Matt Levine and Felix Salmon on Michael Lewis and HFT

In my alternative Michael Lewis story, the smart young whippersnappers build high-frequency trading firms that undercut big banks’ gut-instinct-driven market making with tighter spreads and cheaper trading costs. Big HFTs like Knight/Getco and Virtu trade vast volumes of stock while still taking in much less money than the traditional market makers: $688 million and $623 million in 2013 market-making revenue, respectively, for Knight and Virtu, versus $2.6 billion in equities revenue for Goldman Sachs and $4.8 billion forJ.P. Morgan. Even RBC made 594 million Canadian dollars trading equities last year. The high-frequency traders make money more consistently than the old-school traders, but they also make less of it.

There is more here.  Here is Felix Salmon on the book:

Similarly, Lewis goes to great lengths to elide the distinction between small investors and big investors. As a rule, small investors are helped by HFT: they get filled immediately, at NBBO. (NBBO is National Best Bid/Offer: basically, the very best price in the market.) It’s big investors who get hurt by HFT: because they need more stock than is immediately available, the algobots can try to front-run their trades. But Lewis plays the “all investors are small investors” card: if a hedge fund is running money on behalf of a pension fund, and the pension fund is looking after the money of middle-class individuals, then, mutatis mutandis, the hedge fund is basically just the little guy. Which is how David Einhorn ended up appearing on 60 Minutes playing the part of the put-upon small investor. Ha!

Lewis is also cavalier in his declaration that intermediation has never been as profitable as it is today, in the hands of HFT shops. He does say that the entire history of Wall Street is one of scandals, “linked together trunk to tail like circus elephants”, and nearly always involving front-running of some description. And he also mentions that while you used to be able to drive a truck through the bid-offer prices on stocks, pre-decimalization, nowadays prices are much, much tighter — with the result that trading is much, much less expensive than it used to be. Given all that, it stands to reason that even if the HFT shops are making good money, they’re still making less than the big broker-dealers used to make back in the day. But that’s not a calculation Lewis seems to have any interest in.


If an ordinary person traded on inside information they could go to jail.

Yeah, I didn't read the post either.

I read the post.

And this one too:" FBI Wants to Know If Robots Are Trading Illegally" the investigation alone means there must be some illegality going on.

That's exactly right. Since people don't understand what HFT is (and that the SEC now compels all designated market makers to be HFT shops - they all have to have algos now), then it must be something at least a little evil and any investigation, no matter how idiotic, just confirms that.

There is no "inside information" here of any sort. The order book is about as public as you can get. The computers just read it first and react.

Right, it is all totally transparent. The HFT see you place your order to buy shares and use their high cost comm network to buy the shares 3 milliseconds before your order reaches sellers so they can sell you shares for a few pennies a share more.

After all, if the HFT traders had failed to buy the shares you ordered in the 3 milliseconds it takes your order to reach the trading computer, no shares would exist in the market place for you to buy because it was the HFT buying the offered shares 3 milliseconds earlier that caused the seller to offer the shares, while no seller offering shares would fill your order.


I worked on projects where telecomm link speed was a big issue back in the 90s, and back then hedge funds were spending tens of millions to install fiber links to ensure high bandwidth, short round trip, and highly reliable computer transactions. So, I completely believe a fiber link was run using horizontal drilling to cut 5 milliseconds out of the comm time between Chicago and NJ trading centers. $400 million was only risky if you were not going to hold the hedge funds hostage and legally extort millions from them.

Regardless of how inane and misunderstood this example, it's clear that people think this type of inter-exchange latency arb is the raison d'être of HFT. If you believe that's so, how can you explain that HFT is just as prominent in futures markets (60% of the volume, see link below) where all trading occurs on a single order book?

There's no one raison d’être for HFT. Other than making money, of course. And I'm not saying this in any pejorative sense. It's misleading to imagine HFT as a monolithic endeavor either good or bad. HFT uses a bunch of strategies, and so obviously some part of that toolkit works even in a single order book universe.

OTOH there seem certain HFT strategies which sound outright obnoxious / unfair or at least dangerously destabilizing or questionable in some incarnations e.g. quote stuffing, stub quotes, flash orders, momentum ignition, etc.

Sure, some of these are banned now, but we know they did exist in the recent past and weren't exactly voluntarily forsaken by HFT players. What other corpses might be in the closet? I don't know. More worrying, with the amount of secrecy surrounding HFT strategies I'm not even sure how we would even know. The strategies we openly discuss & dissect must have been cutting edge two HFT generations ago?

A constructive discussion might focus on identifying & rooting out such specific practices from HFT rather than vilifying the entire enterprise? Where exactly do we need the next regulatory nudge?

Fair enough. But I'd say the 'evil strategies' get far more mindshare in the public than they occupy among actual HFT firms. The idea of some evil hackers manipulating the market with vast capital pools and stealing people's orders sounds a lot more tantalizing than the common reality of a bunch of nerds sitting around their HPC clusters trying to slightly titrate some alpha parameters to increase PnL by a few percent.

The reality is most of the evil examples that you mentioned affect normal market making HFT the most of anyone. The problem is any time a respected HFT firm, e.g. GETCO who release comment letters all the time, brings up a problem they are pretty much ignored. Since they "make money" everyone assumes they have nothing useful to say. Than ZeroHedge or Nanex completely mutilate an issue with the understanding of a five year old, and they get wall to wall press coverage. If it wasn't assumed that all HFT is evil, many of these problems could be brought to attention earlier.

To its credit the SEC does do a much better job than the public press of actually listening to people who work in the industry. Of course then the SEC then gets vilified as being in the industry's pocket for actually listening to experts and not cranks on the Internet.

You completely misunderstand what's going on. That would be perfectly alright if outraged yet uninformed people like yourself didn't have the potential to drive idiotic regulation.

And yet there are some very bright readers of this blog who remain unconvinced of all the benefits you're bestowing upon the world. Mainly because nobody can adequately explain it.

Strikes me as a feature, not a bug. I will say this: HFT Trader is doing his best to spin multiple plates.

Hey look: it's great for small traders!
And wow: not just small traders, but big traders!
And liquidity, and historically low bid-ask spreads, and ponies for everyone!

I mean, come clean and explain what on earth it is you're actually doing - unless you can't, in which case I'll continue to believe you're collectively up to no good.

"I mean, come clean and explain what on earth it is you’re actually doing – unless you can’t, in which case I’ll continue to believe you’re collectively up to no good."

I datamine historical market data based on various market metrics. Those are used used to build models which come up with realtime predictive returns and probabilities of fills for various permutations of orders (limit buy joining the bid, limit buy inside the bid, market buy, limit buy outside the bid, mid-price pegged, hidden order, etc.). Those metrics get fed into a module that tries to maximizes expected PnL given capital and risk limitations.

When that's done the system is backtested using various assumptions and methodologies on historical data. If that looks good those models get uploaded to production servers. Those production servers run programs that receive real-time updates from the market and feed it into the models, which generates the orders. If all goes right and you modeled correctly you earn a statical edge on your trading activity above your transaction costs and make money.

This is it, there's no secret hacking, market manipulation or order stealing. Does that adequately explain it?

I see the "if it's not illegal, why don't you just tell us what it is" argument quite often. The intense industry secrecy does not exist because HFT closes ranks to protect illicit activity. The secrecy exists to protect intellectual property in a highly competitive, zero-sum game. Most basic trades can be explained in two or three sentences. The details needed to make them work, however, require a strong mathematical and programming background and would mean nothing to the vast majority of humans. tldr; no trader can go into specifics about what he does, and even if he could, there would be no point--you just wouldn't get it. If you're really interested, there's a large literature on market microstructure that goes into trading considerations in more detail. Start with "Trading and Exchanges" by Larry Harris.

