Hacker News Reader

Comment by bbosh | Hacker News Reader

The primary function of the stock market is to exchange ownership (shares) in a company. It's odd that we seem to have forgotten that. What value is there in a computer owning a stock for 10 milliseconds?

View on HN

I've read elsewhere on here that the value of high-frequency trading is that it reduces transaction costs and increases liquidity in the market. I.e. it makes it easier for the humans to buy and sell at the prices they wish to buy and sell at.

Assuming that's true, the question then becomes what are the costs and externalities of HFT and, in balance, are we willing to make those trade-offs? I haven't seen anything addressing those issues yet, but I haven't been looking either.

HFT experts want to weigh in, pro/con/otherwise?

Since most of the people commenting here don't seem to even understand HFT, I'll just leave this here (a tutorial I wrote a while back):

http://www.chrisstucchio.com/blog/2012/hft_apology.html

http://www.chrisstucchio.com/blog/2012/hft_apology2.html

http://www.chrisstucchio.com/blog/2012/hft_whats_broken.html

you should probably link part 2 from the end of part 1, if not it's not obvious that you did write part 2.

Thanks.

HFTs tend to add liquidity when it's least valuable, and they tend to consume liquidity when it's most vital.

For example, an analysis of the 2010 flash crash shows it was worsened by HFTs fleeing their positions once volatility increased, which is the exact time that liquidity and market-making are most valuable.

If there is any social benefit to HFT it is utterly trivial.

Actually, it's arguable that among the primary functions of the markets in addition to exchanging ownership is price discovery.

"Price discovery." What a sweetly tight little piece of propaganda. Hnngh.

That computer is artificially raising the price of the stock for someone who would buy it ... so they marginally either buy less stock, or pay more for each share.

I'm not going to argue that HFT and hedge-funds are necessarily bad (though I don't buy the liquidity argument for stocks that have reasonable volumes), but I don't see how it's helping to fund the company behind the stock at all (or make it attractive to future investors).

In the same way as the bid price is increased by HFT actors (which I think is what you're referring to), the ask price is decreased as well, and your hypothetical buyer would benefit from that.

These are two sides of the same coin, namely HFT decreasing the bid-ask spread, making trading (and thus capital allocation) more efficient across the board.

Sure, but now it's on you to say how long a stock should be owned for, and why that is.

I mean, you could ban HFT and say that you can trade no more than once a second, or a minute, or whatever. Then people would get upset because computers could trade exactly on that second...

> Then people would get upset because computers could trade exactly on that second...

I would think the way to do it would be to hold every trade open for five seconds during which time either party can cancel it.

Then we'd race to see who could cancel as close to 5s as possible.

Regulating trading on speed or time is inherently silly.

> Then we'd race to see who could cancel as close to 5s as possible.

That would reduce the scope of the problem by an order of magnitude or more, because it would limit the benefit of being fast to only those trades to which you yourself are already a party and to which you so happen to receive actionable information at exactly the point in time that the trade is about to close.

And then on top of that, you can put a small penalty (e.g. 1c per share) on canceling a trade to be paid to the other party, which should put a quick end to thoughts of initiating and then canceling several million trades per second that you don't actually want to make.

The primary function of a market is to set prices to facilitate that ownership. The computer is providing value by setting a more accurate price faster.

Reply on news.ycombinator.com