Wall Street Profits by Putting Investors in the Slow Lane | Hacker News Reader

Wall Street Profits by Putting Investors in the Slow Lane

nytimes.com

160 pointschollida1221 comments

Heh. My second project at my first employer out of college (back in 2006) was building a system to detect violations of this rule (Reg NMS, if you're curious). We found that there were trade-throughs happening on a daily basis, it was so common that it appeared to be just how the markets worked.

Tried to sell it to the SEC and they weren't interested.

Then we pivoted to try to sell to traders, so they could prove to clients that they were getting the best execution possible (or occasionally better than the best possible, but that gets washed out of the aggregate stats). They were very interested, until the data showed that most of the ones who don't already have a proprietary version of this were actually doing terribly on their trade executions. Then they weren't interested at all.

I ended up leaving the company - and the financial industry - at that point. My take-away from the whole experience is that the game really is rigged. I remember reading a non-mainstream economics paper in college that modeled the world not in terms of price equilibria or value-add, but assumed that all actors were basically bandits who would try to take whatever they could by force or deceit. It was horribly depressing at the time, but it actually seems like a more accurate model of how the world really works, the remarkable part being that democratic capitalism has managed to channel the impulses of those bandits (while still being utterly crooked) into a system that on a macro-level basically kinda/sorta works.

> I remember reading a non-mainstream economics paper in college that modeled the world not in terms of price equilibria or value-add, but assumed that all actors were basically bandits who would try to take whatever they could by force or deceit.

If you can find the paper I'd love to see it.

Ooh, I think I got closer, though I still haven't found the exact paper. Search for the work of Mancur Olson. I think "Dictatorship, Democracy, and Development" was the paper, but I just got a JStor link, so I'm not sure it was exactly it.

I disagree that it works--it's just that it hasn't utterly failed yet. Big difference. If you look at failed states, the bandits won. Argentina comes to mind, maybe Venezuela. Brazil is headed there.

Having "not utterly failed" for 240 years is a fairly big accomplishment, precisely because the lifespan of many other societal organizing principles is about equal to a human lifespan (~80 years).

Uh it kinda failed big time in the '20s and '30s

The results were War, big ones at that. So a pretty bad failure mode for millions.

> Having "not utterly failed" for 240 years is a fairly big accomplishment

No. Not even a little. The dinosaurs lasted more than a million times as long. If we talking hominids do not last half a billion years or more we are pathetic losers.

The article isn't about trade throughs. It's an unconvincing argument that rebates bad and that IEX is good because it does not use rebates (while failing to mention any other exchanges that don't pay rebates).

I did some regulatory work as well, for an upcoming regulation called Reg CAT. CAT will require all brokers, traders, exchanges, etc. to report essentially everything that happens--orders, cancellations, trade executions, etc. Right now the SEC really has no way to catch the majority of regulatory hijinx (look at the years-long attempt to track down the cause of the 2010 "flash crash").

Hopefully this will improve the ability to regulate in the future. But it probably won't result in significantly more enforcement--the regulation is written in such a way that the exchanges and FINRA will carry the primary regulatory burden rather the SEC itself. This means that despite there soon being a system that could, say, give you every reg NMS violation via a database query matched to a log of historical latencies, very little will likely change.

I think this is well known. The real problem is that the traders are much smarter than the regulators, and their ability to obscure far exceeds the regulators' ability to untangle. Incentivise your regulators better and you might end up attracting some real talent who can unearth the tricks the crafty traders pull every day (I am an ex trader, from a bulge bracket IB, and trust me, what some of these guys do is not at all kosher).

Also, there is an argument for more time controlled auctions rather than giving everyone a continuous look at the order book. Much like the start of day auction, you could have something similar every minute, where traders submit prices but matching only occurs at the end of each minute. This would certainly put an end to the annoying high/low ticking that happens all too often.

There is no good argument for this. The problems you think it might solve...it doesn't solve.

Hint: if the minute long auction closes at precisely the minute boundary at what time do you want to put your orders in?

Hint: If you can't see the orders during that minute, it doesn't matter when you submit...

Might other things be happening in the world besides orders on a single exchange?

It seems like that would only matter if you assume that a bid can't be canceled or changed. Assuming you're free to cancel or modify a bid right up until the close, whether you put the bid in at the start of the window or the close of the window wouldn't matter as you could always modify or cancel the bid as new information comes in.

If you can always modify or cancel a bid then people will compete to see who can ingest information the fastest and update all of their bids as close to the auction time as possible.

even if you can't modify or cancel a bid people will still compete on speed and submit their one (and only) order as close to the auction time as possible.

i have no idea what he's trying to come up with. there's no need to belabour the point plenty here have tried to show that the power of information (of all kinds: central bank actions, Donald Trump's tweets, corporate filing, obituaries, declaration of war, natural disasters, etc) and the ability to push order submission to the last moment is going to be king. the only way i can imagine one would nullify the value of information is to assign every participant of the market a random price on their trade that has no correlation to the value of the asset that they are trying to buy or sell.

Of course it does. A trader (not an investor) is incentivized to wait to the last possible moment when he can still expect a fill. This is why exchanges disseminate auction imbalance information, to attract people to the auction prior to its completion.

In what way is he incentivized? What advantage does he get from waiting till the last possible moment? He has no clue what the current bids are, so whether he bids at the start or the end he still has exactly as much information. He'll have more information after the bidding closes and the winner is announced of course, but by then it's too late, only thing he can do then is participate in the next round of bidding, but that puts him right back in the situation he was in before where until the close the only thing he knows is what the previous winning bid was.

There are lots of other forms of information in the world other than the orders in a closed auction.

Serious question: Why do people always forget this? This comes up ALL THE TIME when people talk about quantized auction times.

Your definition of a quantised auction is a bit different to mine. In my definition, you submit at the start and get an outcome at the end. You have no ability to modify the contents of the auction during the auction itself. I feel what you're describing is still a continuous auction... I mean if you can continually modify your order, or indeed submit at different times, at which point in time do you do your matching?

TBH I don't really understand what you are trying to say, but if you submit at "the start" and that is a single moment in time then participants have every incentive to wait until the last possible microsecond before that moment to determine their bids/asks.

There always has to be a moment when you lock in the bids, and a non-infinitescimal window before that time when bids are accepted. As long as that window has a start and an end, there will still be traders who write algorithms to deliver their orders at the final microsecond the window is open. 99.999999% of the time, nothing happens. But, in that one case where the articles of impeachment are released just before the window closes, those algos will pay for themselves a billion times over.

Separating the last moment when bids are accepted from the time when the auction closes has no practical effect, other than just a simple time delay.

First off a trader with the results of all previous auctions should be able to make some educated estimates about what the auction contains. Secondly prices are formed by incorporating information into the price. Supply and demand information is only one type of information. Say that we have an auction expiring at 10AM. At 9:59:59.500 Goldman's equity research desk announces that it is downgrading a security. Our trader can now in the remaining 500 milliseconds place aggressive asks to take advantage of what is likely to be a downward price movement. Other's in the auction with resting bids are now going to trade into a price that is moving lower. And arguing against colocation will not help here. There will always be traders who are able to place themselves at whatever the closest allowed point, whether inside the exchange or not.

Sure, but this is combined with all other information over that minute/hour. Suppose you had exactly 1 price exchange per day, now yes you might gain incite waiting for that last millisecond, but you have to weigh all other information over the day, approximate it's changes in the price, and then make a bid. Further if the ticks are say every 15 minutes it's easy to not disclose information on a tick boundary. (Simply because of relative time frames.)

Further if the ticks are say every 15 minutes it's easy to not disclose information on a tick boundary.

This is not true. Lots of things are happening in the world all of the time. You can't tell everyone to stop what they are doing every 15 minutes and wait for the stock market auction to close.

If ticks are 15 minutes, then 99.9% of all things don't happen in that last 1/2 second. Compared to the current situation where every single piece of news is a race for 1/1,000th of a second head starts.

So rather than a race vs time, it's a race to better interpret information. And considering that's basically the point of the stock market in the first place I would call that a net win.

"What advantage does he get from waiting till the last possible moment? "

He gets more advantage from outside information sources.

You are correct that you cannot gain info from the bids that are happening on THAT market, but you can instead get bid/price info on OTHER markets/auctions.

You seem to be describing a standard, continuous auction. Consider a more discrete time auction, like the opening auction described here: https://uk.advfn.com/Help/the-opening-auction-68.html - I'm not saying this format is perfect, but with modification, it may be reasonable

Actually the continuous cross is the special one from strictly the auction design perspective. I am very familiar with discrete auctions and their advantages and drawbacks. It turns out there there are many times to participate in such discrete crosses. Nasdaq for instance runs 3 or 4 every day in addition to the opening and closing crosses.

Just remember that from a market structure perspective, the continuous cross is the best for liquidity. All markets for things like derivatives try to move in the direction of a continuous cross over time.

As well as making the current bids invisible, you can reduce the auction times considerably - once per second, for instance.

Yes absolutely. The SETS order book (London SE) now runs an intra day auction at 12pm where it limits the amount of info available during the auction. It is a price forming auction - it aims to decide a fair price for the stock while limiting information to participants. See: http://www.lseg.com/markets-products-and-services/our-market...

> once per second, for instance.