I understand what a model is, that it is necessarily complex and involves a great deal of IP that you'd want to protect.

You said in one post or another that HFT is a way of weeding out "less informed" investors so as to maximize efficient allocation of capital. Perhaps I am extrapolating but I take "less informed" to mean "less able to follow corporate events." Or were you inferring something else entirely?

Assuming that is what you mean when you say "less informed," what I'm having trouble digesting is that any kind of infrastructure is capable of not only digesting but running through the model a news event in real time such that nanoseconds later you're poised to take a position. I'm really having a hard time believing that. I'm also having a hard time with the notion that there are that many news events occurring every hour, every minute, every second, every microsecond that would trigger actions in your model.

Or are you more focused on patterns in upticks and downticks? In which case, is that really "more informed"?

Good question. "Informed" is a very broad term meant to indicate a market participant with a good knowledge of what the price of an assert "should be." The means by which they come by this knowledge are left unspecified. There is also the question of the horizon over which the knowledge is relevant. The traders of SAC were highly informed and could predict prices because of illegally obtained information. This information was actionable for several days, up until it became public and the market was able to incorporate it into prices. So, SAC's "informedness" came from their grasp of corporate events. The information that matters to HFT is very different than that which a hedge fund might be interested in. It consists of a very intricate understanding of supply and demand over brief periods of time. Our information arrives much more frequently, is considerably more diffuse, and is hidden in a huge amount of noise. You are right to wonder whether this really makes HFT "more informed." The inputs are each rather trivial on their own. However the sum can be quite revealing. HFT can infer the emergence of news (and thus correct prices) through the order flow it sees (upticks, downticks, orders in the book, other securities, etc). Is someone trading because they *know* (i.e. an SAC or someone similar) the price will shortly move up? Or is her or she trading for some less urgent reason, in which case the price shouldn't go anywhere? This is the informed vs. uninformed dichotomy and lies at the heart of market-making.

Regarding speed, a news story takes several microseconds to process. However, a new order in the market can be processed in several hundred nanoseconds. That this can be done is fairly remarkable but requires nothing more than a good knowledge of hardware architecture, operating systems, and software design.

I think HFT Trader's description of his activities is much less innocuous than he seems to think it is. It excludes very few possible strategies and includes strategies that many would immoral and some that are illegal. For instance, it doesn't exclude various forms of market manipulation including spoofing. It also doesn't exclude using ISOs to queue jump by deliberately locking the book on the slowest lit market.

Just wanted to thank you both - HFT Trader and Anothe HFT Trader - for the informative and civil dialogue.

Not that one opinion matters but I have a better understanding now what it is that you're doing.

>You said in one post or another that HFT is a way of weeding out “less informed” >investors so as to maximize efficient allocation of capital.

You're mixing up two different things. The HFT is not trying to weed out "less informed" traders, he's trying to trade with them. HFTs are happy to to sell share of MSFT to people who are buying because they're investing for retirement. They're happy to buy shares of MSFT from the guy who is selling his shares to make a down payment on his house. They do these trades and they make a penny or something per share round trip. This is good for the "uninformed" guy who is buying or selling because he can complete the transaction quickly and cheaply. It is good for the HFT because he makes his half a cent a share.

What the HFT doesn't like at all is trading with Steve Cohen at SAC Capital who maybe just learned something that no one else knows and that will have a big effect on MSFT's price. They don't like it because when they for example sell a share to SAC Capital or some other hedge fund or generally someone who is more informed about short term events than the HFT, the price is likely to go up after the trade, and the HFT will lose money on the trade.

The HFT would like to know whether he is trading with the guy who needs to sell his shares to make a down payment on his house (in which case the spread will be very tight) or whether he is trading with Steve Cohen (in which case the spread will be much wider). But there is no way of knowing this. All orders and trades are anonymous. You have no idea who you are trading with. On the other hand, there is so much data available that it's possible to detect very small, very subtle patterns in the data. To take a trivial example, equity returns on the scale of say a minute are slightly negatively auto-correlated. You can use very small effects like this to make a predictive model, given the current order book as well as historical orders and trades, of what is going to happen in the future. From these predictions you can develop a strategy. If you are very good and very careful, you can make some money doing this. Sometimes the strategy will tell you to change your prices or sizes or widen the spread, presumably because the model thinks it detects a Steve Cohen (I am anthropomorphizing this for clarity - the model of course knows nothing about informed or uninformed). But most of the time, markets are very tight - much much tighter than they were in the past when market making was a monopoly run by specialists.

Is all this activity nefarious? I'm at a loss to see how. At the end of the day, a bunch of different HFTs are competing with each other to buy your shares of MSFT at 35.19 if you want to sell or to sell you shares of MSFT at 35.20 if you decide you want to buy. It's your choice to trade one way or the other and it's your choice to trade at all. Or perhaps the HFT decides that he wants to buy the share that you are offering to sell at 35.20. Again, how is that wrong?

Perhaps the objection is that the HFT has spent a lot of time and money and effort assembling the best data sets, and developing the best models and strategies and the fastest data connections that he or she can, and that the retail investor can't hope to compete with that. But that is to misunderstand roles. You aren't competing with the HFT when you buy or sell a stock. The HFT is also buying and selling, just like you, but for entirely different reasons. You are buying because you think that MSFT will be a good investment over the next ten years or whatever. You are (or should be) thrilled to be able to buy a thousand shares of MSFT with one transaction at a cost of half a cent a share (20 years ago, 12.5 cents a share, 25 times higher). The HFT on the other hand is buying in the hopes of selling at a penny higher to someone else. The HFT is trying to maximize his own profit of course, but because of the intense competition between HFT firms, the only way he can make a profit is by offering you the best tightest markets on which to trade. Now of course, he needs to be careful of people like Steve Cohen. He can't afford to make markets a penny wide and put up huge sizes because he will be picked off - subject to adverse selection. So he puts up modest sizes, but certainly large enough for any retail investor to trade his or her fill on, and also enough for any institution to trade its fill, provided that the institution does not demand to be filled in very large size in a very short amount of time (which can be a tip off for adverse selection).

>what I’m having trouble digesting is that any kind of infrastructure is capable of not only >digesting but running through the model a news event in real time such that nanoseconds >later you’re poised to take a position. I’m really having a hard time believing that. I’m also >having a hard time with the notion that there are that many news events occurring every >hour, every minute, every second, every microsecond that would trigger actions in your >model.

News arrives relatively infrequently. Most predictive inputs are based on what I suppose could be called technical data. Changes in price and things of that nature.

And yet there are some very bright readers of this blog who remain unconvinced of all the benefits you’re bestowing upon the world. Mainly because nobody can adequately explain it.

It's OK for you to be unconvinced. But if we start trying to regulate it then the burden of proof needs to go the other way. In a free country things should be allowed unless there is a reason to ban them, not the other way around.

"But most of the time, markets are very tight – much much tighter than they were in the past when market making was a monopoly run by specialists."

"You are (or should be) thrilled to be able to buy a thousand shares of MSFT with one transaction at a cost of half a cent a share (20 years ago, 12.5 cents a share, 25 times higher)."

20 years ago a lot of things were different. Most of those differences aren't a result of a HFT shop getting a slightly faster internet connection. Can you really show that if your HFT firm shut down that spreads would be worse for Joe Schmoe? Especially in light of the fact that your goal is to make as much money off anyone (including Joe) as possible. I think this is a hard case to prove, especially for the marginal HFT shop.

"It’s OK for you to be unconvinced. But if we start trying to regulate it then the burden of proof needs to go the other way. In a free country things should be allowed unless there is a reason to ban them, not the other way around."

