[nimble]

There is an obvious problem with HFT: it gives an advantage to lower latency analysis and communication, leading to huge amounts of wasted effort getting faster pipes and software, co-locating automated trading, etc. None of this is delivering real value to the economy, as far as I can see. If you have an argument for why trades that happen at millisecond time scales help liquidity and price discovery over what you'd get with a system that handled trades on the scale of minutes, I'd like to hear it. I have a hard time imagining that the benefits (to society at large) could outweigh the costs of the waste when I see what's being done.

That said, I agree that the high frequency of trades themselves are probably not the problem. The problem is rewarding a latency/processing advantage.

[snowwrestler]

If a HFT firm makes a profit then they are, by definition, adding value to the economy.

All that technology investment is not a waste; it allows firms to condense money out of information asymmetries with ever greater efficiency.

Edit to add: if you prefer industries that build "real things" then look at the revenue that HFT delivers to companies who make high-speed networking and computer equipment.

[nimble]

Economic value is a measure of the benefit that an economic actor can gain from either a good or service (first sentence at http://en.wikipedia.org/wiki/Value_(economics) ).

My post was comparing the current system that allows millisecond scale trading with a hypothetical one that slows trades down to something on the order of minutes. Your observation that people are willing to pay to exploit information asymmetries in the current framework doesn't address the question of how my proposed change would affect net economic value.

[snowwrestler]

The metric of measurement is wealth, which profits create.

Your proposed change would reduce the benefit because it would eliminate the profit (wealth) that HFT firms create.

Your proposed system is not hypothetical; it is how all trades used to happen. HFT did not arise from nothing. People developed it because it created new profit in addition to that generated by slower trading.

[judk]

Are you seriously claiming that profits create wealth but counterparty losses don't destroy wealth???

[snowwrestler]

Are you claiming that the stock market is zero sum?

[yummyfajitas]

Suppose I'm a speculator willing to pay $0.03/share for liquidity. It's valuable for someone (I don't care who) to provide that liquidity.

If HFTs were all 100ms slower, I wouldn't care. But if HFT 1 is 100ms slower than HFT 2, then HFT2 will capture $0.03/share and HFT 1 captures nothing. It's a wasteful arms race. It's also fixable simply by eliminating the subpenny rule.

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

[nimble]

Eliminating the subpenny rule sounds like a good idea, but it wouldn't address the advantage of responding to news quickly, would it?

[kasey_junk]

It's not that it would eliminate it, but it would make the margins to due pure latency arbitrage so small as to make it not worthwhile (in theory anyway).

That said, there is and always will be an advantage to players that can respond more nimbly to changing market conditions (news being one input to this) and there is no inherent problem with that (and there may be lots of positives to the market as a whole).

[minimax]

Have you considered this post in light of the inverted exchanges BYX, EDGA, and BX? You can effectively make your liquidity cheaper by posting to an exchange with an inverted fee schedule (pay to add, rebate when you take), but we still see the "latency arms race" at the traditional exchanges.

[snowwrestler]

Moving the decimal point to the right just increases by 10x the volume necessary to create the same profit. I think a likely result would be massive consolidation to a couple big companies, who would then compete once again on latency but at much higher volumes.