That won't change anything, he who trades first still wins, HFT will still exist.

Sure there is. The idea isn't to quantize to arbitrarily long times, but to make it long enough compared to the time of information generation and more importantly, communication. Otherwise you may as well give up the notion of a common market. The number may not be a minute but it sure as hell ain't a nanosecond.

The real world is continuous. It cannot be quantized.

> The real world is continuous. It cannot be quantized.

I thought modern physics leaned more toward “the real world is quantized, but the quanta are small enough that things usually seem continuous to human perception.”

Heh. Perhaps that is true. But since we're all up here operating at speeds far slower than planck time I am gonna cheat and call it continuous. For the discussion at hand it is true enough.

What a fatuous comment. The real world satisfies laws of physics. Your fantasy of a continuous single market doesn't exist. This (information coherency) problem has been solved many times in engineering and the solution is quantization.

I do not have that fantasy. I'm sorry if I gave you the impression that I did. I have communicated poorly somehow if that is the case.

like 32 bit fixed point continuous?

Finance is NOT the real world. Lets not pretend that stock movements obey some sort of "universal laws" or such nonsense.

There are some good economic arguments to be made in favor of frequent batch auctions. Check out the following presentation by Eric Budish from the University of Chicago ( https://simons.berkeley.edu/talks/eric-budish-2015-11-19 ).

A couple of facts about the auctions:

- They often end at a randomized time in a given window

- They often cross a very large proportion of a day's trading

So in some sense the things you are after - hard to game, transparent auctions - are already in existence. If you're happy with waiting, you can pretty much ignore the continuous trading when you're buying/selling your shares. The exchanges I look at all have an opening and closing auction, and a few have a midday auction as well.

I'm literally coding a system that uses the auctions right now. To game the randomized end would not be easy, though there are a number of particular market models that open for it.

In general, the exchange system has gotten out of hand. There's a lot of weird rules that only make sense if you're told what they're for, and it will undermine confidence if they continue to grow. It's become a catch-22 though, as the markets need the market makers, and the market makers can't make money so easily without an advantage of some sort.

> I think this is well known.

I hadn't heard of this. But then again, I'm not an institutional investor and I don't place a lot of trades.

you have to target the disgruntled employees that already know the tricks and get them to flip.

You know, I've always thought this! Even better, get the guys that have been convicted of serious financial fraud to help the regulators out!

Or we can just ban anything insufficiently transparent and be done with it. After all, the point of a market transaction for pricing is its ability to convey transparent, accurate information to other market participants -- right?

In theory, sure.

How's that working out for us?

EDIT: Meta, I will never tire of hearing people be all, "But markets!" as if they aren't just as corruptible as any other human institution — if not more so.

Well, I agree that it's not working out very ideally for us. This is why I think all those hyper-complicated trades don't necessarily have much real economic value, and banning them is unlikely to be very harmful.

"But but the regulations impeding economic growth... "

> After all, the point of a market transaction for pricing is its ability to convey transparent, accurate information to other market participants

Arguably, the transparency for pricing that matters is only the final trade price, but I'd argue, in any case, that providing liquidity and actually pricing trades that occur is the more important function of markets, not providing pricing information.

OTOH, transparent pricing information brings some participants to the market, which increases liquidity; conversely, though, so does providing various dark trading vehicles. So, in the liquidity-focussed view, there is a trade-off and a balance to be struck.

Eh, somebody in their spare time figured out Bernie Madoff's scheme in the 90's and wrote them an explanation such that ignorance was no excuse. The SEC regulator marrying Madoff's daughter seems relevant.

Madoff didn't have a daughter.

Not to support GP's point, but he's probably referring to Shana Madoff, Bernie's niece, who was a compliance officer at the firm and married an SEC regulator. I'm not aware of any report "exposing" the Ponzi before it actually happened, but Shana met her husband during an SEC investigation into allegations of front-running by Madoff's fund.

> The real problem is that the traders are much smarter than the regulators

They're not smarter, they have massively more resources. The SEC investigates and takes to court the tiny, tiny portion of investors which are the most egregious and easy-to-prosecute criminals. This isn't like hackers fighting against security systems.

Remember when Sanders proposed a small fee on every trade on Wall Street to discourage high frequency trading and to recoup some value from the market? Remember how he was widely pronounced deranged for suggesting that there should be a fee associated with trades? How it would destroy the market?

Funny, that.

Common sense and basic regulations to protect people and economies have no place in this laughable economy. What you're describing, what Sanders proposed, would put a dent in the profits of some of the wealthiest, most powerful, and most soulless people in the world.

Of course it was shot down, that's one way to tell that he was cutting close to the bone.

would put a dent in the profits of some of the wealthiest, most powerful,

This is most certainly factual.

and most soulless people in the world.

This is a factually unsupportable adhominem. Sentiment that contributes to outrage on social media is a form of cultural pollution. People use it for short term gain, but it's a kind of externality which is tearing society apart. (FWIW, I dislike this situation as well.)

(Yes, this is obligatory: https://www.youtube.com/watch?v=rE3j_RHkqJc )

Outrage is a form of cultural pollution? So no one should ever be outraged by peoples actions? Or only in the right circumstances? Who gets to say when its justified, if ever?

All of those questions apply to air pollution.

That's true. I guess the difference is that outrage has a function, and can have a utility, so I don't think it's a good metaphor. Viewing it as just 'pollution' implies that it has no value. It's only 'pollution' to those who don't agree with the outrage.

That's true. I guess the difference is that outrage has a function, and can have a utility, so I don't think it's a good metaphor

CO2 is the best analogy. There needs to be a certain amount for the utility. Too much and too little are detrimental.

Viewing it as just 'pollution' implies that it has no value.

This is an all-or-nothing fallacy. It's the amount produced which is the issue in the analogy. In reality, there are also finer grained quality issues.

To further demonstrate the application of your fallacy, I would agree that there are problems with under-prosecution of certain crimes.

https://www.youtube.com/watch?v=rHMGbtGGdbQ

However, when the outrage which has reached a fever pitch such that people start calling for abrogation of Innocent Until Proven Guilty based on inherent characteristics, something has gone wrong. Our culture has known, since the times in which the Magna Carta was written, that the protection of the individual from arbitrary imprisonment and prosecution is essential to prevent totalitarian abuses of power.

Outrage is easy to over use, its over-use is readily rewarded and such over-use is clearly everywhere, even despite the fact that it's only the excesses of the "other side" that are easily discerned.

>This is an all-or-nothing fallacy

Not really. Pollution is uniformly unwanted by definition (without you changing the goalposts to CO2, which is naturally occurring, and the naturally occurring CO2 would not be considered as pollution, whereas human created CO2 would). In fact, let's stick with the wikipedia definition:

"Pollution is the introduction of contaminants into the natural environment that cause adverse change."

Outrage, on the other hand, may be very much wanted, or even required. I'm not defending ALL outrage. I'm defending that some outrage may occasionally be warranted. You saw an all or nothing fallacy where there was none. To recap -

Argument: Outrage is cultural pollution.

My Response: All pollution is unwanted, some outrage may be occasionally wanted or warranted.

Your response: Saying pollution has no value is an all-or-nothing fallacy!

The worst kind of bad reasoning is the false accusation of a fallacy. Because the person making that claim should know better.

In fact, let's stick with the wikipedia definition:

"Pollution is the introduction of contaminants into the natural environment that cause adverse change."

This is either an honest mistake or a pedagogical trick you're pulling. In the general point, I mean pollution in the sense people mean when they say something like "noise pollution." "Pollution" in my analogy (which isn't the same referent as above) would be excess CO2 -- in large enough quantities this is a bad thing, and everyone should know that fact. The validity of the underlying point really has nothing to do with your nitpick. Just substitute "bad thing" for that word in your head. Your whole argument vanishes, and my point remains.

Your response: ... is an all-or-nothing fallacy!...The worst kind of bad reasoning is the false accusation of a fallacy.

You do have an all or nothing fallacy, and your falsely claimed refutation is actually an irrelevant language nitpick. However, I don't find that a quarter as disturbing as the seeming attachment you have to outrage as some kind of tool for convincing others. That's not convincing. That's coercing.

Here's a primer on intellectual hygiene you might enjoy.

https://www.ribbonfarm.com/2017/03/02/the-limits-of-epistemi...

Outrage is a form of cultural pollution?

On the 21st century internet, an alarmingly large portion of it is. Some outrage is justified, clearly. However, the incentive structures online are so extremely skewed in favor of producing outrage, we need a new form of skepticism. I was once outraged by the notion, "Pics, or it didn't happen!" But on reflection, I realized that the new incentive structures made the rewarding of internet fakery far too likely. Young people, realizing this, reacted in a rational way!

As with "pics or it didn't happen" this is going to be hard for many to hear, but there needs to be a more rational approach online. As it is, the lowered bar for producing online commentary and media has meant a general drop in quality, and this extends to commentary and media produced for activism and activism itself.

Fees on trading would not meaningfully discourage high frequency trading and would just be passed on to buyers and sellers of stock in the form of larger bid/ask spreads.

Something like a minimum holding time is a more reasonable approach. However, there's not a clear cut answer in any case.

Before you start talking about what would be effective in reducing the amount of high frequency trading you have to make the case that reducing HFT is a good goal in the first place. This is a challenging case to make.