Not really. Finance, most especially finance specifically divorced from facilitating real world productive investment, tends to be a zero (or negative with transaction cost) sum game. We aren't talking about some consumer product or a factory getting built. No wealth is created by HFT. The only claim that HFT can make is that maybe somehow they make less of a spread then some world where they don't exist (even then, no new wealth, your just claiming that the distribution leaves the little guy with a bit more of the pie).

HFT, and things like it, take in a lot of very rare and scarce technical and intellectual talent. The best you can claim is that they are putting all that talent towards a zero sum game that maybe, through some really hazy assumptions that can be questioned, results in that zero sum pie possibly getting a little more into the common guys 401k, and only because the HFT trader failed at doing his job the best (taking everything possible would be best for his P&L). Considering that economic growth is mostly a function of such people coming up with real world technological breakthroughs that result in productivity growth (whose slowdown we have seen in very hard data for a long time) is HFT really something that should receive the benefit of the doubt.

Can we show that if an HFT shop shut down spreads would be wider? Uh...yeah. First of all, there's elementary economics: fewer competitors means less competition to trade with you and wider spreads. If that doesn't convince you, recall the "flash crash" of 2010 when several HFT shops turned off their algos. My stub quotes where hit out (I bought stock at $0.01) because there were no bids.

You fail to understand that benefits of traders, including HFT, because you don't understand liquidity. Liquid markets (many competitors) allow investors to exit positions without worrying about moving the price against them in the process. Liquidity reduces volatility (risk), transactions costs and, ultimately, the cost of capital for firms. If you don't think that this is valuable, then you'll have to explain why you think higher volatility, huge transactions costs and high cost of capital is valuable.

Anon - Those sentiments could apply to tax policy, redistribution policy, environmental policy, natural resource extraction restrictions (or in the USA bans), education, etc., etc. Welcome to what passes these days for a republic. At least Barney Frank and Chris Dodd are not controlling levers of federal power in this area any longer, but they did inflict much damage during their times and one ought not expect better management in the future.

I do trust correctly that the sentiments are not condoning the state of affairs at the federal level in regards to banking and finance, and that you are not a fan of financial repression and zombie cartels?

@Joe Teicher makes a crucial point, I think. From a very lassiz faire point of view, almost no HFT strategy is fraudulent. After all, what you are doing is "merely" analyzing info & making a series of buy / sell decisions. Innocuous enough.

But that's very different from actual SEC rules, I think. e.g. If, say, @HFT Trader's models told him to churn by issuing thousands of buy & sell orders simultaneously at identical prices that might run afoul of actual rules?

My point is, the devil is in the details & you cannot extrapolate from everyday commonsense rules of what's legal behavior to something complex like HFT. e.g. multiple ads on Craigslist offering to buy / sell the same bike model at same price simultaneously, isn't illegal.

Alternatively, what @HFT Trader described about his work sounds innocuous enough, but without deep examination of the actual strategy it might be impossible to say if it is clean or not.

"You are (or should be) thrilled to be able to buy a thousand shares of MSFT with one transaction at a cost of half a cent a share (20 years ago, 12.5 cents a share, 25 times higher)."

See, we were all playing nice until you had to go and conflate HFT with decimalization, which happened in 2001. Unless you can show me some proof that this was HFT-led effort, you don't get to take credit for that.

You also don't get to take credit for the lowering of broker's fees that were largely the result of the burgeoning of the internet in general - again, something that happened in the early 2000s and had nothing to do with HFT.

You might be able to take credit for increased tightening of the bid-ask spread. But again, if you're going to compare trading back in 2000 with trading now, the biggest gains to traders (and losses to market makers) occurred when markets moved to decimalization.

When you guys lay out these spurious claims my ears start tingling.

"See, we were all playing nice until you had to go and conflate HFT with decimalization, which happened in 2001. Unless you can show me some proof that this was HFT-led effort, you don’t get to take credit for that."

The average spread on S&P 500 stocks at the start of 2001 was around 35 bps. In 2002 it was around 20 bps. In 2003 it was around 15bps. Today it is around 3 bps. Decimalization occurred in April of 2001. How can decimalization account for the drop in spread from 20 to 3 bps that occurred between the start of 2002 and today?

I didn't suggest that brokerage fees were lower because of HFT.

No comments on any of the other points I made?

Source for bid/ask spread info:

I just watched a CNBC interview with Michael Lewis, Brad Katsuyama (the guy who is building the new exchange designed to beat them), and the CEO of BATS (a big electronic exchange that works with a lot of HFTs). The studio is right on the NYSE floor as it usually is.

After interview Katsuyama gets hero's reception walking by the NYSE specialists. Bootleggers and Baptists.

I enjoyed reading your posts and those of Cliff Asness and especially HF Trader in the previous post on this topic.

One thing that hasn't been discussed much is the irony in Katsuyama running a dark pool while simultaneously claiming he's the good guy. Dark pools allow brokerage firms and banks to internalize their order flow - i.e. avoid competition with HF Trading firms to fill the order. In doing this they weaken the public, displayed markets because they do not display order book information. Dark pools have a parasitic relation with exchanges and ECNs because they use quote and trade information from the exchanges, but they do do not contribute any of the information (order book and trade information) that is vital to efficient price discovery. If the fraction of trading in dark pools grows too large, it seems very possible to me that markets will become less efficient, to the detriment of investors. I don't know for sure (I can only speak with certainty about one case), but I strongly suspect that most HFTrading firms are doing the bulk of their trading in the public markets (exchanges and ECNs), where they ambiguously improve outcomes for passive uninformed investors (as outlined by HF Trader) rather than in the broker and bank owned dark pools, which I suspect do the opposite.

In my opinion both HFTs & the traditionals (brokerages, banks etc.) have facets & practices that are crooked / sleazy / anti-competitive, unfair etc. Clearly, neither party is a saint in this debate.

OTOH, that shouldn't make us blind to criticism just because an interested counterpart with its own ulterior motives brought it up. There's crap to be cleaned from both sides of the pigsty.

Well said.

Rahul, glad I had a chance to continue the HFT discussion with you (and hopefully HFT). Can we also step back and talk about price discovery? A lot of people in the last thread (you, HFT) seemed to be talking about price discovery as the main, or one of the main, benefits of HFT, which rings false to me.

Haven't we all spent hours and hours talking about the irrationality of markets? What should we care if we discover the wrong price, quicker? I'm being a little facetious, obviously, but it comes back to the old thing about precision and accuracy.

We haven't gotten more accurate thanks to HFT's; we've gotten more precise. That isn't price discovery, as I think of it.

If markets were so highly irrational it would be a lot easier to beat them over time. Financial economics makes a big deal out of times it finds market inefficiencies because doing so is difficult and rare. Sure there are inefficiencies and irrationalities, but market based price discovery is highly important. If it wasn't we could just centrally allocate capital, call it a day and go home. To that end reducing the noise in the price discovery process is a very worthy endeavor.

I wrote a comment but it's kind of non-sensical. So I'm putting the summary up top, and the comment below. Maybe together they make sense.

(a). I do not think markets are efficient/rational at all.

(b). Even if they are efficient/rational, there is huge disagreement about correct prices even in the absence of new information (there are many ways to prove this, including volatility and correlations between stocks, disparity in analyst recommendations, etc.). If that is the case, the precision that HFT brings to any particular millisecond of trading is moot in the face of large inter-temporal lack of accuracy. If that is the case, how does HFT support price discovery (i.e. discovering the fair value of a security)?

___original below____

It's not inefficiency that's rare. Inefficiency is rampant(1), but obvious, indisputable inefficiency that is newsworthy is rare. That is what gets reported. Anyway, I'm not sure I buy your argument. If price discovery were occurring, you wouldn't see the volume of trading that we see (2). A discovered price would be found and reached, until new information was found (or minor supply/demand of investors jiggled prices). There are probably a million ways to prove that stock prices aren't discovered (their volatility, inter-stock correlation), per se. They're agreed upon within a band, but not discovered.

