Trading Fortunes Depend on a Mysterious Antenna in an Empty Field
bloomberg.com
bloomberg.com
If 2 person read the same news and send a trade order at the same hundredth of second, who wins the trade should be attributed at random, not to whoever is closest to the exchange.
Super fast trades only allow high frequency trading firms to skim the market...
Who are all the market participants? How do you know there is no value in sub-millisecond trades? Why is trading different from the many other computational tasks humans strive to shave milliseconds off of that they can't possibly notice?
> If 2 person read the same news and send a trade order at the same hundredth of second, who wins the trade should be attributed at random, not to whoever is closest to the exchange.
Why?
> Super fast trades only allow high frequency trading firms to skim the market...
Is your claim that HFT does not provide any value to the market?
Because trading execution is not the critical path in the execution of a financial strategy!!!!
> why?
Because the intention of the traders in the same and it is impossible to measure correctly who had the idea first.
> Is your claim that HFT does not provide any value to the market?
Yes!!!
Okay, so what is?
> Because the intention of the traders in the same and it is impossible to measure correctly who had the idea first.
Why do we want to reward the person who had the idea first? Why is that a more optimal reward mechanism than whoever arrives first?
> >Is your claim that HFT does not provide any value to the market?
Yes!!!
You implicitly admitted that HFT does provide sub-second liquidity benefits elsewhere in this thread, but questioned why we'd want sub-millisecond liquidity. Can you illustrate why we'd want one and not the other (what is especially beneficial or harmful about one versus the other)?
1. non HFT traders are the fundamental players in the market, without them, HFT traders wouldn't exist. The exchange itself wouldn't exist. The exchange's objective is to connect non HFT traders.
2. subsecond liquidity is useful because non-HFT traders want their orders to be done as fast as possible: they want to their orders to feel immediate.
3. non HFT traders do not care about submillisecond liquidity because the timeframes in which they invest is much larger.
4. submillisecond liquidity is harmful because it incentivize HFT firms to stupidly race to reach the exchange without adding any value for the non HFT traders, while at some point the exchange has to say "we received these 2 orders at the same time".
Sure: there are buyers, sellers and market makers. Naturally the market exists for buyers and sellers. Market makers can't exist without anyone to make a market for, and buyers and sellers can't exist without a way to find each other (more generally, the market requires liquidity to be healthy). The market is an ecosystem.
Market makers used to make markets in timescales that required me to wait minutes before I could complete an order. Now I can wait less than a second. So this brings us full circle back to my question: if they benefit the market at sub-second granularity, and you can't demonstrate a harm in sub-millisecond granularity, why do I want them to stop?
> 2. subsecond liquidity is useful because non-HFT traders want their orders to be done as fast as possible: they want to their orders to feel immediate.
Yep, I'm with you there.
> 3. non HFT traders do not care about submillisecond liquidity because the timeframes in which they invest is much larger.
And now you've lost me again; I am an active algotrader, and I don't work in HFT timescales, but I appreciate the liquidity benefits I get from HFT all the same because I do trade in timescales that make it meaningful.
Furthermore, I can't really consciously think productively in sub-second or sub-millisecond resolution.
Retail investors who trade by hand don't need sub-second or sub-millisecond resolution. If you have an argument that both are harmful, make that argument. If you have an argument that sub-millisecond is harmful, make that argument. But so far you're saying it's bad because...people don't need it, which is odd. I don't really try to stop people from doing things that I either don't care or about or which don't harm me. There's a lot of that in the world.
> 4. submillisecond liquidity is harmful because it incentivize HFT firms to stupidly race to reach the exchange without adding any value for the non HFT traders, while at some point the exchange has to say "we received these 2 orders at the same time".
I'm still not following you, because you're not demonstrating harm here. It's harmful because it's stupid? Why is it stupid? Who can deem it to be stupid? Even if it is stupid, what makes it actually harmful to anyone?