Why is it challenging? What conceivable valuable function does front-running serve?

High frequency trading is not front running.

It's becoming truly kafkaesque how often this claim comes up on Hacker News.

Or kafkaesque how often these separate concepts are conflated in popular usage.

First you have to tell me what you think front running means. Originally front running had a very specific definition: Someone who has a fiduciary responsibility to me cannot use my trading interests to execute their own trades before they execute my trades. This sort of thing is clearly wrong because it can result in someone I've hired delivering subpar execution prices to me.

However the term has expanded a lot over the years to mean a lot of different things to a lot of different people. So what does it mean to you?

You don't know what front-running means, and HFT's aren't doing it.

These need to be discussed together. Otherwise, the defence of HFT tends to cite the economic damage caused by measures to reduce it.

Anyway, I agree with you. I don’t see a significant downside to using a small transaction tax or one of the other suggestions. The real hard question is about the benefit or harm of HFT itself.

I haven’t heard a decisive argument yet, but I would say that the “liquidity defence” of HFT is in unconvincing to me. I don’t see how liquidity can add value past a certain point.

For the record I do actually see a downside to a trading tax and do not think we should have one. Trading is a useful activity. There is no need to single it out separately from all other forms of economic activity for a special tax.

Trading is probably essential, it definitely is given our economic system which involves trading. In my opinion, HFT does easily merit singling out for evaluation. It is a relatively marginal economic activity, but accounts for a large portion of transactions, mostly in highly liquid markets purposely designed/evolved to be highly efficient like stock markets and other tradable securities markets. I don't think pure theory economics can say much about what happens at extremes like that. Transactions are prices and information. If most transactions are based on HFT strategies divorced from anything even remotely related to an understanding of the underlying asset, who knows what wheels fall off the theoretical model and/or reality. It's worth looking at, at least.

As I said though, I agree with you that there's an onus on those proposing a HFT tax to convict it convincingly. I don't think this has happened yet. The argument can't be "weird and scary." I don't see the liquidity defense. How can liquidity beyond a certain point be meaningfully more useful, but that's not a conviction. In my mind, it rests on how HFT impacts economic fragility, and increases the likelihood or impact of busts.

We're talking about something like a 0.02% of equity tax on HFT specifically or lower if it's going to be everyone. That does not impede the ability to price in real information.

An artificial restriction on trading is not unlike natural ones. Did we have liquidity issues when trading in and out multiple times within tenths of seconds was impossible? I don't think the cost is high. The risk is ..unproven.

Did we have liquidity issues when trading in and out multiple times within tenths of seconds was impossible? I don't think the cost is high.

Yes. Spreads were a dime (or more). Now they're a penny.

Your argument boils down to they're trading too much, it scares me, make them stop. That's not a logical argument. Whether they're economically beneficial or not is simply not relevant, people have a right to buy and sell stocks for whatever reason they want whether it's beneficial to the market or not. They shouldn't have to defend themselves for buying and selling stock. HFT'ers simply do this a lot, so any objection against them is an objection to the either the speed at which they do it or the volumes they do it in, neither of which should be relevant to you. You shouldn't be able to tell someone they can't trade because they're too fast at it.

Said proposed tax is nothing more than people who can't compete trying to punish those who can because they don't understand why they're losing.

HFT has improved the market for everyone involved, spreads are lower, liquidity is higher, everyone pays far less for trades than ever before. There's no reason at all to regulate it that isn't simply fear based.

HFT should be presumed innocent until proven guilty; the question isn't to defend HFT, it's to show it's actually a problem to begin with. Those wanting to regular HFT really have no argument that makes sense other than "we don't understand that thing so it's bad".

Proposing a tax on trades is punishing those presumed guilty without a lick of actual evidence they are. HFT don't need to prove they're good, they're just traders making trades in the market like anyone else, that they do it faster than a manual trader doesn't make them bad. To try and regulate them should require an actual case be made against them and all such cases I've seen so far are completely irrational emotional arguments by people who just want to point a finger at someone to explain why they're no longer able to compete.

High frequency trading is basically capitalizing on inefficiencies in the market. Ideally, those inefficiencies would be addressed directly, and the benefit of a lower friction marketplace would be distributed among all participants. Then all the energy going into HFT could be redirected to something more productive.

Yes, and this is happening. The low hanging fruit in HFT is effectively gone, and you can no longer compete on just speed. The industry has been shrinking since the early 2010s.

http://www.thebigquestions.com/2014/04/21/high-frequency-ren...

This seems pretty convincing to me. The argument is that, based on the amount that firms are willing to spend on fiberoptic cables to perform hft, they put an extremely high value on hft. On the other hand, reasonable back-of-the-envelope calculations show that the social benefit of making the trade slightly faster are much less than the private cost. This indicates that almost all of the private benefit from hft comes from value accruing to the hft firm at the expense of other hft firms. We therefore expect to see overinvestment in hft.

What about this do you find objectionable?

HFT has drastically reduced the costs of trading and improved the returns of nearly all investors especially retail investors like you and me. These benefits are huge. This blog post ignores them.

Here is Vanguard's CEO on the topic:

http://www.cnbc.com/2014/04/25/vanguard-chief-defends-high-f...

That blog post also makes a mistake of claiming that the advantage of these sorts of fiber optic cables is to let people complete their trades faster. That is not the case. They enable people to execute their trades at better prices. The way he is looking at this issue is almost silly.

The entire argument is extrapolation based on the cost for a firm to lay optic cable, are you serious?

Do we really measure the social good of Google by how much capital they put into their fancy server clock syncing (the exact name of the project escapes me)?

How did we get to a point where "reasonable back of the envelope calculations" are what people seriously consider when trying to develop economic policy for the biggest economy in the world?

> Something like a minimum holding time is a more reasonable approach. However, there's not a clear cut answer in any case.

Why do you want to reduce high frequency trading?

flash market crashes.

Okay, fair enough. But do you believe these happen often enough, or will increase in the future, to outweigh the improvements to liquidity provided by HFT?

http://www.investopedia.com/articles/markets/012716/four-big...

This article has a lot of the reasons. I think loss of confidence in Market Integrity is the most important one.

That article has two reasons, though it spreads "amplification" out among several factors. In fairness, I do agree with the problem of flash crashes, whether they are caused by algorithms moving too quickly or too similarly.

I'll contest the "confidence in the market" hypothesis, however. As more investors move to index funds, I don't believe "confidence" as defined would have any significant impact by increasing or decreasing, because fewer participants overall will be actively engaged.

That's a good point about index funding.

I do think if confidence in the market gets low enough it is possible that people will stop investing all together or invest less than they would have. But that is me sidestepping the issue a bit.

I think higher spreads would “slow” highest frequency trading. They cross the spread many more times per dollar than regular investors to it costs them more.

High frequency trading only works inasmuch as transaction costs are low, at least that’s my understanding.

> I think higher spreads would “slow” highest frequency trading. They cross the spread many more times per dollar than regular investors to it costs them more.

If I understand your thesis correctly, you want to decrease liquidity in order to reduce the speed at which high frequency trading can be executed?

I'm not following your point about transaction costs - or do you mean that you'd limit HFT by reducing liquidity, which in turn would reduce their volume, reducing their trade discounts?

I don’t want anything in particular, just speaking to the earlier post (speaking on a Bernie Sanders proposal, I think).

The “thesis” is that by putting a small tax on transactions, you will decrease the profitability of trading strategies that involve trading securities many times therefore discouraging them.

Liquidity is a trickier concept. I’m not sure if it has a consistent measurement definition. If you define liquidity as turnover or something close, then lowering trading frequency lowers liquidity by definition. How that affects liquidity in the practical sense for a “regular” investor is an uncertainty. It’s hard for me to imagine that a trader that holds stock for an average of 1 year will have more trouble getting in or out of Google stock, but maybe I’m wrong.

I think that's about half right. A 0.5% tax on all equities trades, which is what Bernie was proposing, would put a bunch of HFTs out of business basically overnight and at the same time dramatically increase costs for investors via 1) the tax (obviously) 2) wider spreads 3) reduced liquidity.

4) increased volatility 5) less efficient price discovery

The HFT shops put millions of dollars into research to attempt to ascertain correct prices (e.g. ETF pricing, derivatives pricing, etc). If they are disincentivized from trading in the equities markets, they will no longer be a conduit of relevant pricing information from other global markets into the equities markets. That means investors (big Wall Street firms catered to by IEX) and retail (you and me in our individual accounts) are more likely to be trading mis-priced markets.

What is the value of trading at increasingly marginally more accurate prices? And more importantly, what is the cost?

You seem to take it at face value that trading at accurate prices is an unalloyed good. But for the extremely overwhelming majority of retail investors — whose only sane strategy is buy and hold — buying at a few tenths of a percentage points closer to the most-accurate possible price is worth nearly nothing (and has negative worth half of the time, practically by definition).

On the other hand, Wall Street has been raking in tens if not hundreds of billions in profits from this service. What value we get from more accurate pricing may very easily be offset by these costs a hundredfold.

You are drastically overestimating how much money HFT market makers earn. Virtu Financial (one of the biggest firms in this area) earns about 100 million a year.

I think you are also underestimating the costs to retail investors to not getting accurate pricing. Shaving a few tents of a point off of every trade will have a huge effect on the lifetime earnings of an individuals.

