I don't understand the gains for the economy as a whole of smoothing out the price curve in the sub-second range.
I'll make it simple for you. There are none.
HFT traders as a whole make no gains from trading in microseconds as opposed to milliseconds or even seconds. If an investor wants to buy 100 shares right now with a spread of $0.02/share, market makers will earn $2.00 over the next few seconds. The only question is which market maker will earn the $2.00, and the answer is whoever is fastest.
Trading fast doesn't increase profits, all it does is redistribute them. Of course, one HFT firm could also outbid the rest, lowering the spread to $0.01/share and allowing the retail investor to pay only $1.00. That's why competition among HFT firms is good.
But you are correct that all the effort put into lowering latency (rather than improving trading strategies) is a deadweight loss to society.
I'll make it simple for you. There are none.
HFT traders as a whole make no gains from trading in microseconds as opposed to milliseconds or even seconds. If an investor wants to buy 100 shares right now with a spread of $0.02/share, market makers will earn $2.00 over the next few seconds. The only question is which market maker will earn the $2.00, and the answer is whoever is fastest.
Trading fast doesn't increase profits, all it does is redistribute them. Of course, one HFT firm could also outbid the rest, lowering the spread to $0.01/share and allowing the retail investor to pay only $1.00. That's why competition among HFT firms is good.
But you are correct that all the effort put into lowering latency (rather than improving trading strategies) is a deadweight loss to society.