As to your last point, what do you think is more likely - timestamp collisions at second, sub-second or sub-millisecond resolution? Frankly your point here is an argument for greater resolution, no lesser.
Strong disagree. There are plenty of non-HFT participants whose critical path is the strategy execution (and is latency sensitive).
When you buy a stock from your banker, you don't care that you buy it at 65 or 65.00001.
Same thing for time, you don't care that you send the order at 15:14:25.0000001 or 15:14:25.5
Markets should be tailored to the needs of its users.
Well, they are. HFTs do 70% of the volume, so the markets are tailored to their needs.
1. Of course it's arbitrary. And the current time resolution (a computer tick = one ten-millionth?) is just as arbitrary.
2. The market is a man-made system, it's made by and for humans. It would make sense that the time resolution of events is mapped to human perception: You need subsecond liquidity because traders want to perceive their orders as being immediate. But as soon as the order looks immediate, you don't need further precision. The additional resolution is useless to anyone making decisions and only encourages a stupid race that requires immense investments.
3. Can you explain what you think is the value in submillisecond trading?
So why should we change it?
> The market is a man-made system, it's made by and for humans. It would make sense that the time resolution of events is mapped to human perception: You need subsecond liquidity because traders want to perceive their orders as being immediate. But as soon as the order looks immediate, you don't need further precision. The additional resolution is useless to anyone making decisions and only encourages a stupid race that requires immense investments.
I don't follow - by that logic, everything created by humans shouldn't have resolution greater than what we can perceive. Does this hold for, say, HTTP requests, or pings?
> Can you explain what you think is the value in submillisecond trading?
It allows me to complete trades before the market turns against me. In the absence of any particular harm, faster execution is better.
Also, to your third point -- as a market maker, more information is strictly better than less information (as a corollary, shorter time resolution is better than longer time resolution). Any time I have less information, my predictions are worse, my risk goes up, and I have to either quote less or wider, making liquidity worse and hurting other (non-HFT) participants.
That's really not true. The microwaves in this case are being used to communicate between NYC and Chicago to trade products in increments of <1bp. Retail investors may not be as concerned about such fluctuations, but "actual market participants" most certainly do.
> Same thing for time, you don't care that you send the order at 15:14:25.0000001 or 15:14:25.5
Again, this is not true. If Fed decides to adjust interest rates at FOMC (14:00:00), they likely do so in a 25bp increment. Market makers absolutely care whether the order is received at the matching engine at 14:00:00.01 vs 14:00:00.5, as such a move can cause large fluctuations in a portfolio's value.
> Markets should be tailored to the needs of its users.
Agreed. Market makers need to be able to adjust their pricing to reflect market conditions. This is critical to the price discovery process. Were they prohibited from updating their pricing rapidly, they would either cease providing liquidity entirely or show much wider spreads to reflect their increased levels of risk. Clearly such an outcome is not advantageous to the market.
Who are the "actual market participants"? Why do you believe they don't care about sub-penny differences, or sub-second differences?
1. It actually reduces HFT profits.
2. It actually benefits the markets.
Who, exactly?
For HFT: they wouldn't have to spend all that time and energy trying to be as close to the market as possible.
Can you describe, in your own words (i.e. not pointing me to an article or book), why HFT constitutes front-running?
So, since you seem to think I'm just a BSer, here goes what I saw:
A trader is looking to buy X futures at a certain price P. There is nothing happening on that market, the order book has been stable for minutes. The trader clicks to buy the futures at market price (which the trader assume is P). Immediately, the price goes up by 1 tick (P+dP), and the trader gets filled at P+dP. Immediately after that, the price goes back down to P. The whole thing last ed 1 ms. What happened is that the order got sent to multiple exchanges at the same time. And on it arrived first on an exchange where it has not been executed. But HFT traders saw the change in the order book on that exchange and "front ran" the order to all the other exchanges, buying everything lying around at P to make the price go up by dP, and selling to the trader at P+dP.
So that is annoying (and illegal).