> earns about 100 million a year.

They made 147M in the first quarter of this year. 197M in the first quarter of last year. They might only make 100M/year after costs, but that doesn't represent the 600-900M they take from the market.

OK, so gross-not-net, Virtu is making about $2M/day. The value traded on a typical day in US stock markets is north of $100 billion / day.

HFT is rounding error.

> Virtu is making about $2M/day. The value traded on a typical day in US stock markets is north of $100 billion / day.

You can't estimate it that way as they don't win or profit on all their trades. At it's height HFT was estimated to responsible 15-25% of daily volume by best guesses (it's some what obfuscated.) I'd be surprised if it was less than 5% today. Not just Virtu of course - all players big and small.

Why does this change whether or not you can estimate it that way? The tick size in the US is at least $0.01 for reasonably priced names so it's completely unrealistic to expect an HFT to make money on every trade.

When people talk about making $0.0001 per share that's their ex-ante expectation. It accounts for the fact that you're not going to make money on every trade.

That's not the point. The point is that the industry is vanishingly small compared to Wall Street proper, yet it has an outsize target on its head due to FUD and emotional appeals like the ones presented in the article.

Furthermore, in exchange for "taking" that money from the market, they enhance liquidity, which is directly helpful for price discovery and facilitating trading among both retail and institutional investors.

People are continually moving the goalposts in this thread and others like it. If you're going to talk about Wall Street and fraud, high frequency trading is not the place to start. All of the legitimate arguments against high frequency trading have nothing to do with fraud, they have to do with the dangers of runaway algorithmic trading that coalesces into the same market movements.

But we can't reason about that issue while half the people talking about HFT (almost none of whom actually have experience with trading whatsoever) still think it's front running, or believe it constitutes some sort of fraudulent con over "the little guy."

Liquidity is willingness to make a trade others aren't. To be useful, it has to linger in the order book for a long time. If you're winning a race by milliseconds, you are trying to interpose yourself into a trade that was already going to happen that day.

> If you're winning a race by milliseconds, you are trying to interpose yourself into a trade that was already going to happen that day.

This is an inaccurate framing of how high frequency trading propagates liquidity in an otherwise illiquid (or strictly less liquid) market. The claim is not that liquidity is contributed on a strictly trade by trade basis, but rather than the low-latency activity has meta-reactive effects owing to enhanced price discovery that increase overall participation by drawing in other traders at different time resolutions. For example, where there may a stagnant order book on one equity (and consequently, few human traders able to fulfill orders without significant pricing penalties), the same order book may draw in competing market makers. They attempt to predict the next price movement - some win and some lose on the immediate sequence of trades, but the consequent activity narrows the bid/ask spread by heightening local participation in the order book and improving the pricing confidence. This has practical ramifications for "human" time resolutions, because the human traders now have a better opportunity to fulfill orders without overpaying. This in turn reduces overcautious traders from participating, and so on and so forth.

For what it's worth, your line of argument has been rehashed for years now on Hacker News, going back to when Chris Stucchio wrote his HFT apologia. Instead of lazily linking to that thread, I'll do one better by walking through research on the subject. Fortunately there is a handy paper that explicitly examines the question, "how does the interaction of these traders in the millisecond environment impact the quality of markets that human investors can observe?"[1] The data is constructed using NASDAQ TotalView with equities in the S&P500 in periods of varying volatility. Both reactive and periodic trading algorithms are reviewed.

Here are a few critical passages:

By tracking submissions, cancellations, and executions that can be associated with each other, we create a measure of low-latency activity. We use a simultaneous equation framework to examine how the intensity of low latency activity affects market quality measures. We find that an increase in low-latency activity lowers short-term volatility, reduces quoted spreads and the total price impact of trades, and increases depth in the limit order book.

IV.B. Results Panel A of Table 4 presents the estimated coefficients of the pooled system side-by-side for the 2007 and 2008 sample periods. First we note that the two instruments have the 25 expected signs and are highly significant. Specifically, the coefficient a2 indicates that when liquidity off NASDAQ is higher, our NASDAQ market quality measures show higher liquidity and lower volatility. Similarly, the coefficient b2 is positive in all specifications, indicating that higher low-latency activity in a specific stock in an interval is associated with higher low-latency activity in other stocks on the NASDAQ system. Second, the estimated b1 coefficients tell us that low-latency activity is attracted to more liquid and less volatile stocks.

The fact that low-latency trading decreases short-term volatility and contributes to depth in the 2008 sample period where the market is relentlessly going down and there is heightened uncertainty in the economic environment is particularly noteworthy. It seems to suggest that PA activity creates a positive externality in the market at the time that the market needs it the most. Panel B of Table 4 presents roughly similar results from the estimation of the system with SpreadNotNasi as the instrument for market liquidity.

It is possible, however, that the impact of low-latency trading on market quality would differ for stocks that are somehow fundamentally dissimilar, like small versus large market capitalization stocks. Table 5 presents system estimates in subsamples consisting of four quartiles ranked by the average market capitalization over the sample period.22 There is not much pattern across the quartiles in the manner low-latency activity affects short-term volatility in the 2007 sample period. The picture in the 2008 sample is different: It appears that during more stressful times, low-latency activity helps reduce volatility in smaller stocks more than it does in larger stocks.

Lastly, Table 6 shows summary statistics for the stock-by-stock estimations. The results suggest similar conclusions concerning the effect of low-latency trading on market quality. In particular, an increase in low-latency activity decreases short-term volatility, decreases quoted spreads, and increases displayed depth in the limit order book. This is true both in the 2007 and 2008 sample periods.

_______________

1. http://people.stern.nyu.edu/jhasbrou/Research/Working%20Pape...

Do you think Virtu takes more or less money from the market than the market makers they displaced?

If one is employing a buy and hold strategy, then it doesn't make sense to perform buy operations for every pay cheque. At the level of a retail investor, making buy trades once a quarter or once a year will do far more for a retail investor looking to have money on market making operations.

A few tenths of a percent is on the order of less than $100/year assuming that a retail investor invests the maximum amount allowed inside a 401k each year (ignoring for a second that typical 401k plans do not permit investing directly in individual stocks and also ignoring catch up contributions for older folks). It's just not a significant amount of money at the level of an individual retail investor.

Whats wrong with them making money from their efforts.

Completely delusional about Hedge Fund profitability.

Automating the work done by human traders, market makers, and specialists has not only facilitated decimilization resulting in smaller spreads, but also allowed trading fees for the retail investor to reach historic lows (the worst you'll find at a retail brokerage nowadays is about $8/trade, far better than the $35/trade), with many even being free. Wall Street had always been raking in billions in profit from market making activity, if anything that profit margin has been significantly reduced by automating market making.

> ...also allowed trading fees for the retail investor to reach historic lows (the worst you'll find at a retail brokerage nowadays is about $8/trade, far better than the $35/trade

The average retail investor should not be making enough trades for this to matter.

Bringing down the price of trades like this only makes it cheaper for the suckers — day traders — to think they're playing the game. It is of marginal utility for the average retail investor.

That's not a fair assessment. Let's say you are doing well and actually investing $1000 per month into the market. Now let's say you are diversifying into three index ETF's, and you buy those quarterly, so that's three trades a quarter on a total of $3,000. If the commission is $35/trade, you are being taxed 3.5% of your investment up-front to get into the market. If the commission is $4/trade (a more typical fee these days), then the up-front loss is 0.4%. You'll pay those same fees again on the way out, and assuming that you sell on the same schedule, that's now 7% vs. 0.8%.

Yes, there's some nice compounding in between, assuming that you buy and hold with no further trades. But, there's a wide gulf between "day trading" and active stock-picking. Assuming that your average hold time is 5 years per stock, you're still going to rack up a lot of commission costs at $35 per trade.

There's no real problem at all to professional investors, they will figure out how to trade at the price they want regardless. It mostly hurts retail investors who rely significantly more on market signals to be correct. Professional investors almost prefer inefficient prices as it provides more opportunities for arbitrage against dumb money.

How is mispriced markets a problem? If institutional investors are buying and selling at roughly the same rate, and the mis-pricing occurs in either direction, there will be more noise, but over time it would balance out. Sometimes you'll pay 1% too much, sometimes 1% too little, but it would balance itself out in the long run.

Am I missing something here?

Your assumption about institutional investors buying and selling at roughly the same rate is wrong. When one firm decides that Microsoft is overprices, a lot of firms are probably going to decide the same thing. Then it's a race to see who can extract the most value out of the market before the price corrects.

Not really missing anything, but if you're willing to accept buying/selling 1% away from the "ideal" price, it's hard to complain about an HFT making a penny because you can't back-off when the market is moving.

> Am I missing something here?

Yes. By paying relatively small amounts to high frequency market makers in return for enhanced liquidity and price discovery, you won't be overpaying by 1% (or more). I also challenge the idea that it would just "balance" itself out, in the absence of evidence supporting that thesis. In actuality you'd likely just amplify the costs you already have and either fill fewer trades or have higher costs for doing so.

Choosing to lose $1 due to low liquidity instead of a few cents due to market makers is both petty and nonsensical. There are legitimate arguments against HFT, but they don't begin by trying to reinvent economics such as to de-emphasize optimal price discovery.

as far as I can tell, HFT shops aren't producing anything tangible. Considering almost nothing is as zero-sum as a stock market at sub-second timescales, their revenue is exactly equal to the costs for other market participants. For Virtu Finance, mentioned by another comment, that's around $800 million/year.

