certain amount of human consideration about the state of a company was necessary to legitimize a trade. Forrest Bennett convinced me in a conversation that there is a legitimate public purpose to more frequent trading. The stock market is a channel for many kinds of signals, in the form of price signals. Low latency is good in any signal system and these signals improve the economy even when those signals travel via traders who are proud of the fact that don’t know what the companies do. If new information about the cost of some industrial component becomes available that bears indirectly on some other venture, it is good for that signal to get thru the system quickly. Traders presumably have correlation matrices relating stock prices between companies. This is the old Input-Output matrix idea in action. Often several steps are involved and then my rule would cause a delay of several days. Relevance of such signals to various ventures is likely to vary with conditions in ways unknown to any specialist, or small collection of specialists, but ‘known’ to the collective market. Here is more detailed logic on futures.

Just now (2011 Sept) I see the advantage of fast trading for equities less so than for futures.

There is underway now (Aug 2011) the “Hibernian Express” (? .) a new trans Atlantic cable whose sole advantage is a few milliseconds less transit time. Also new cables are planned under the Arctic ocean to save 60 ms from London to Tokyo. Some have speculated that the business plan is to sell to traders who plan to exploit the advantage of milliseconds in price signals. It seems clear to me that there will be such an advantage but probably much less net collective advantage than the cost of the new cable. If I traded in such domains I would be tempted to sell several of these traders short; they each think they have the best system and they can’t all be right. It would be an interesting exercise to estimate the worth of a millisecond advantage, not to a trader, but to the economy.

Technical traders seek patterns in price histories and exploit these. These indeed ‘train’ the market to move signals quickly. I presume that this is usually to the advantage of true hedgers that receive the signals. I think this may be identical to Hebbian learning thought by many to be how the brain learns.

This article provides some interesting new information but not much insight. Quote:

I think not. It is only a tax on the fast traders which you can choose not to join. They suggest that the prices of these new circuits are secret as well as their customers. They also mention some new land based fibers with smaller latency.
High Frequency Trading and its Impact on Market Quality by Jonathan A. Brogaard.
Long Term Greed
recent note