(1) Depending on what level of the EMH we want to believe in, which I don't really want to discuss because all of the hypotheses sometimes true and sometimes false, sometimes.

(2) I would argue that if markets were rational/efficient, it would hardly ever be rational to trade (due to trading costs always being positive), except for demand/supply reasons or the discovery of new information, a relatively rare event compared to the frequency of trading. Hedge funds wouldn't really exist. Volumes would be way down.

brokers have always internalized order flow. Long before HFT and dark pools. They have always preferred to cross trades internally if they can rather than sending them to the floor.

Also, large trades have always been worked off floor and then just showed up on the tape. This isn't a new feature that suddenly arrived with the advent of dark pools and dark pools themselves are largely a response to regulation, btw. Large orders have the potential to move price against the customer, so customers with large orders always had brokers working their orders off the floor.

You're right that dark pool participants often trade on the market posted on the exchange and it's frustrating to do all the work and have everyone else trade on your market all day without getting a piece of the action yourself. All I can say is that we don't just sit and complain. We figure out ways to trade more and the exchanges themselves have innovated in response to this parasitic behaviour. People need to remember that we live in a dynamic world and the reason financial markets work remarkably well despite moronic regulation and pressure for increasingly moronic regulation is because we adapt. Those of us who can't simply cease to exist. So, I disagree that dark pools are as big a problem as you imagine.

@methinks: I've enjoyed your comments on this thread. I'm not sure about the dark pools. I agree internalization has been happening forever. It is quite profitable for banks and brokers to internalize their flow because it allows them to eliminate competition. Although I agree the markets are dynamic and always changing, on very general grounds I find it difficult to avoid the conclusion that the larger the fraction of total market volume traded on in dark pools, the less efficient the overall market will be. There will be less public information, less transparency, and less competition, all of which seem undesirable from the point of view of market efficiency. This is obvious in limit where all the volume were to be conducted on the dark pools, and should certainly operate to some degree at volume fractions well away from that limit. How to quantify this, short of running real world experiments (eg the SEC mandating no dark pools on some large randomly selected group of stocks and then seeing what happens), I'm unsure.

I've noticed that the Chairmen of IEX is Jim Clark, the founder of Netscape who Lewis profiled in The New New. If Lewis' good friend is a large and early investor in the venture, that begs the question of what financial ties Lewis might have to venture.

Overall the anti-HFT hysteria has become the financial equivalent of the anti-vaccine crowd. A combination of paranoid people with a tenuous and overconfident grasp on the subject material, a blatant disregard for evidence and logic, and a hanger-on crowd of conmen willing to play into people' fear to sell them overpriced and worthless junk. It's made all the more morally wrong by the fact that they're scaring people into seriously hurting their (financial) health, by not vaccinating their kids or investing in equities.

So, other than attacking the character of the author, what did you do to provide information to the debate?

If you go to the comments Tyler's post from March 30, you'll see that HFT Trader provided huge amounts of information, in a series of very informative comments, all for free. Go read his comments, I expect you will learn a lot.

Ed. I did, and didn't learn a lot.

And, what was your defense for HFT's adhominem?

Talking your book again I see...

Yeah, I was pretty unconvinced by your arguments in the last thread - eloquent as they were - much as I am Tyler's "well, one set of frontrunners is making much less than another, so we should just thank our lucky stars" conclusion.

The costs of trading for the small investor are not a whole lot lower than they were back in the 2000s when shops like eTrade and Fidelity came online. The cost of trading for retail folks dropped almost overnight by 75% or more. Decimalization was another huge boon. HFT injects undue systemic risk into the equation and the benefits aren't obvious to a small player like me.

The liquidity arguments might (might) hold with more off-the-radar type stocks, but if your argument is that this is all for the benefit of the small investor, most of those people aren't buying stocks with small floats and/or low volumes.

Here's a graph I found of the declining trend in bid-ask spreads since 2002.

"The costs of trading for the small investor are not a whole lot lower than they were back in the 2000s when shops like eTrade and Fidelity came online."

Really? Are you going to bother citing anything for this? Or just throw it out there. First let's talk about average bid-ask spreads. In 2014 it averaged 7 bps for S&P 500 stocks, in 2013 it averaged 3 bps. (And certainly since we're talking about S&P 500 stocks these aren't names with small floats or volumes). A decade ago bid-ask transaction costs were 133% higher than today.,d.cWc
(Page 3)

Now let's talk commissions. And before you object, yet HFT lowers commissions cost. Lookup payment for order flow. Retail brokerages are heavily subsidized by HFT internalization pools. Correct me if I'm wrong but in 2000 I believe E-Trade's commission was $15, and it was the best game in town. Today TradeKing offers $5 commissions, and many other places offer free commissions on ETFs. Cumulative inflation has been 36% since 2000. That means on a real basis retail commissions were 300% higher than they are today.

So, retail sales could be cheaper if HFT were to buy and sell on eBay and Amazon tens times more than real consumers do? Bezos could setup a system that allows HFT to buy items in the warehouse and keep them in the warehouse so they can sell them back 5 milliseconds later?

Bullet-proof argument. Your perfectly crafted analogy just completely made irrelevant decades of research in financial economics and market microstructure. Shut it down, everyone go home. No need to do research anymore, HFT is evil (citation: mulp's Amazon analogy April 1, 2014).

Are you (or is anyone) able to prove that HFT is the root cause of bid-ask spreads being tighter?

Here's a question: how rife was HFT during the 2008 crash in the period between when Bear Stearns imploded and not long after, Lehman?

I don't remember hearing that as a possible culprit amongst the raft of proffered causes of the crash. I guess I don't believe that the decline in the bid-ask spread as shown graph is largely related to HFT. Especially in the early- to mid-2000s.

I think what mulp is trying to say is that many other industries have gotten along just fine at increasing efficiency and reducing cost, without pointless (I don't say 'pointless' to be disparaging, but I think we can agree HF Traders are speculators, not investors) mechanisms.

Trading is a zero-sum game. HFT is profitable, too. Knight and Getco are making $1.3bn. Call it, along with bulge bank HFT desks and the mom and pops like the one you are getting paid by, a $5bn (revenue) operation across US equity markets. This money must come (pretty much by definition) from less sophisticated investors, and is done by processes that are at worst harmful (order cancellation), and at best stupidly wasteful (setting up lasers to Chicago). Can you understand why we're not giving you a medal for our 4bps reduction in bid ask spread? And can HFT take credit for the reduction in retail trade costs, or is that a massive increase in retail investors spurring economies of scale for the big guys like Schwab, and inviting low cost competitors such as Tradeking?

Let's call a spade a spade. You guys are just another class of hedge funds. Like hedge funds, you make money in a zero-sum speculation game, and produce liquidity as a side effect. Like hedge funds, I think most of you are pretty nice guys, and very smart, but if the earth swallowed up your entire existence tomorrow, society would keep on chugging without a hitch.

Okay fine, let's say you completely reject the gains to capital allocation efficiency. You're committing a fallacy by comparing the "huge" $5 billion in revenue (which is overly generous, since banks engage in very little HFT, but let's go with it) to the "tiny" 4 bps bid-ask reduction (and again you're ignoring the commission gains). But approximately $20 trillion in equity value gets traded a year (and your HFT revenue numbers include, futures, while this does not). So even by your overly generous lower bounds HFT has saved end consumers $8 billion a year. Even pretend that there would be no overhead cost to the previous HFT market makers (there was, floor traders need to eat, and most common estimates was that they earned about $20 billion a year in aggregate). That's still a net consumer surplus of $3 billion a year, using by far the most unfavorable estimates we can.