That's not really that much in the scheme of things, but it's only one company, and I doubt the other investors would be willing to spend similar on insurance against volatility.

That's such bs--as if the markets couldn't determine prices at anything but sub-microsecond velocity. No, HFT is the ultimate rent seeking.

> HFT is the ultimate rent seeking

It really isn't though. It's been maligned as part of a smear campaign by the actual rent-seekers, Wall Street proper, as other commenters have noted.

It's a bit hard to say to be honest. The wider spreads should make a lot of HFT strategies more profitable. Profitable enough that they can successfully absorb the costs from the tax and not go out of business. You are probably right that this wouldn't be true in all cases though.

Here are some numbers. 0.5% of a $50 stock is $0.25. So to break even on the tax alone you need to sell $.50 higher than you buy. That alone will blow out the spread any market maker is able to quote at. The other problem is that now scratching (you buy at the bid and now it looks like the price is going the other way so you aggress and sell back into the bid for no profit) is also extremely expensive (you lose $.50 per share on a $50 stock just scratching). That's going to really kill your profitability. Maybe someone could figure out how to make it work, but it would be an extremely painful regime for market makers.

Are there really people proposing a 0.5% tax on trades? Jesus. That's 1-2 orders of magnitude bigger than I thought we were talking about. That's crazy!

0.5% on equities was the Bernie proposal. Here I'm assuming that both sides pay 0.5% but even if it's half that (each side paying 0.25% or only one side pays) the numbers are crazy.

Easy fix - only tax the takers. I agree the 0.5% is a bit steep on both sides. If only to takers and makers aren't taxed, seems like it could work. Would also probably add a ton of liquidity to the markets.

HFTs are almost entirely makers; retail and institutional do almost all the taking, so taxing the takers may not have the desired impact on the industry.

How is that a fix? A market with huge displayed sizes but no trading (because nobody wants to pay the massive take penalty) is not a "liquid" market.

> Would also probably add a ton of liquidity to the markets.

What? No it wouldn't. You are disproportionately rewarding makers in this scenario. You would find plenty of listed orders, which somewhat looks like liquidity, but it would not be a liquid market. The end result would be a market that is actually less liquid because no one wants to fulfill orders. It would be utterly lopsided.

Does it dramatically increase costs for investors or speculators?

How many times do you trade a year? Actually perform trades? Even including mutual funds, I think it's < 100 yr.

Those mutual funds trade on your behalf every day.

What you call "recouping some value from the market", others call "stealing some value from the market".

I've always thought a better solution would be a voluntary market "quantization" at some reasonable human-scale time frame. It would be a little random (to avoid gaming), so all put and asks get resolved "about every hour".

There would probably have to be a law to prevent people from running markets at faster time-scales on top of this.

This has existed in various forms for more than 10 years. Nasdaq I believe has 3 or 4 discrete intraday crosses in addition to the opening and closing cross. You don't want it to be random since participants need certainty about when the cross will occur. In addition there are a variety of venues that allow block trading in various forms that market participants can use.

This solves fewer problems than you might think.

1) Let's say that trades are resolved at time X. Participants have every incentive to submit all bids/asks as close to time X as possible (microseconds possibly).

2) How do you handle a mismatched number of bids/asks at a given price? Resolving this difficulty without creating bigger problems than the problem you were trying to eliminate is challenging.

3) I'm just a regular guy who wants to buy $1000 of stock as part of my monthly savings plan. With an up to date market I can just buy and sell at the market price and not worry about it. But now I have to be afraid that there has been some big news event in the past hour that will make this hours price much different than last hours. One of the biggest services markets provide is up to date pricing information. Why do you want to take that away from me?

1) You only get to see and make a potential buy/bid for last hour's bid/asks, only.

2) The system itself resolves it. Either it prevents you from making a buy/bid for something that was already "fulfilled" (though this would leak information). Or it accepts them sequentially, and refunds you at the end of the hour.

3) We're not taking that away from you. We'd be taking it away from everyone. You are welcome to see last hour's trades and try trade on it.

Additional points:

2) Another option. If your order didn't make it, the system can simply let it "stay" on on the system to be fulfilled in the future. If someone wants to take you up on the offer you made, they'll do it in the next hour, or any subsequent hour.

1) Yes, but as I've said elsewhere there is a lot of information in the world other than the last hour's orders (why do people always forget this?)

2A) Preventing unmatched bids/asks will not work. Remember that you are starting at zero. All bids/asks are unmatched when you start with an empty order book. If you relax this some then yes you will leak and you are back to where you started.

2B) If you handle things sequentially then you have reintroduced a speed imperative. I, again, have an incentive to go fast to get first in line.

3) Yes, of course you are taking it away from everyone, but that matters more to me (an unsophisticated guy with no real time market research) than it does to BIG_HEDGE_FUND_GUY who does have such things and can, more easily, figure out what the market price should be without the help of the market. You are putting me at a disadvantage and him at an advantage. Is that really your goal?

There are some exchanges that have tried this, but it's hard when the auction price cannot result in a price better than the NBBO (which is just one way the Reg-NMS has stifled innovation). They are having more success with this in Europe: http://www.businesswire.com/news/home/20150906005008/en/BATS...

A variant on the Tobin Tax https://en.wikipedia.org/wiki/Tobin_tax

Your comment, while snarky, isn't a refutable statement. As it stands you seem to be endorsing some position, which I'm inferring is in favor of Sanders and maybe in opposition to HFT. I can't really tell what you're getting at precisely though.

> Remember how he was widely pronounced deranged for suggesting that there should be a fee associated with trades? How it would destroy the market?

Who said this, specifically? What is your point in bringing it up?

> Remember when Sanders proposed a small fee on every trade on Wall Street to discourage high frequency trading and to recoup some value from the market?

I'm getting the sense that you'd be in favor of this - can you tell me why, in your own words, you believe we should be trying to "recoup value" from the activities of high frequency traders?

It's mostly about how skimming a tiny bit of cash off of every trade is perfectly good for the exchanges to do to john Q public, but heaven help us if real traders have to put up with that.

I have no idea if Sanders' plan was good or not, I'm more grousing about how critically important principles of how the market is supposed to work seem to vary depending on whether you're talking about consumer-level investments vs. institutional investors.

I still don't get what you're trying to say.

Who is "skimming a tiny bit of cash off of every trade"? Do you buy into the notion that HFT is somehow so fast that it can travel back in time and jump ahead of orders that have just executed?

Why is it OK for the exchange to charge fees from every trade motivated by profit, but if the government does it to redistribute, it is somehow labelled as "market distortions"?

Either fees on every trade distort the market, and in that case the exchange is distorting the market, or they don't have appreciable effects and the government can levy that tax.

My opinion is that this a political opinion, and pretending that this is an economic concern is just a smokescreen.

The exchange owns the exchange, and provides the service of running the exchange, and can set whatever fees they like, and people are free to use other exchanges that set different fees.

The government already takes (more than) its fair share, from existing taxes. It has no right to also levy a tax on every trade on every exchange.

I'm not OP but I sympathize:

> Who said this, specifically? What is your point in bringing it up?

Sanders. The point in bringing it up is to point out the irony that Wall Street already does this behind the scenes, but people called the Senator crazy for proposing it.

> I'm getting the sense that you'd be in favor of this - can you tell me why, in your own words, you believe we should be trying to "recoup value" from the activities of high frequency traders?

Shortest terms I can put this is: our country needs the money that Wall Street siphons off of the economy. We need more government money to educate, house, feed, and care for people. High frequency trading, unlike traditional investment, is not a "mom and pop" thing, it's a tool only accessible to the wealthy to enrich themselves. Thus, we should disincentivize it in order to generate tax revenue and discourage practices that are not accessible to shareholders.

All you did was move the goalposts...

> but people called the Senator crazy for proposing it.

Who called him crazy? That is my question. And the corollary to that question - why do we care about this party, and why is it relevant to the point? A lot of people say plenty of idiotic things, but that doesn't mean they have any real authority in the matter.

I'm looking for precision and an understanding of why it's relevant.

> Shortest terms I can put this is: our country needs the money that Wall Street siphons off of the economy.

This is an emotionally loaded claim, and it's also not axiomatic. How precisely does Wall Street "siphon off of the economy" without providing value in exchange?

> High frequency trading, unlike traditional investment, is not a "mom and pop" thing, it's a tool only accessible to the wealthy to enrich themselves.

High frequency trading is a very small industry compared to all the types of trading that occurs on Wall Street. It has outsize publicity for a variety of reasons, many of which circle back to FUD.

Moreover, every single type of trading firm is only accessible to the wealthy - that is very nearly what defines institutional trading. You are no more going to start a discretionary hedge fund than you are going to set up colocation and an FPGA for high frequency trading, which means that HFT is not nearly alone in being inaccessible.

You're not actually explaining your point here. You're just repeating claims without defending them. On the contrary, market makers (who almost exclusively use HFT these days), provide liquidity to the market, just as the institutional investors in charge of endowments and pension funds provide value for retail investors' retirement savings.