Strategies used by HFT are usually legal, but all tend to use the same mechanism: - look at what's happening with the price / order book - guess what kind of operations are happening - execute operations that benefit from the operations that should happen afterwards.
- A trader looking to buy futures is almost certainly not going to send their order to several exchanges. The main exchanges list distinct contracts (with, I believe, some cross-exchange listings that are just cross-continent). A person sending an order is not going to get it routed to several different exchanges.
- Your example itself would result in a losing trade for whomever was doing this trading. Let's call this HFT guy, "HFT", and pretend like you were talking about equities(which does have the multiple exchange property). In order to accomplish what your trader was doing, HFT has to:
1 - buy up all existing shares at P
2 - probably buy up all the shares at P+dP as well (because all US markets have price-time priority)
3 - place sell orders at P+dP that would fill this guy's X contracts
4 - According to your story, they (or someone else) then try selling at price P again.
1 - Took a large long position
2 - Took an even larger long position at a worse price, paying commission for this
3 - Managed to sell some of their position at the same price as (2)
4 - Are now trying to sell back their position at P.
1. perhaps it was not a future, I don't recall exactly
2. ??? If HFT guy buys N lots at P, sells N-X at P+dP, sells X at P, he still makes X*dP as profit. The risk he takes only depends on the size of N.
(2). That is not what I said. To sell at P+dP, you need to buy up all the shares at P+dP as well. That's because of price-time priority, which almost all the major markets have (with some exceptions, most notably NYSE -- and even then, their book is mostly price-time priority). This is the idea that a person who placed their order first, has priority over those who placed their orders after. A strategy that makes the trade that you are proposing must then also take out all of P+dP and sell against the incoming order. In a usual case, you pay commission when trading into that position, and (maybe) earn commission when you sell it back (as the liquidity provider). In no market is the commission earned greater than the commission paid. The shares you buy-and-sell at P+dP are straight up loss, and that's before figuring out if X (the incoming shares that you are supposedly front-running) is greater than the #shares at P+dP anyway.
"The risk he takes only depends on the size of N" -- that is true -- that is always true, if you are making an unhedged trade.
Putting aside whether or not your scenario is possible, profitable or actually front-running; by your own admission, not all of HFT is illegal, so why do we need to care about HFT trading based on the order book? What is sinister about formulating a trading strategy based on the order book itself?
In case this hasn't become clear - I am trying to demonstrate (Socratically) that the premises you're deriving your conclusions from aren't entirely coherent.
Of course, if you do that retail investors would no longer have access to $5 stock trades and actually have to pay for the "fair" services they are receiving. Probably some other cost as well to be absorbed under this model.
This is similar to, but strictly worse than pro-rata allocation as used in quite a few products in the US. Pro-rata allocation has its own set of problems, namely that you encourage people to quote in larger size just in the hopes of getting a partial fill.
There's also a larger speed advantage under this model because participants who are able to enter orders at the end of the centisecond time period will earn outsized profits.
My estimate is that your proposed market model actually rewards speed to a greater extent than the current market structure.
You also haven't demonstrated that pure price-time priority is bad in any way.
My opinion is that in making that time resolution coarser, we would avoid having people racing to trade in micro seconds, and spend all that stupid energy in being super fast.
Did you read my response? I explained why your proposal would still involve people racing to trade in microseconds. I also claimed, but didn't prove that your proposal would reward fast traders to a greater extent than the current system.
This part of the GP's view is definitely correct: at some (theoretical) point, the exchange will be operating in Planck-time units, which are quantized.
> Look, my view is that people put too much energy in shaving off microseconds to get to the exchange.
> This has no benefit for the non HFT traders in general.
Also, not every fund can fit their computers into an exchange building, nor would they necessarily want to.
Pretty cool but crazy at the same time.
Here's another shitty comment you can downvote. Have fun. This is what you get for frivolous downvotes. More bile.
Fuck this karma-whore rating shitshow.