This is generated by maybe 5000 total people working in the industry. Each HFT worker is generating $600,000 a year in consumer surplus. That makes it one of the most economically productive endeavors on the planet. Whether society would "keep chugging" or not is a stupid metric. In aggregate society would be a fair less well off, which is pretty much what you would say about 99.9% of any other type of worker.

Is that 5,000 high frequency traders or 5,000 people including the large army of IT drones laying cable, configuring servers, optimizing router code, etc.?

Just curious what number you're using.

Employees of HFT firms or HFT divisions of larger firms. Virtu has about 150, KCG has about 1200 of which maybe 600 do HFT, Citadel has maybe 150 working in its HFT divisions, RenTech has about 300 at most half do HFT, Jump has about 400. That's 1350 employees among the mega-shops, who collectively make up 30% or more of HFT activity. That gets you to about 5000 employees industry wide.

Hey don't get me wrong. I like you guys. I know how the traditional equity traders at the banks work, and I'd rather you have my money than them. So cool. All I'm saying, and what I think mulp was saying is that the relative benefit isn't great. Hence why I'm saying no one is really throwing you a parade.

Your calculations -- and they seem good enough for gov't work -- indicate $3bn consumer surplus on $20tr of trading. That's 1.5bps of surplus gained. Let's say I'm wrong and off by 5x, that's 7.5bps more efficient. It's nice, if that full 4bps can be granted entirely to HFT. mulp might say that 7.5cents on his $100.00 ($99.93!) Amazon purchase seems like a bad use of resources. I don't really care either way, but I sympathize with those people that are less than willing to sing your praises.

Keeping in mind that mulp probably trades a handful of times a year, just as he makes a handful of $100 amazon purchases a year.

Also, quick aside, am I stupid, or if there is $5bn extra in the market than there would normally be (supply and demand), doesn't that increase the equity markets' money supply? Does HFT cause price inflation? Been a while since I took macro.

Amazon does indeed use dynamic pricing based on algorithms.

This is quibbling, but Amazon's product is the book itself. A market maker's product is liquidity in the stock, not the stock itself. So a reduction of 4 bps relative to the price of the stock doesn't mean that the service being performed has only become slightly cheaper. The analogy would be if a different payment system comes along and cuts the transaction fee of credit cards from 1% of 0.25%. You would rightly say "That's Amazing!" not "Oh, big deal. I went to buy a shirt the other day and it was only fifty cents cheaper."

"It’s made all the more morally wrong by the fact that they’re scaring people into seriously hurting their (financial) health, by not vaccinating their kids or investing in equities."

Agreed. "Vaccines cause autism" is an apt analog for the kind of hysteria that this topic invites. If you believed the nonsense pedaled by people like Lewis, you would predict that people who work in high frequency trading firms never buy stocks in their own personal brokerage accounts because they would realize that "the market is rigged". Based on my own experience and that of others I know, the opposite is true. People who work in HFTrading firms have no hesitation about buying stocks or etfs in their personal accounts. They know they can do so at very low cost because spreads are so tight. Of course, most of them are also smart enough to know that investing is not trading so that they tend to hold these stocks for a long time in order to minimize the performance drag caused by taxes and transaction costs (as low as these are on a per trade basis, they still add up if you trade a lot).

So the HFT guys pool all the small investors creating secondary dark markets, and it's a good thing.

The HFT guys talk about liquidity but drive large investors out of the market. Makes perfect sense.

How can a contractual arrangement where someone pays fees for service be compared to an HFT who front runs a trade? Because the HFT are taking less money?

Isn't this whole thing a tempest in a teapot, something like a major crisis of confidence of the Salt Lake City exchange, or the Vancouver stock exchange? Isn't the stock market tiny compared to the debt asset markets?

I'm not sure "less dishonest than prior thieves" is a defense anyone would choose, unless innocent is unavailable. That said, the opportunity to steal (legally or otherwise) is why the market exists. If the buyer did not think he was getting the better end of the deal, he would not be a buyer. If you read the book The Money Game, written fifty years ago, you will see that there's not much new to what the sharps are doing today. Different tools, but the game is the same. The smart money profits from the dumb money.

There are some reasonable steps the exchanges could take to blunt the criticism. They have been out there for years, but ignored. That's nothing new either. reform comes after a crisis, never before.

I don't think "steal" means what you think it means.

Also, you seem to have made a leap from "profit" to "dishonesty".

I'm sure you have many unusual thoughts running around in your head. Perhaps that's why you prefer striking a ridiculous pose, rather than making declarative statements. It's a defense mechanism of sorts. Like wearing wool socks with sandals or cargo shorts in the winter. I'm just throwing it out there of course. I don't mean to imply you need your meds adjusted. I'm sure your dosage is just fine.

Z you got to raise your trolling game...this board is for pros not tyros. ;-)

Even Babe Ruth had slumps.

I question "the opportunity to why the market exists" and I am the one making a crazy statement? How about explaining how a system consisting entirely of voluntary transactions between willing parties is stealing?

The fact is, your comment had little to do with the argument made in the post, was illogical, and was not supported by any sort of evidence. And your response to criticism thereof is telling.

Maybe if you had just asked for an explanation, I would have provided one for you. There's nothing wrong with putting your hand up and asking for help. It's OK. We know.

By symmetry, if the seller did not think he was getting the better end of the deal, he would not be a seller. Where's the theft? I've seen a good deal of hyperbole and no actionable criticism from you in the past two or three days.

HFT is going to get vilified by people who don't work in finance and don't understand what it is or how trading works.

Michael Lewis should be ashamed for profiting on this public misunderstanding--and the NYTimes should be ashamed for promoting IEX Group and its investors, who are also trying to profit from the public's misunderstanding.

Funnily, a fair bit of vilification is actually coming from *within* finance.

How long do HFT hold on to shares?


How much can conditions in the real world change in the time the HFT hold shares to justify the claim HFT are adding value by providing "liquidity"?

Is holding a stock for one second really adding liquidity to the market?

Would you argue that Bezos was adding value by increasing liquidity if he allowed HFT to buy and sell items in his warehouses for small transaction costs without anything moving even in the warehouse?

HFT is primarily the business of providing liquidity (read, immediacy) to people who want to trade. The short holding period is a reflection of frequent incoming buyers and sellers, not the HFT "flipping" stock for whatever reasons people claim. Consider the case of one market maker always willing to buy at 100 and sell at 101. Alternating buy and sell orders come in every 100ms. The market maker's average holding period would only be 100ms, yet he is most definitely providing liquidity.

If HFT trading is driving other traders out of the market, (is it?) how is it providing liquidity?

how do you figure HFT is driving out traders when financial markets today are far more liquid with far tighter spreads? That only comes from more competition (read: more traders), not less.

I keep seeing this liquidity claim: is the (un)-stated goal of HFT to put market makers out of business?

HFT is driving (indeed, in the case of US equities, has largely already driven) a certain class of traders out of the market: the humans who used to perform its functions. For example, the trading floors of the past were populated by market makers who did exactly the same job at a much higher cost. Their only remaining purpose is to cradle their heads in their hands for CNBC cameras on down days in the market. Lest we have too much sympathy for this crowed, I note there is zero overlap between them and "retail" investors. These were professionals who collected huge monopoly rents. This process need not in any way hurt market depth or liquidity, and empirically, it hasn't. In plain terms, HFT provides no fewer trading opportunities to others than humans did. HFT's goal is not to put other market makers out of business (I say "other" because HFT *is* largely market making), though it is a natural consequence. The mechanism is very much akin to Amazon destroying traditional book stores.