> Shortest terms I can put this is: our country needs the money that Wall Street siphons off of the economy. We need more government money to educate, house, feed, and care for people. High frequency trading, unlike traditional investment, is not a "mom and pop" thing, it's a tool only accessible to the wealthy to enrich themselves. Thus, we should disincentivize it in order to generate tax revenue and discourage practices that are not accessible to shareholders.

If I understand you correctly, there are a couple issues you're bringing up:

  1. Wall Street "siphons" off money from the economy.
  2. The US government needs higher revenues.
  3. HFT, as distinguished from "traditional investment," only benefits the rich.
  4. 1-3 are problems whose best solution is to tax HFT specifically.

First, I think it's important to distinguish between "Wall Street" and "HFT." "Wall Street" is composed of the largest banks in the world (Goldman Sachs, Morgan Stanley, Bank of America, etc.). Wall Street owns trillions of dollars worth of assets and has income in the 100s of billions of dollars each year. All HFT revenue in the US is estimated at less than $2 billion per year [1].

As for "siphoning" - do you mean to imply that making money by buying and selling a financial asset is somehow cheating someone, unfair, or something else? Or do you think that by sometimes functioning as "middle-men," HFT and Wall Street's are somehow cheating someone, unfair, or some other bad thing?

What is "traditional investment?"

What do you think of the fact that Wall Street and HFT firms combine to pay many billions of dollars in income taxes?

[1] https://www.dbresearch.com/PROD/DBR_INTERNET_EN-PROD/PROD000...

You are aware that HFT is essentially dead? It isnt profitable anymore and most of those firms are failing. Trading situations that allow for easy outsized profits like that disappear very fast.

Is this true? Any links? Not doubting you, just interested.

Source is I work in the industry. Look at the price of VirtuFinancial stock, who was at one time a premier HFT firm.

The amount of disinformation surrounding HFT is staggering.

Unrelated, but I have a question for you. I've been unable to find a good answer for how exactly HFT benefits the average person.

Stuff like antibiotics, electric lights, refrigeration, washing machines, phones, computers all have all led to direct and immediate quality of life improvements. Often on the order of a tenfold improvement for that activity, and they are easily within reach of the majority of the population.

What would become 10x worse for the average person if HFT were to vanish overnight? Would mortgage rates massively spike? Would bond rates plummet? Is there some quantifiable financial thing that would regress to whatever terrible situation we were in back in 1990 before HFT was substantial?

Right now the newest investment strategies are taking alternative data sets, like mining website data, and using it to predict stock prices. What value does that provide to the average guy? Nothing but market efficiency. It's just how the industry works.

Right, but how does "market efficiency" trickle through the rest of society? I can't even find anything that quantifies market efficiency, let alone anyone who has tracked it over the years.

Increases market confidence, which in turn promotes economic activity.

Too vague!

Since 1990, US population has increased 30% while the (inflation corrected) GDP has increased 90%. So something improved. I'm going to (arbitrarily) say it was computer literacy, since home PC ownership went from 15% of households to 85% of households in that time period.

What can you counter with to say that the improvements were from market efficiency? If we never had HFT, how much lower would the GDP be?

To be completely honest with you, I think you are trying to back someone into a logical corner on something you dont understand. You are arguing one thing and missing the whole picture.

The proper allocation of capital to working business. Are you arguing against the existence of the investment industry? How would capital move around, how would growth occur?

I'm arguing against the existence of things that can't be measured. How is "market efficiency" different from Carl Sagan's "Dragon in my Garage"?

It sounds like you saying that private companies that don't participate in the stock market are incapable of managing capital or achieving growth.

Market efficiency means everyone gets cheaper trades with less slippage and less money going to middle men. All of society benefits from market efficiency. If you have a 401k or any stocks of any kind, you benefit from a more efficient market. Efficient means less money is lost in the transaction to middle men like traders. HFT make less per trade than old school manual traders did because they've out competed them and lowered the spread which means you get a better deal when buying and selling stock.

This is more specific and useful. But it is still fairly hand-wavy. How is this efficiency measured? Is "market efficiency" really nothing more than the average USD to conduct a trade? Or the number of microseconds? (In either case it isn't an efficiency, efficiency is always a unitless number between 0 and 1.)

401k is a great indicator, but I'm not sure it makes HFT look very good. The median trade time is more than 1000x faster than in 1990. But people's 401ks are not doing 1000x better. Did we hit a point of diminishing returns long ago? If so, is HFT pointless?

I feel like you are conflating HFT with electronic trading. Electronic trading can be HFT or slow, either way it is cheap and cuts the middleman out.

If you want something more specific, how about something like the following -- for a fixed number of shares N that you want to buy (or sell), the average "cost" (as measured by the difference between the price you pay and the midpoint between the best bid and offer) of performing that transaction has gone down (true whether you buy all N shares in one order or across a block of time T).

You're asking all the wrong questions. HFT is just trading really fast, we live in a free country, some people want to trade really fast because they've found it profitable to take on the risk of doing so, it doesn't matter one bit if you or anyone else thinks it's pointless, you have no right to tell someone else they're trading too fast. They aren't hurting you, they're making your trades cheaper, they're making it easier for you to get into and out of any position you're in by being there to buy or sell exactly at the moment you want to buy or sell, and you're complaining about what? You don't even know.

You seem to have this backwards notion that they need to justify their existence, they don't, they're just traders executing their rights to buy and sell like everyone else. It's those like you seeking to regulate HFT that need to justify yourselves. You don't even understand what HFT is really as you're asking basic questions like what does market efficiency mean and does HFT help it, and you think you're in a position to question someone else's trading habits? Really? HFT doesn't need to look good, those of you trying to punish them need to show some actual evidence they're doing something bad, but they're not and you can't.

Efficiency in a market means things are priced accurately and you're not getting ripped off when you buy or sell; if you find that vague and hand wavy, well, sorry but that's what it means and you should be able to understand that without further explanation. HFT traders make the price more accurate benefiting you and everyone else, they don't need to justify themselves, you need to justify your witch hunt against them.

HFT is a form of electronic trading, I'm not conflating them, they're just different forms of using tech to trade and there's valid reason at all to single either of them out as bad. HFT are market makers, they're providing you and everyone else liquidity for a vastly smaller fee than you've been provided it ever before. So say thank you to HFT, enjoy your cheaper trades and increased liquidity, and go find a real problem to complain about instead of attacking those who make your life better.

> Is there some quantifiable financial thing that would regress to whatever terrible situation we were in back in 1990 before HFT was substantial?

Yes, trades would cost a lot more. Every time your money was put into a stock, some middle men would take more of it than they do now. Trades would take longer, you might not get the price you thought you were getting when said buy or sell because the price might change in the time it took some dude to go manually buy or sell the shares you requested. That directly affects everyone with a 401k or stock. HFT has eliminated a large swath of useless middle men who were gouging you for money: those middle men are pissed they've been obsoleted and lost access to easily profit and are the now pushing to regular HFT, so they can go back to the good old days of bigger profits and more room for middle men taking a bigger bit of the average mans investment money.

HFT saves every market participant money, except the old school traders whom they've largely obsoleted.

To say it's dead is a bit extreme, but it's not like 2008 when anybody kinda fast could find success.

Speed doesn't make money by itself anymore, so a lot of players whose only trick was being fast and not-so-smart are having problems.

https://www.bloomberg.com/news/articles/2017-07-13/they-re-t...

Its not dead. But like any other maturing industry there are now scale advantages. Firms will merge and market making can go back to being the boring activity it has always been.

Did it serve any higher purpose? Did any meaningful software/hardware advances come out of this? Has it evolved into something else now?

I worked at an HFT firm. We didn't cure cancer, but we tried hard to improve the technical ecosystems we were a part of. Many other firms do the same.

We sponsored the x64 port of LuaJIT and kicked off a sponsorship system for it [1]. OpenResty took nginx and integrated it with LuaJIT. Ten years later, CloudFlare started using OpenResty and LuaJIT to protect massive swathes of the Internet. This kind of butterfly effect makes me smile.

The same HFT crew discovered critical issues in the circa-2009 Linux kernel, wherein there was significant packet loss on multicast workloads [2]. We were using bleeding edge Debian/Ubuntu distros whereas much of the industry used more stable kernels (which didn't have the issue). So because of our hard work, along with much love from the incredible Eric Dumazet, we figured it out and everybody benefitted (especially RedHat and SUSE who got to put the fixed kernel in their stable releases a couple years later).

We also sponsored other open source projects, Debian packaging, etc. HFT firms from 2006-2010 were early adopters of the advanced network and computing technologies that now power the clouds (e.g. Arista, Solarflare, various acquired storage/network companies); those companies might not be around now if they didn't get those early wins from the finance community.

[1] https://luajit.org/sponsors.html [2] http://www.spinics.net/lists/netdev/msg90771.html

You worked at Athena?

The guys who were fined for manipulating the NASDAQ closing auction?

https://www.bloomberg.com/view/articles/2014-10-16/high-spee...

I did. Although I left years before that, Athena was not "fined". They settled without admitting any wrongdoing.

Sanders lacks fundamental understanding of the stock market, indicated by his unsubstantiated refrain "the business model of Wall Street is fraud." First, there are already per-trade regulatory fees, so this isn't a novel idea. Second, this is necessarily a regressive tax: wider spreads mean worse prices for the ultimate owner of stocks; that tax is applied whether trades are retail (you and me), institutional (big Wall Street firms, i.e. IEX's clients), or HFT (small tech firms).

