The point is that whatever function you think HFT serves can be done just as well at fixed tick rates. This whole latency optimization mini game is just a pure waste.
Btw, a "tick" in trading is the minimum up/down movement of an asset (e.g. 0.25 points on NQ, or 0.01 on TSLA).
This is true, but the term is overloaded: if I say "the exchange slid the price one tick to comply with Reg. NMS" then I'm using it in the sense you mean; if I say "the model recomputes on each tick" then I'm using it in the sense of the GP.
Increasing market friction is not a good idea.
As for price discovery, a lot of information can happen in one second. Quantizing just adds risk, which adds cost, which harms everyone.
> Individuals and corporations aren't making tradable decisions at subsecond intervals.
No, but they're benefitting significantly from the fact that the cost of trading is now minuscule compared to what it used to be.
No, it adds risk at the scale of the quantum, adding cost for those trading at the scale of the quantum, harming a very small subset of traders who would normally benefit from trading at or below the scale of the quantum, like HFT firms and short-term market makers. Should such institutions cease to exist because the cost of trading at such scales becomes prohibitively expensive, I highly doubt that society will suffer.
The spread represents the amount of uncertainty and risk associated with a price. Every bip away from the midpoint your bid/ask sits is pure cost to the participant, and that's true for all participants.
Even if I only come to market once a year to splurge my bonus on GME, I'm worse off paying 20c over the midpoint than I am paying 1c.
Would this really hold true linearly across trading frequencies?
As we get to shorter and shorter delays, the technological cost for participating in HFT will just keep going up (e.g. building your own microwave transmission towers et. al. sillyness), decreasing the amount of potential participants.
Won't this at some point start having the opposite effect - limiting participants in the market to those with enough cash to play? Wouldn't there be some theoretical level of trading delay that should be regulated in order to ensure optimal liquidity as well as market participation?
Okay, can you please use this evidence to say how much the interval has to increase for spreads to increase by 1 cent?
That's easy! Come time to match and fill for orders placed at identical prices, there's no reason a random number generator couldn't do the trick. Trading second-by-second need not be reliably profitable in the extremely short term to ensure an orderly and efficient market. The whole point of a static tick rate is to disincentivize such myopic trading in the first place!
I agree, though, that the logical conclusion of all this would be to replace the trading session with a market auction. However, while I appreciate the superior allocative efficiency of a Vickrey-style sealed-bid second-price multi-unit double auction[1] as much as the next guy, I have more faith in iterative improvements coming to pass before revolutionary ones.
> Btw, a "tick" in trading is the minimum up/down movement of an asset (e.g. 0.25 points on NQ, or 0.01 on TSLA).
Right, I apologize if I confused anyone with overloaded terms, but I think it's appropriate to expect the average Hacker News reader to correctly parse "tick rate" in the context of telecommunications and networking rather than confuse it with the trading concept of "tick size."
> That's easy! Come time to match and fill for orders placed at identical prices, there's no reason a random number generator couldn't do the trick. Trading second-by-second need not be reliably profitable in the extremely short term to ensure an orderly and efficient market. The whole point of a static tick rate is to disincentivize such myopic trading in the first place!
Adding deliberate randomness to the market makes it neither more orderly nor more efficient.
And you're confusing "myopic" trading with market making. Buying now and selling 10 minutes from now brings together two other participants who want to trade but not at the same time. That seller may have needed to have accepted a worse price in order to execute when they needed to.
But if you just look at the market maker, they briefly held onto something they didn't need for 10 minutes. What's the point!
A one microsecond or one millisecond time window would serve all legitimate trading purposes.
Sorry, does a "random number generator" mean some other type of randomness? And what's true for 10 minutes is also true for one second.
> A one microsecond or one millisecond time window would serve all legitimate trading purposes.
The fact that you're throwing around quantizations that are three orders of magnitude different and just saying "this'll do" makes me think you don't actually understand how trading works.
Which then? A millisecond? Why not a microsecond? If a microsecond is fine, why not a nanosecond?
Why are you so keen to put arbitrary restrictions on when people can buy/sell things from each other?
> makes me think you don't actually understand how trading works
whoo-hoo! Getting snooty there, aren't you? I think I understand it as well as you do.