I feel like improvements in technology, alterations in the exchanges, and changes in the financial markets among other factors all caused spreads to go down. Do we have any proof that HFT itself is responsible to some quantifiable level of spread reduction outside these other factors? Is such proof even really possible (and I mean good proof, not I'm an academic with some massaged data and generous assumptions - the kind of thing Tyler would post in general). And would that same logic apply to the marginal HFT firm (could we have gotten the same reduction if fewer HFT firms existed).

I mean certainly your not claiming that without HFT we'd still be at 70s spread levels.

My firm used to make markets in equities. We weren't driven out of market making by HFT. We were driven out by the SEC. The cost of regulation increased sharply while the exemptions (that enabled us to perform our obligations) disappeared. When portfolio margining came into existence, a lot of independent firms withdrew their broker dealers and began making markets as customer firms. We still post markets and use algorithms, just not as official market makers.

HFT firms ARE market makers - even if they don't have the official SEC designation - and every firm's goal is to beat their competition. I know of no equity market maker (designated or otherwise) that doesn't employ algorithms.

I will pick one nit with Another HFT Trader's response. It was electronic trading, not specifically the HFT's algos that were the demise of floor traders. Back in the days of the floors all orders were funneled through the crowd on the floor where there was a tremendous amount of collusion (particularly in not very liquid stocks) and orders were regularly raped (there's just no softer way to put it). Electronic trading eliminated the cozy relationships on the floor and increased competition. Many floor traders were just not good enough traders to adapt to a more competitive environment. Those who were former floor traders and successfully left the floor employed algorithms to compete in todays much faster and more competitive environment. While the ex-floor guys who couldn't adjust moan at every opportunity, customers benefit by a more anonymous and much more competitive market.

HFT guys often claim credit for the tight bid-offer spreads in recent years. But how much of that is also due to the markets becoming electronic per se versus the actual HFT part? Is there a way to tease out the effects?

Also, do the non-HFT algorithmic traders contribute to the tighter spreads? Or not at all.

HFT can pretty much be defined as the people that provide liquidity in an electronic market. You would need to articulate what exactly you imagine a non-HFT electronic market to be, specifically how would liquidity be provided. Various anti-HFT measures, like transaction taxes, minimum holding times, etc. wouldn't necessarily lead to a paradigm shift, but simply the same electronic algorithms adding liquidity to the book. The only difference is that they'd adjust their parameters to reflect the additional burden or the regulation or tax. (To the great reduction of market liquidity).

Non-HFT algorithmic trading refers to people executing an order using an electronic strategy. Since they're working a fixed orders they're still liquidity takers int he sense that they don't have the discretion to ultimate buy or sell depending on market demand. Non-HFT algorithmic traders are the customers of HFT, not the competitors.

...says the man who profits from HFT.

As for the bad effects of HFT, see the Wikipedia 2010 Flash Crash entry, and note that HFT only is beneficial when there's nothing to lose, but harmful exactly when you need volatility. In short, HFT is parasitic.

Ray, are you going to bother responding to the last comment I left on this exact same conversation we had in the previous thread. Or are you just going to keep making the same arguments, then I'm going to make the same responses, then you're going to leave the thread?

Projection noted HFTT. Besides even Goldman is against HFT, see "Fault Runs Deep" in today's Dealbook. HFT benefits a subset of investors, in the same way insider trading does. You may argue that insider trading is not bad (as I have) since Coase's theorem of economics says society is no better or worse off if a rich person makes money off a stock tip rather than a random grandma (since grandma sold her stock regardless of the inside information, as is self-evident, and in countries where insider trading is legal, there's no real evidence it discourages the retail investor), but that's not the world we live in.

In short: if you are against insider trading--and I bet you are--you have no moral defense for HFT, which also suffers from the same moral arguments and further suffers from demonstrated physical bouts of injecting instability into the system. You have no argument against that, never did.

Ray, were you ever going to respond to the challenge I left to your comment on the previous thread? Or are you just going to keep changing the subject.

My knee-jerk reaction is that if Goldman is against something, I'm probably more inclined to favor "it."

That said, your reading comprehension is sorely lacking if your takeaway is that Goldman is "against HFT."

I'm wading into this against my better judgment, but here goes. When Ray says "HFT benefits a subset of investors", we can replace "HFT" with any profitable strategy without changing the truth of the statement. Markets are fundamentally zero sum, and every dollar of my outperformance is matched by precisely one dollar of someone else's underperformance. When you take issue with the "moral problems" of insider trading and HFT, you seem to be upset that people have gone to enormous lengths to obtain and use information that others don't have. Note, these things aren't closed to anyone. If Joe Schmoe really wanted to, he could go and cultivate an expert network like Stevie Cohen did, or he could go and pore over statistics papers each and every night like I do. If he doesn't do these things, is he right to complain when he inevitably loses to those who do? You appear to suggest, yes, he is. Care to explain?

I'll leave you with a choice quote from Ray Dalio: "We have 1,500 people that work at Bridgewater, we spend hundreds of millions of dollars on research... We've been doing this for 37 years... if you're going to come to the poker table, you're going to have to beat me"

I know what you wrote is meant for Ray, so he should answer.

That said, all of these trading strategies are antithetical - maybe that's not the correct word - to the main purpose of the stock market with regard to our capitalist society.

The market exists because companies need to raise capital and the market is a way to efficiently draw in capital that otherwise sits on the sidelines - in bank accounts, under mattresses, whatever. And further, with feedback to maximize the efficient allocation of that capital: invest wisely and reap the rewards. Invest unwisely and suffer the consequences.

None of what I'm saying is in any way a suggestion that I personally favor more regulation, or taxing, or banning of something because it is 'antithetical' (or maybe 'peripheral' is a better word) to the core purpose of the market.

It just seems to me an incredibly inefficient use of not only capital, but talent to have the best and brightest (and best infrastructure imaginable) devoted to activities that are, let's be frank, completely peripheral to the core purpose of the market. I'm not convinced that HFT does much in the way of improving the efficient allocation of capital.

I guess you can argue that those hundredths of pennies saved add up if you're a large entity buying millions of shares which could then be used to 'invest' in another company. That still doesn't account for the deadweight loss of having people like you, who could probably be brilliant in a variety of other fields, focused on schemes to wring out a little extra alpha.

This isn't meant to be preachy or prescriptive, or as I said, proscriptive. I don't know - just seems like a large waste of talent and resources.

What you say is very true but, I think, far more applicable to the primary markets. This work is done by i-banks, VCs, angel investors, etc. I tend to think that's where capitalism really happens. The secondary markets (like our stock exchanges) are only important insofar as they allow primary investors to pass their securities on to the next guy. This is why liquidity is valuable--it encourages real capitalists to put money into enterprises by lowering the barriers to get money back. I admit that it can all seem like a stupid paper shuffle (and I'm not going to argue that the amount of shuffling going on right now is in any way optimal), but it is towards a useful purpose.

I think it's fair to worry about the talent wasted on this enterprise. If it helps at all, most people don't last all that long in this industry. Trades are always going away, and profits are always decaying towards 0. The skills gained in the business are readily applicable to problems faced by the Google's and Amazon's of the world. That is, I think most traders/quants/programmers will eventually end up contributing to the real world.

@HFT Trader

One question, possibly naive, is how exactly do you define liquidity? In another thread you mentioned volume as a reliable surrogate for liquidity (if I read you correctly!) But is that true? For example what fraction of typical trades are HFT-to-HFT? And should such transactions really count in a useful metric of market liquidity?

Also, the way I understand it, HFT firms have no traditional contractual obligation to play market maker, right? Say, unlike a traditional designated market maker or specialist? If so, an HFT is free to pack up and abandon it's liquidity injecting obligations when things get messy? How robust is the liquidity injection function of such HFTs?