>Second, this is necessarily a regressive tax

You're putting your own ideas into his mouth. Nobody said anything about the fee being a flat fee, or that it would apply to every single trade in existence.

I'm saying this as a Sanders supporter, but maybe he should have been more specific and detailed about his proposals then. My two biggest frustrations with Sanders were 1) his inability to understand that it was going to take more than just mere backbone and higher taxes to reform wall street and 2) his foreign policy

>but maybe he should have been more specific and detailed about his proposals then.

I'll agree with you there, his proposals were very similar to Trump's in that they did not have much substance, just broad ideas.

Your reasoning seems to ignore the evidence in this article; there does seem to be a business model of fraud and, even if he did lack a fundamental understanding of the stock market, I would hope we could all agree that action needs to be taken to regulate and restrict these illegal practices. Perhaps it would not be the specific terms Sanders was describing, but unfortunately he was the only viable candidate that ostensibly wanted to do so.

> Your reasoning seems to ignore the evidence in this article; there does seem to be a business model of fraud

What evidence? Specifically, where do you see any evidence?

The article is filled with emotional appeals, doesn't quantify its "calculations", conflates queue size with a lack of liquidity, professes obvious and heavy-handed hero worship for Brad Katsuyama and IEX, and (to top it all off), claims that high frequency trading is front running.

Frankly, I'm shocked it was even published, even as far as op eds go. The article perpetuates the same tired FUD about high frequency trading that IEX continues to push out, and in doing so preempts reasonable discussion about the real negative externalities caused by HFT. The author either doesn't understand, or willfully misconstrues the way in which HFT operates.

The main criticism of HFT I have heard is that it makes life harder for institutional investors, which makes them less likely to take effort to invest well.

This article is a pretty transparent advertisement for IEX.

I suspect this debate was/is largely conducted by uninformed parties. There already is a small fee imposed on every trade due to gov't regulation. https://www.sec.gov/fast-answers/answerssec31htm.html

At best, there could be a meaningful debate over increasing the appropriation to the SEC and adding new regulation mandates for them.

The question over whether we should have transaction fees should already be settled by the fact that the SEC is currently funded by such fees.

IIRC, the bill he proposed had taxes on the value of trades, such as 1% per stock per trade, .05% (maybe off by OOM) on options, etc.

Someone will call this a submarine article for IEX but I don't think this is far, it was written as an oped by Dave Swensen, the famous head of Yale's endowment fund.

Just so the issue is clear, almost all hedge funds don't do active/passive also called maker/taker, but rather they pay a flat fee per share traded to their sell side broker.

The sell side broker will then collect/pay the exchange fees. This means that the sell side broker has an incentive to post the order to a market that pays them the largest rebate rather than the market with the shorted queue.

The buy side clients are not getting worse prices necessarily as you still need to fill orders at the NBBO.

So the argument would be, why not post on the exchange with the shortest queue always. And the response would be that

1) markets move fast, and what is the shorted queue when the order is dispatched may not be the shortest queue when the order arrives.

2) Other markets may be more active, ie more orders routed to them first so the shortest queue may not be the best place to route at all.

To be fair its not a consensus that the IEX approach is better than maker taker, there is no clear consensus as to what the correct approach is even when you take out the HFT opinions.

In case anyone wants to see here's a link to the BATS cash equities execution quality page.

https://www.bats.com/us/equities/market_statistics/execution...

Worth noting that the head of Yale's endowment fund and IEX's interests are aligned. The business model of IEX is to convince naive investors that high frequency trading is somehow screwing them and that they should choose to route their orders to IEX in order to avoid it. They then facilitate trading between these naive investors and large institutional investors (like Yale) who want to move large blocks of shares before the market moves against them.

Parent is correct. Uninformed order flow is what HFT wants to trade against: people buying because they have cash and selling because they need it. HFT doesn't want to trade with people moving prices at the old prices.

Most brokers charge some all-in commission per share to clients. If you take away the rebates, brokers will simply raise commissions to compensate. Whether a fee is passed through or not is irrelevant. Brokers don't pass through direct costs of exchange licenses, servers, development time on their algos, etc. but clients still end up paying for them.

Odds are clients like the current system because they know exactly how much executing a 100k share block will cost in commissions. If they didn't, there are plenty of competitive equities brokers who will offer pass-through pricing.

And saying the shortest queue is best makes no sense. Most exchanges with short queues have laughably low market share. The lines are short for a reason, because they get serviced more slowly. Otherwise it'd be free money for fast HFTs to fill those lines up and get trades before someone on another exchange.

My post and your post seem to agree on all the points you just made:)

The real reason the article is unserious is because IEX is basically purpose built to game the execution quality metric they're looking at (effective spread... trade price vs NBBO midpoint). Imagine you built a dark pool that only executes at the midpoint of the NBBO. It would have a perfect "effective spread" score but contribute nothing to the displayed NBBO. That's essentially what IEX is. Their midpoint order type only has value because of the displayed liquidity at the other rebate-paying exchanges.

The problem with this essay is that it conflates the needs of a large institutional investors like Yale with the needs of small investors (like you and me). In many cases these needs can be inverted.

Yale wants to buy and sell large blocks of stock without the price moving away from them.

I want to buy and sell stock at the best possible price with all of the latest information transmitted to the market as fast as possible.

This is why Yale would prefer to trade in "darker" exchanges like IEX and most retail investors should prefer other exchanges.

Yes. This isn't Wall Street is bilking the common man. This is one side of Wall Street trying to squeeze the other side, and appealing to the public to help their case. It's a tough market for big money right now. Opportunities are scarce and their customers are pushing down costs, so they're putting pressure on their suppliers.

> small investors (like you and me)

Uhmm, maybe you, but definitely not me. Most people I know (I agree though, I live outside NYC or any big financial center) don't give a rat's ass about where the market is going, which stocks are good and which are bad. We invest a lot into our 401(k)'s and the rest we put in index funds. So my interests are definitely aligned more with large institutional investors.

BTW my personal investment strategy is not to get some kind of windfall during retirement. The only thing I really look for is that if I saved well, I will have enough to retire on (i.e. not ask for others for monetary help). I just want my funds to not fail.

That makes you like me. You aren't buying and selling based on proprietary information about where you think the market is going, but because of your cash flow situation (You save $1000 a month now, and hope to withdraw $5000 a month in retirement or whatever).

I realize this in the op-ed section of the paper, but where the issues are complex and most people are totally unfamiliar with them, printing one side of the story feels a lot like endorsement.

When Flash Boys came out there was extensive discussion here on HN about IEX's claims. I was convinced that Lewis at the very least exaggerated the benefits of thier speedbump model.

On the issue of rebates, I'd keep on eye out for other takes (especially from Matt Levine at Bloomberg) before forming any firm opinions.

More generally, consumers have been fleeced and left out to dry over the past 10 year economic cycle. Take a look at a graph of the S&P500 from 2009-2017, then look at consumer interest rates of 0.5% (in a 'high yield' savings or CD account). It simply makes no sense. We simply cannot build wealth any longer. Baby boomers had CD rates of >10% [0]. Think about that for a moment.

[0] http://www.forecast-chart.com/rate-cd-interest.html

> Baby boomers had CD rates of >10% [0]. Think about that for a moment.

My aunt and uncle (who happen to be baby boomers) bought their first house in the early '80s when inflation was rampant. Their first mortgage had something like an 18% rate. It should be obvious that a 10% CD rate is worthless if inflation is nearly as high or higher.

> It should be obvious that a 10% CD rate is worthless if inflation is nearly as high or higher.

Well, yes, that would be useless, but that's not what happened for most of the time with high CD ratee. The period of CD rates ranging from just under 10% to over 17% in the 1978-1984, saw inflation peak at 14.8% and spend much of the time below 5%.

Recently CD interest rates are not only low but for many years below inflation most of the time; in the period of high interest rates and high inflation, CD rates were still above inflation.

This is one of the most disturbing charts I've seen recently. Especially with the 45yr history.

> Although the harm suffered on each trade is minuscule — fractions of a cent per share — the aggregate kickbacks amount to billions of dollars a year.

That's the money quote. 99% of retail investors should be buying stock infrequently, maybe once a month when the paycheck comes in. Ideally you're buying one or a few index funds, so the total number of transactions is small. If you're in that boat, this order-of-fulfillment tax really doesn't affect you and can be entirely ignored.

Your returns are really only in danger of being dragged down by this thing if you're executing many trades per day. But if you're doing that, you had better be a sophisticated investor anyway, or else you're definitely losing money.

This is true, but incomplete. Such unsophisticated investor has to control his frequency of trading, and also the frequency of trading of his etf or mutual fund. Any fund needs to do trading, re balancing, etc. So there's a double layer where ppl get charged 'the vig'.

Whenever I see these IEX articles, I think it is useful to point out that they generally describe 3 types of traders: 1) Retail - you and me, who make money through asset appreciation 2) Institutional (IEX's client-base) - big Wall Street firms, big money managers, big pensions, who make money by charging fees 3) HFT - small technology firms, who make money through trading

Reading further: this really feels like PR for IEX.

