The problem with HFT is that practically speaking, the value of a company can't change in timeframes shorter than a couple of hours (barring exceptional circumstances like a big announcement). So a HFT firm is not actually benefiting from information asymmetry about the product of the company being traded or real-world conditions, it is benefiting from information asymmetry about the conditions at the exchange. In theory this is transferring money from subject-matter experts to experts in how exchange servers & the local fiberoptic grid works.
I'd be alone in a room of HF traders, but this isn't useful economic activity vs. subject matter experts making money.
I don't understand all the arguments about 'liquidity', I don't see why a slight price spread is a bad thing; provided that I can sell stock in an hour or so. There isn't a magic 'efficient price' determined by the gods of the market - it is whatever the buyer is willing to pay.
As far as I can tell that isn't hurting anyone more than normal arbitrage.
That's effectively describing the problem is a slightly reformulated way. HFT basically gives traders priority access to the market - they can receive outside information at the same time as you and just react to it quicker.
Moreover, they have priority access to information about how others in the market are behaving.
>I'm not sure if it's possible to fix that, but one idea would be to do trades in discrete increments of a few seconds or whatever.
That simply changes the race from about being the quickest trader to being first in the queue during the increments.
True, it would have to be in a nondeterministic order. I posted another possible solution above, randomly fuzz the time trades are executed to within a few seconds after they are received. Completely disrupting any millisecond advantages.
I just realized the problem with that is someone could just send millions of trades through and increase the probability one of them is executed first. So maybe not.
So we want lower barriers to entry in trading and moving capital to the productive sectors of the economy and smarter trades. That is why the transaction tax could be helpful - we want to move the largest amount of capital to the economy and not the maximum possible numbers of transactions.
That's backwards. Exactly synchronising the dissemination of information is a hard problem. HFT leverages even small sub-second differences.
Even simple physical factors (i.e. the speed of light) mean that being closer to the exchange is highly desirable as this will mean getting the information sooner. People did not create the phsyics of that problem.
It's not perfect because someone could still get information at the end of the time block before anyone else does. I can't think of a solution to that. Perhaps instead of time blocks, execute trades randomly within a short period of them being posted. Completely disrupting millisecond advantages with random noise.
I don't know if the problem is real, and I don't know if this tax would fix it, but that is in my understanding why taxes on financial transactions are thought as useful.
I'm not sure if it's possible to fix that, but one idea would be to do trades in discrete increments of a few seconds or whatever. Similar to how exchanges close overnight and companies can release important information, and everyone can be prepared by the next day when they start trading again.