[cmdrfred]

From your source: "The opponents of high-frequency trading feel that whatever liquidity HFT creates is superficial because the securities are held for a very brief period (seconds or fractions of a second) before being sold back again into the market. Most of the time, securities are bought and sold very frequently between high-frequency traders until they are bought by an investor. Opponents say there is thus no ultimate creation of liquidity but a mere facilitation for order execution.

HFT results in what is called hot potato volume. Positions are being ping-ponged between high-frequency traders and the other marketmakers. Thus there is the creation of great volume and no concurrent depth. For orders to be absorbed, buyers must hold their positions for a longer time than just a few seconds.”

And then merely counters that with a argumentum ad populum:

"With more than a decade in existence, high-frequency trading is now more or less an accepted part of the stock markets. There is a consensus that, on average, HFT has added liquidity to the markets and reduced trading costs."

The evidence appears rather weak for that position.

[mrchicity]

That "mere facilitation" is extremely valuable. Do you think stock exchanges themselves are valuable and worth paying for? They bring buyers and sellers together in a centralized place, merely facilitating a pre-existing desire to trade, taking a small cut in the process.

Imagine a world without them. When you want to buy a share of Microsoft, you could call all your friends seeing if they want to sell any or know someone who does. That could take forever or you might never trade. Search costs are real.

Much like the stock exchange helps buyers and sellers find one another in the same security, HFT helps buyers and sellers find one another across exchanges, securities or risk factors. They basically create a meta-market for you, at very little cost. Typical HFT profit margins are fractions of a cent per share traded, similar in magnitude to what the exchange earns.

To give a real example, imagine you want to sell Valeant Pharmaceuticals after their latest scandal. You only have access to the US markets. There's an abundance of sellers in New York pushing the price down, but over in Canada, the price is a little higher. Someone in Toronto just read the latest report, thinks the price will rise, and put a bid in, but he only has access to the Canadian markets. The two of you wish you could find one another, but you can't.

An HFT algo with real-time data from Canada and the foreign exchange rate places a bid on the NYSE, tightening the spread. You sell to them. They turn around and hedge their stock exposure by selling to the guy in Toronto, and trade an FX futures contract in Chicago. This is "mere facilitation", but it helped two traders transact at a better price than they would have otherwise, and forces convergence between two related markets to make them more efficient. Automation lets him run this same trade over thousands of securities for very low margins, making a few dollars in each every day. How is that not a great thing?

[fsloth]

"The evidence appears rather weak for that position."

Sure, and it's exactly not a scientifically backed site. "Not obviously pathological" was the intended message from my part. I have no strong opinion on the matter. My weak belief is that it's non-pathological, and thus not worth chiding.

Any real world system that has enough degrees of freedom to function and not be too brittle and cumbersome is bound to have loopholes and such. Which are ok, as long as they are not grossly parasitical.