High-Frequency Trading: Benefits and Harms

Today, I came across this blog post by Felix Salmon of Reuters, which talked about a GIF of high frequency trading activity from January 2007 to January 2012. As you can see, trading activity became incredibly large as time went on.

Short definition: HFT is essentially the use of computer technology and algorithms to analyze market data and trade securities, like stocks and options. The practice is usually characterized by trading in very small assets, with an incredibly large amount of trades done at a time.

This fits in with a number of things I've heard about high-frequency trading, with people claiming that it causes excess volatility in the market and harms long-term investors, or that it's a democratizing force among traders compared to the old boy's club that used to characterize traders.

I'm curious to know what people here think of it, so what's your view?

edited 7th Aug '12 11:11:12 AM by Wicked223

Could you add a definition of high-frequency trading to your OP?

EDIT: Thank you.

edited 7th Aug '12 11:22:14 AM by BestOf

As a computer programmer, I've got to say that anyone who trusts $420 million to an algorithm with no human override is pretty much asking for it.

Apparently, this trading is so fast that letting a human actually check what's going on would mean that you're always too late.

This is mad.

My company actually has a division that carries out a completely automated stocktrading service. As in, no human interaction is required except for dealing with actual clients. All investment decisions are completely handled by the computer, the algorithms built into which are ridiculously complex and can handle basically any change in circumstances. In 20 years of operating, it has only ever lost money in two quarters.

It's certainly possible for humans to be completely cut out of the loop with great success.

COUGHROBINHOODTAXTHATSHITCOUGH

I read somewhere that the average high-frequency trade is so fast that physical constraints on the speed of light over distance are often the deciding factor in how much money you make...

edited 7th Aug '12 1:28:42 PM by Wicked223

For the first time I am glad we haven't invented quantum computers yet.

If we had then these companies would find themselves in a superposition of profitable and unprofitable, and they would probably start lobbying against requirements to report earnings because "you'll collapse the wave function!"

Well, I mean, if you do HFT then a software bug will destroy you. I think it's more of a matter of lacking fail-safes. As the financial world gets more complex, I think it only makes sense to use computers to perform the investment work.

That being said, I'm all for a "tighter margin" society by using computers but we have to use them in places that actually improves lives. HFT doesn't improve our lives. Computerised trading makes sense, just like any algo-trader, it's a matter of how you apply it. You don't actually make money in a day, because nothing in the physical world can reflect that change.

Some people claim that HFT does improve our lives (or at least the lives of investor), though.

Well okay, so what is their claim based on?

Lower trading costs, more liquidity, more accurate stock prices, and a number of other things.

Okay so I'm looking at what it is talking about and I'll admit that I have to rely mostly on friends in the stock market to know these things better.

Here's what I think

• The purpose of the stock market is for investors to put money into various business ventures and hope to get a long-term return for the investment
• The concept of stock trading was to introduce liquidity to make it easier to buy/sell stocks when you wanted to switch between or diversify your portfolio in your investments, because equities is still the number one way to get the best return on your hard-earned savings
• The vast majority of the population is middle class, even though they do not make up the majority of the wealth in the west, we should not impose conditions that are detrimental to them and beneficial to a select few and preferably we can find a solution that is beneficial to everyone
• Stock trading is not zero-sum. The whole market is intended to rise over time. Fractional penny arbitrage and the more ridiculous micro-second ahead trading is not helpful.

I really love seeing awesome algorithms competing in a death rally race of doom in a digital world of win/lose with obvious metrics for success. In a game. This is real money, from real people, who work 8-12 hours a day, put their money in an investment and then get hurt by some guy purchasing a bunch of servers filled with FPGA hardware so they can get a micro-second ahead in trading and make millions by causing long-term investors to lose money.

There's no way for HFT to make money except off the backs of long-term investment firms. It's not possible for physical reality to have had any increase in the way HFT makes money, and so the profit they make that is siphoned off long-term investors is another assault on our economic system for the benefit of the few.

HFT is the extreme result of our lack of placing regulations on stock trading. They're getting money based on how fast long-term investors place their trades.

I should point out that HFT is also used to reduce risk. Monte-Carlo simulation based Value at Risk projections, specifically. Essentially, this is using Supercomputers to calculate trillions of possible transactions in order to establish and reduce market risk - risk caused by forces specifically related to market trading and associated factors. (This is distinct from the other main forms of risk, credit risk which is lending related risk, such as that of someone not paying back a loan you give them, and operational risk, which is internal stuff like your controls not being up to scratch and employees losing or stealing money without being caught in time).

HFT as used by speculators is incredibly low margin, which is why it has to be, well, high frequency. Arbitrage, which is making money by exploiting price differences over periods of time, is as close as can be found to the perfect ideal of guaranteed reward with no risk. It's also an incredibly low reward since everyone knows about and does it, so it has to be done in volume and over ridiculously tiny periods of time in order to make any real money. These days, humand are too slow for the margins required. But as I've just said, it can also be used for positive reasons, calculating the transactions without actually doing them in order to reduce risk. All major banks do this.

My degree and job are in market risk management, which explains the jargon filled post. Happy to clarify if anything isn't clear.

So basically, it is an argument for the correct uses of HFT?

My concern is that these price differences (outside of commodities) arise heavily from structural issues with how we conduct investments in corporations, largely on the basis of "introducing greater liquidity". I'm not sure how helpful that is to joe-schmoe working 9-5. What difference does it make to him?

None. But it buys bigger Porsches for those who sign the cheques that buy the kit needed.

I want a quantum computer now.

That said, this sounds like a perfect reason to set up a 1% or so tax (actually, most economists would suggest something like .0001%) on stock trading, assuming that that tax won't damage liquidity markets. A similar tax was already proposed on currency markets (to stabilize them against speculative fluctuations); it's gone nowhere because such a tax would have to be worldwide in scope to prevent an exchange haven.

Then again, an agreement between London, New York and Tokyo alone would probably do pretty well at capturing both currency exchanges and stock trades, so it's not all bad.

edited 8th Aug '12 11:57:42 PM by Ramidel