I see no social benefit that accrues from allowing trading (vs. investing) in the stock market. The usual arguments that traders make in its favor are just a plain attempt at legitimizing their stealing from market participants who are less sophisticated and resourceful.

1) Efficient price discovery - Individual stock prices often move 20,30,50%+ in short periods of time, prices can't be efficient with such high volatility. Trading more likely causes inefficient pricing due to speculations, margin calls, trigger orders.

2) Liquidity - Sure they do increase liquidity in the market but as long as sellers can find buyers I don't see any benefit from increased liquidity other than (3)

3) Bid/ask spread - I'll admit that lower spreads are desirable and traders to play a vital role in keeping it so, but I'll question how important it is for a long term investor if has to cover a spread of an extra 0.1% and then weigh that against the social cost of tens of billions of dollars worth of wealth transfers from retail traders to some HFT shops in New York.

This is a quasi-inevitable side issue. Generally, investors eat transaction costs and aren’t party to decisions that affect them. There are related issues in many 3-party systems. In medicine, doctors, insurers and other people tend to decide on (prescribe) products and services. The consumer eats the costs (sometimes it’s a 4th party) and decision makers have an incentive to take kickbacks.

The regulator could do stricter rules/policing, for a partial fix.

Every article I read on the topic has a plug for IEX, an exchange that "refuses to pay kickbacks" which make me think that these peices are basically advertisements. Either kick backs serve no purpose and IEX will thrive especially in this environment where where investors are clamoring for yield. Or it does have some purpose or is not significant. Putting in regulations would likely lead to more complexity and regulatory capture

Kickbacks to brokers that are sending out retail flow allow those brokers to reduce commissions on trading. These reductions have gone all the way to zero in a lot of places.

> Kickbacks to brokers that are sending out retail flow allow those brokers to reduce commissions on trading. These reductions have gone all the way to zero in a lot of places.

I think you are wrong here. Most consumer orders are active meaning typically the broker would pay the maker fee instead of collecting a kick back. I mean, inverted exchanges are a thing but do very little volume.

Now pay for order flwo from wholesalers like Citadel is a big win for brokerages and does help to reduce trading fees.

Some of the underlying data here [1] on the effective spreads of the various market centers. Bats also publishes the definitions of these metrics [2]. They seem to be difficult to game, since they're derived from actually executed trades compared to the midpoint NBBO (that is, the national best bid/offer -- by definition, at that time, cheaper than any exchange to buy, and more expensive than any exchange to sell).

The low volume numbers make it slightly more difficult to determine significance, but the effect does appear to be real, which I found slightly surprising.

[1] https://www.bats.com/us/equities/market_statistics/execution...

[2] https://www.bats.com/us/equities/market_statistics/execution...

So lets say they extend Reg NMS to require sending the order to the NMS exchange with the shortest queue at the BBO?

I can imagine some potential ways that could be gamed as well...

Articles like this are so mind-numbingly frustrating because they completely muddle the water for any reasoned debate about the true advantages and disadvantages of high frequency trading.

> Wall Street has developed a new way, clouded in obscurity, to fleece the hundreds of millions of Americans who have money invested in company pension plans, mutual funds and insurance policies.

Well, that sure is a neutral way of presenting it, isn't it?

> Instead, brokers routinely take kickbacks, euphemistically referred to as “rebates,” for routing orders to a particular exchange. As a result, the brokers produce worse outcomes for their institutional investor clients — and therefore, for individual pension beneficiaries, mutual fund investors and insurance policy holders — and ill-gotten gains for the brokers.

"Kickbacks"...that's a strategic word to use. Technically true, but more importantly, emotionally loaded. "Kickback" is not often associated with positive sentiment. "Union leaders receiving kickbacks"..."politicians receiving kickbacks"...

More importantly, this claim is neither axiomatic nor defended by the article. How precisely do these rebates harm investors?

> The diffuse harm to individuals and the concentrated benefit to Wall Street create yet another way in which the system is rigged, justifiably eroding public confidence in the fairness of the financial system.

What the hell? The rhetoric is so heavy-handed - is there no attempt at an unbiased presentation here? I understand this is an opinion piece but come on.

> And yet, brokers choose longer queues hundreds of thousands, if not millions, of times a day. Publicly available trade and quote data show that the queues to buy or sell stock are considerably longer on exchanges that offer kickbacks. Even though the queues decrease the likelihood of getting a trade completed and impair the price performance after the trade is executed, brokers still direct trades to these places because of the kickbacks they receive.

Yes, that's interesting. But how does that correlate with the liquidity available on these exchanges? If you have reduced liquidity, do you want to be in a smaller queue with less price competition? This isn't even addressed.

> One exchange, the IEX, refuses to pay rebates. Created by Brad Katsuyama (whose odyssey to defy the ethos of Wall Street was told in Michael Lewis’s “Flash Boys”), IEX has a speed bump that prevents high-frequency traders from front-running ordinary investors. (Yale University, where we work, has a de minimis exposure to IEX through an investment by one of the university’s external managers.)

Okay, so we have blatant hero worship and the claim that high frequency trading is front running in 2017. And this article has reached the front page of Hacker News.

> BATS (a rival stock exchange founded by a high-frequency trader) posts data on this measure of execution quality for the major exchanges on its website. According to our calculations, in the six months before IEX’s arrival, Nasdaq led the effective spread rankings in the widely used Standard & Poor’s 500 index, with the number of top ranks ranging from 169 to 216 stocks.

Okay, where are these calculations? What is the point of making the claim if you don't quantify it whatsoever?!

>More importantly, this claim is neither axiomatic nor defended by the article. How precisely do these rebates harm investors?

The rebates are effectively being taken out of retail investors' money. If the rebates did not effect brokers' behavior, there would be no point in offering them. If the rebates do effect brokers' behavior, then it means brokers are willing to accept a marginally worse price for their investors in order to get the rebate.

The money you lose in the cost of rebates being passed on is less than the money you gain through enhanced price discovery (so you overpay for securities less often) and enhanced liquidity (so you can actually fulfill your trades more often).

There is no neutral comparison being run. You should be comparing against a market where every trade is happening on lit exchanges.

Can't practices like these not be considered in the same way as "hacking" and thus made illegal?

After all, hacking = influencing electronic systems to make them function in ways they are not intended to function. Replace "electronic" by "legal/financial" and there you are.

Sure, but the people standing to make millions of dollars on this particular illegal activity are also the people making the rules and laws.

One immediate problem -- this is how these systems are made to function. It's not like the kickback is happening behind the back of the operators of the exchanges. It's not "hacking" to follow the rules as they're laid out.

So you are saying these systems are bug-free by design/definition? That's quite a statement :)

That's so crazy far from what he is saying.

If you enter a building through a front door, you are not "hacking", you are entering the building in the way the designer intended.

If you enter a building through a window, you are "hacking" because you are exploiting an unintended ability that the designer did not intend to give you.

Just because the comment you replied to said that "kickbacks" are a front door intentional design, you can't then claim he said that the building has no windows. He has said nothing about windows. He just said don't call that door a window, because its not, its a fucking door.

Yes, it's a door, but they told the regulators it was a window. That's a bug in the spec.

> IEX has a speed bump that prevents high-frequency traders from front-running ordinary investors.

Anyone care to explain, in precise terms, how a high-frequency trader front-runs ordinary investors on a typical exchange and how putting a delay on all incoming orders prevents it?

They don't. That term has basically become an advertising slogan for IEX.

HFT is great for retail investors because you trade cheaper and faster. Usually retail investors get price improvement over the market since HFT brokerages compete for retail flow.

HFT is bad for institutional investors who don't want to invest in sophisticated execution since the market reacts very quickly to large orders. Institutional investors include firms like Vanguard or firms that greatly inform price discovery, so it's worth thinking about both sides of the market and not just optimizing for best retail execution.

IEX is by and for large institutional investors. IEX's delay doesn't apply to one (multiple?) of their hidden/protected order types, which allows larger orders resting on the book to avoid market impact and execution at 'bad' prices as these orders can move away from the top of the book in 'unfavorable' market conditions.

Michael Lewis and Brad Katsuyama misuse the term "front running," either willfully to stir people up or out out of ignorance. Front-running refers to the practice of a broker holding customer orders, but trading for their own accounts at a better price before executing their client's orders (i.e. using privileged information for their benefit). Katsuyama's uses "front running" to describe a practice where if an HFT sees a price change happening (i.e. trades are being reported or a level is going away), they quickly go and remove liquidity before others have gotten a chance to react to this new information. An HFT with faster technology will beat an institutional trader in this race (even though all market participants must always send bona fide orders that they intend to have filled). IEX solves this problem by slowing down incoming orders to give special orders on the exchange (D-pegs, which automatically change prices as the prevailing market price changes) an opportunity to reprice. The HFT has put millions of dollars into fast technology, whereas the institutional investment firms have presumably put millions of dollars into long term research. IEX allows institutional investors to outsource this technology investment to the exchange, which puts an artificial delay to give its matching engine time to reprice special orders during market moves.

Taking aside the speed bump that IEX provides, how it is meaningfully different from Liquidnet?

Don't get me started by putting out warnings of stocks they are shorting for no reason. It's despicable.

Not only companies, but also countries. While America will always be aaa

Reply on news.ycombinator.com