I'm not saying if it is wise to have someone obligated to inject liquidity or not but my only point is: Are we mistaking churn during the good times for real liquidity that is oft needed under times of stress?

The below linked paper found that in the S&P E-mini market intra HFT trading accounted for 18.61% of the volume (Table 2). Even assuming that pre-HFT intra-market maker volume was zero, that still means that the amount of trading done by non-HFT liquidity consumers has increased by an order of magnitude. Liquidity is hard to define, but I'd say its the ability to trade, potentially in size, without significantly incurring market impact cost. The reason I'd say volume is a good metric is because if it goes up it means either the supply has had to have gone up or the demand for it. There's no compelling story why the demand for volume exploded from ten years ago (there was maybe 2008, but not in 2014). Plus we know transaction costs, which are a pretty good proxy for the cost of liquidity, have substantially fallen. All of that strongly indicates that there's been a massive increase, shifting the supply curve for liquidity.

As for market maker obligations, I made a point about it from the last thread. Basically obligating market makers to make uneconomic quotes at certain times will raise the amount they charge for liquidity in other times. So it's not a free lunch. Also designated market maker or not, you can't ultimately stop prices from adjusting if the market's preferences truly change. Even if you force market makers to price floor if the new price equilibrium settles below that they are going to sell as soon as they legally can and the price will go through the floor. You'll just get there slower and in a much messier manner. If you're talking about preventing erroneous or panicked trading than circuit breakers are a much better option, with much lower economic costs, than forcing market makers to provide a price floor.

Finally I'm not sure why everybody is so sure HFT withdrawals in times of stress. On the one hand there is the Flash Crash, but on the other hand there's the entire year of 2008, which was the most macroeconomically stressful time for the stock market in the post-war period. HFT activity substantially increased that year, not withdrew, and the equity markets maintained total liquidity availability without fail, unlike the non-electronic markets.

Just prior to the beginning of the flash crash, NYSE data began to be significantly delayed relative to other markets. The CQS data feed, which is responsible for determining the NBBO, used the delayed (wrong) NYSE in their calculations. Over half of the stocks listed on the NYSE had delayed data. As a consequence, the bid/ask spread was crossed in hundreds of stocks. Worse, the CQS data, which many algorithms rely on, did not indicate that the NYSE data was delayed. They just reported nonsensical severely crossed markets. If you were trading algorithmically at the time, whether HF or otherwise, you quickly became aware that there was a problem with the market data. At the time, it wasn't clear what the problem was exactly, but it was very clear that the data was bad. Because the data was obviously bad, many trading firms stopped trading. The alternative, letting the algorithms run, despite knowing that the data they were being fed was wrong, would have been irresponsible. So the flash crash had nothing to do with HF firms "pulling back" because the market was crashing, they "pulled back" because the data they were getting from the exchanges and the consolidated feed was obviously wrong. You could say this is an overall failure of automation coupled with the fact that there are now multiple exchanges, and that if people were still trading by calling each other on the telephone or yelling across a pit maybe this wouldn't have happened. But would the correct lesson to be drawn from this to be to ban automation? Sometime in the possibly near future, a robot driver is going to run over a pedestrian (possibly because a network that the robot uses to get information goes down or is delayed). Will the rational response be to ban robots from driving?

There are more details here:

@HFTT - you lost in both your responses to Rahul. This is typical of the way you "debate"--are you really passionate about HFT or just leading with your glass jaw in the boxing match? Because you are getting knocked out worse than Bradley will against Pacquino.

First point in your rebuttal to Rahul: "well, mini-SP contracts have increased in popularity therefore HFT, which sometimes uses these contracts, have ipso facto increased liquidity" (WRONG LOGIC, applies and oranges)

Second point: "market makers, which have a contractual obligation to do trades (even if they break the rules by not trading during crashes, but that's an argument to penalize these firms, not an argument to do away with market makers), unlike HFT which has no contractual relationship, is more expensive, so HFT is better" (WRONG LOGIC: mufflers on cars also decrease engine power but nobody save street drag racers want to go to a world without mufflers, since the ear pollution will be deafening).

"As a rule, small investors are helped by HFT: they get filled immediately, at NBBO. (NBBO is National Best Bid/Offer: basically, the very best price in the market.) It’s big investors who get hurt by HFT"

It's not big investors inasmuch as it's informed investors, i.e. investors who's trades predict the direction on the stock. Nearly all informed investors are big, but most big investors are not informed. For example index funds, even massive ones, are almost certainly helped by HFT. Even though they have to execute huge orders, they have no fast-decaying alpha, unlike sophisticated hedge funds. They can put in a highly patient VWAP order order and just let it slowly work, so as to incur little to not market impact. It's traders that have highly pertinent information that need to execute in fixed time frames that suffer from sophisticated market makers.

Overall this is an excellent benefit for market efficiency. Uninformed traders by definition have no information to add, but informed traders do. To the extent that prices respond more to the latter and less to the former it reduces noise in the price discovery process and leads to more accurate allocation of capital.

the anti-HFT arguments are unconvincing after you separate the hype from facts . there's hardy any arbitrage opportunities in the markets because all the exchanges are competing with each other which reduces delay/friction

I have tried to follow the writers cited, as well as people like the HFT Trader who commented extensively on yesterday's item. But they all seem to be saying, "This is how HFT trading *really* works, trust me, I know," and I don't see how we're supposed to evaluate competing claims.

The same way you evaluate all claims: to see if they make sense and are coherent. Many opponents of HFT have no coherent story about the harm it supposedly does. Also you can observe that actual trading costs for ordinary investors are at historic lows, so if there are any harms it is mostly the big boys vs the other big boys.

(Not to say that there aren't some informed critics who have concrete and useful ideas for fixing real problems. And the flash crash wasn't exactly a shining moment for HFT either.)

But many of the claims are based on insider knowledge or sources. Lewis says orders are routed a certain way that makes them vulnerable to front running, HFT Trader says Lewis doesn't know what he's talking about, Salmon says Lewis is years out of date. All of these claims seem coherent to me. I'm being sincere when I ask whether someone without personal knowledge of the industry can make a sound evaluation.

I think that's a general observation about most controversial, evolving, "hot" issues. I don't think there is any easy solution. Try making a judgement call which of Lewis, Salmon, HFT Trader you find most credible. Or whose arguments you find most convincing. Are you conflicted on the facts or the arguments? Perhaps just wait for a while and a consensus will emerge in the field?

How? That gray matter between your ears?

There should be no great mystery here. Exchanges publish documents detailing everything one would need to know to trade on them. For example, here is information for BATS and Nasdaq:
If you're interested in the theory behind it all, there's a considerable literature on market microstructure ("Trading and Exchanges" by Larry Harris is a good place to start). There are no good reasons a supposed finance professional like Katsuyama should be surprised by the way his orders are handled. It's staggers me that people entrust their savings to the Katsuyamas of the world--money managers who couldn't even be bothered to learn about the sandbox they're playing in.

Yes, the HFT's make less money than the traditional market makers; but OTOH isn't the risk of making catastrophic losses lower too?

Why is the fact worth mention? Aren't we just talking of riding a different point on the risk-reward tradeoff curve?

That's exactly the point. Reducing the risk is what enabled returns/prices to be bid down. Would you prefer a world where investors pay more for market making, but every once in a while an insider makes a huge profit off the market makers?

You obviously have no concept of Falkenstein's ideas or writings.

Right - it's not as if he's making his thoughts pretty clear right here in these threads.

Let me be more clear. 'Bootleggers and Baptists' is a reference to the economist Bruce Yandle's observation that Baptists and other evangelical Christians were prominent in political activism for Sunday closing laws restricting the sale of alcohol. Bootleggers sold alcohol illegally, and got more business if legal sales were restricted. “Such a coalition makes it easier for politicians to favor both groups. … The Baptists lower the costs of favor-seeking for the bootleggers, because politicians can pose as being motivated purely by the public interest even while they promote the interests of well-funded businesses. … Baptists take the moral high ground, while the bootleggers persuade the politicians quietly, behind closed doors.”