> Why are you so keen to put arbitrary restrictions on when people can buy/sell things from each other?
that's called "market regulation." That's what the SEC's mission is. It's intended to convince unsophisticated investors that they're not being cheated by slick operators.
shrug
You're proposing fundamental changes to the way that markets operate, which would have profound (and mostly negative) effects on the economy, but being entirely casual about the pretty important details across multiple orders of magnitude (a millisecond here, a microsecond there, what's the difference?).
I'm not sure you do understand it.
> that's called "market regulation." That's what the SEC's mission is. It's intended to convince unsophisticated investors that they're not being cheated by slick operators.
Sure, but you have not in any way made any case about how quantizing the market stops unsophisticated investors from "being cheated by slick operators.". What does "slick" mean? What does "cheating" mean? If I buy a computer and colo it with my local exchange to get better latency, is that being "slick"? How about if I hire an army of PhDs to do extensive market analysis, am I "cheating" the unsophisticated investor then?
shrug. you do you. If not agreeing with you is "not understanding," then you're right. I think you're the one who doesn't understand the US securities market.
Nitpicking the precise meaning of every word is a good working definition of "slick."
We're done here.
Surely you must realize that regulating the markets is not as simplistic as you make it out to be. There are certainly always improvements to be had but they have to be carefully thought out because every rule and truly cause new imbalances.
Agree, but to both sides of the above discussion. The other guys literally just says "you don't understand". And I have to say, I don't understand either. If there's someone here who actually understands HFT and has done it, it would be nice to hear your thoughts on the matter.
What do YOU think would happen if we placed and arbitrary minimum tick rate at say 1 second? 1 microsecond? 1 nanoescond? 1 femtosecond?
Why would the US securitieis market collapse? Would it collapse? Why do we need extreme liquidity at the nanosecond level? Where can I get a loan for a nanosecond?
As mentioned elsewhere, lets say you quantize. How do you decide which of two orders that arrive within the same interval at the same volume and price gets filled? Unless your proposal to quantize the market also acknowledges that there's now significant, basic questions to be answered about how a matching engine would even work, then it's not a serious proposal. It betrays the fact that the proposer hasn't fully thought through the impact of what they're suggesting.
But, to answer your question: Quantizing the market will aggregate and dilute market information into the size of your interval. The enforced delay between your order and execution will fundamentally add some uncertainty and risk into every trade, which will increase spreads and the cost of trading. This will obviously happen less at smaller intervals, but I find it hard to see any offsetting benefit whatsoever.
What'll actually happen is that no-one will place orders until just before the interval boundary, so you either have an open order book (and still have an arms race so that you can see the book, process it and submit an order before the deadline), or you have a closed book, in which case you're now running a dark pool. (People tried dark pools, and discovered liquidity fell through the floor, and the thing that market customers really want to be able to do is trade. So they invited a bunch of HFTs to come along and participate in the dark pools, and everyone's now wondering what the point was).
Finally, unless you form a World Government, you'll discover that different markets will end up operating on different quantizing intervals and rules, which will lead to enormous structural arbitrage opportunities across different venues.
> Why would the US securitieis market collapse? Would it collapse?
Of course not. We used to have high latencies and high costs in the olden days. I think it was worse then.
> Why do we need extreme liquidity at the nanosecond level
Because I want to come to market and buy right now based on the available information in the universe right now. I don't want the risk that something will change in the next 500ms (or whatever) factored into my price.
The way they operate right now. There used to be plenty of seconds of delay.
Submit a limit order at a price you find fair and wait your turn or just take whatever's available on the other side of the order book. Markets have long functioned productively and efficiently on slower, human-speed timescales without the need for day traders or intrahour market makers.
For these sort of things were you know there will be an end with a minimum achievable delay, it just makes way more sense to arbitrarily decide it's X and call it a day and let humanity use its resources on better things, than let this waste of resources continue.
Not quite, because the next step is being physically closer to the exchange, and there's only so much real estate to go around.
I think the SEC decreeing "random" for trades within some time window (is the window of millisecond size? microsecond size? not sure) would be totally legit, and it would redirect financial resources to better trades, not faster ones.