Katsuyama's rhetoric is high minded but helps insiders, specialists at the major exchanges. Thus, they love him, but they are surely not sympathetic to your average socialist, and hardly better for widows and orphans or anyone else. I worked on high frequency algorithms in my last job, and it was all about trying to minimize trading costs taking into account the strategic, predictable behavior of others, just as everyone else did. To say that was rigged suggests there's a conspiracy, but we all knew about the importance of having access to retail flow, speed, and better algorithms. The former you can buy, and you better believe it was always sold this to the highest bidder (ie, Schwab's flow is worth a fortune because it's full of noobs, they knew it and sold them out, as all retail brokers did). HFT are basically doing what the old specialists did, at a much lower cost to the consumer because unlike specialists they don't have a monopoly.

I have a belief, a prejudice, for free markets because I believe competition is the best regulator. I am not for 'the financial industry' per se, many of whom, like NYSE specialists, are pathetic crony capitalists, or simply like Dilbert's pointy haired boss, clueless bureaucrats. Why isn't anyone beating up on the CME, which still has a monopoly and a huge market cap? Clearly much more consumer surplus is being lost there, but because there is no competition and so no 'losers', and they can buy nice PR no one seems to mind even thought it generates an order of magnitude of greater costs to the trading public. Why isn't Washington trying to reduce the size of the big banks, rather than create trillion dollar monstrosities? Because those banks are both big enough to benefit from federal legislation and few enough to monitor and control.

Finance is a big industry, and government's involvement is invariably for less competition in the name of fairness, with benefits shared between politicians and elite bank managers: shareholders and consumers lose.

Lewis claims he has no financial stake in IEX.

I always wondered why Billy Beane is the hero of Moneyball and not the villian. The players who were cheap weren't valued properly by other teams, but when Billy saw their 'true' value - he didn't pay them any more. Moneyball doesn't work if the market DOES work and the players are compensated for their realized fair market vlaue.

To answer my own question: I think Beane is the hero of Moneyball because Michael Lewis liked him.

It's a compelling story because he built a championship contender with one of, if not the lowest, salary in baseball - going up against teams spending multiples as the A's did on player salaries. It's a David vs. Goliath story.

There is no room for rational thought in the Democratic Party. They believe if you change a subsidy from being paid to the supplier to being paid to the demand the entity that accrues the benefit of the subsidy will change.

As someone who has lost lots of money with stocks, the speed of the trading makes no difference. HFT can only helps retail investors like myself by keeping trading costs low

Katsuyama is acting obnoxious in this video. painful to watch

from the NYT article in OP: "It used to be that when his trading screens showed 10,000 shares of Intel offered at $22 a share, it meant that he could buy 10,000 shares of Intel for $22 a share. He had only to push a button. By the spring of 2007, however, when he pushed the button to complete a trade, the offers would vanish."

I'm a small time investor, but I had a similar event to that above that made me wonder for the first time about what was happening several years ago. It was an after-hours trade in MSFT. I normally don't trade in after hours - but I was looking to sell MSFT and there was a bid at a certain price that had set in after hours for maybe 10 of 15 minutes. Very little after hours volume and the bid sat out there for a long time and had enough shares to cover my order. I decided to go ahead and sell and I hit the bid price on my sell order (I placed the order to sell at the posted bid price) - and immediately when I processed my sell order the bid disappeared and a new price was quoted. I'd had an idea before that somebody could see my trade coming and react to it before it got there - but given this particular situation I became more suspicious that something odd had just happened. What were the odds that the after hours bid to buy would sit out there for a long time like that, and then immediately disappear when my order was entered? When things like that happen in the middle of the day you just think that somebody hit the bid before you and the price moved, but in that situation with almost no volume in after hours - I couldn't understand it.

After hours trading is not governed by Reg NMS. I honestly don't know the details of what happened in your scenario, what broker you used, whether they tried to route it through a dark pool prior, etc. But without Reg NMS protection some shady sh*t goes down (like the pre-2007 market), I'd honestly advise anyone not to trade outside market hours without a compelling reason.

thanks for the info. I'll avoid after hours going forward.

You probably hit a stale bid. Because data providers don't update as frequently outside of market hours, you were probably looking at a bid that didn't actually exist anymore. This is the most likely scenario.

It's possible that the bid was real and faded when it saw your order, but you have to remember that the rules that apply during market hours don't apply after market hours. For instance, there's no requirement for designated market makers to honour their markets should they have them up after the market closes. Unless you're pretty sure of what you're doing, it's not a great idea to trade outside of market hours because if you make a mistake and trade at a price far away from the last print the only thing you can do is beg the other side of the trade to bust the trade and they usually won't,

Layman question: Why does "after hours trading" exist as a separate category any more? Now that market making is mostly all algorithmic & so is most real time compliance or circuit breaking, why not let markets run 24x7?

Why is this gray market of trades not governed by RegNMS maintained at all? Just curious? Isn't the discontinuity imposed by a closing / opening hour a destabilizing influence especially if material information happens to be generated in the after hours?

"Layman question: Why does “after hours trading” exist as a separate category any more?"

the answer is simple: because exchanges are not open around the clock. Once the exchange closes, the information flow drops significantly (information flow includes orders) and information flow is critical for making markets. You'll still often see markets, but they'll be incredibly wide.

It's expensive to maintain compliance with all regulation - including RegNMS - and it's expensive to have a market up all the time. Once the exchange closes, it is no longer responsible for routing your orders. It's closed! It's just not economic for exchanges to run 24/7.

And, no, there's not really a discontinuity problem because if news comes out and enough people care, they'll trade after hours. Also, stocks that people care about are listed globally and you can always find a way to take advantage of information either through derivative products that are trading or on exchanges that are open.

To have an exchange "run" these days mostly means whirring servers to run algorithms, right? And associated skeletal IT staff etc.

Is the incremental cost of a 24x7 uptime really so high over status quo?, I'd be surprised. The data center stays up & running anyways I bet. It's not likely that they cycle hardware down every night.

No. wrong. Very wrong. You obviously don't know anything about the economics of running an exchange.

bu that isn't hft because no trade was completed, so what you're describing is different

pulling the bids is an old manipulation technique that dates to the early 20th century used by short sellers

the trader places a huge limit buy order to create a floor and waits for other traders to bid the price up because the floor makes them feel confident. Then he cancels the buy order, turns around, and shorts the stock. now the price falls and there is no floor. panic ensues and the price could fall a lot

not sure if related to HFT or not. The bid seemed to disappear almost the moment I submitted my sell. I wonder if the order was viewed on one exchange - then before my order had time to reach the other exchange where the after hours order was sitting - was the order changed in the interim milliseconds? (after hours may not even use alternate exchanges - so I don't know - total speculation)

What I don't understand is how (or even why) a bid to buy disappears when a matching sell order shows up. The article seems to provide an explanation of order arriving in different markets at different time, and the fast market is used to gather data, and the other slower market used to front run the order (hypothetically). I'll heed advice to avoid after hours altogether - although I have had other trades in after hours that have worked out seemingly fine. The time frame of this order was 2010 I think based on my records - about the time the guys at RBC were figuring this out.

btw - this was a tiny order - 600 shares according to my records.


This is highly unlikely. Not only is the trade you describe virtually impossible to profit from, but stale bids are extremely common after the close. Extremely common and all result from slower updates from datafeeds. Thus, an innocent stale bid is the most probable explanation by far. Happens every day in virtually every stock.

Why is it that people immediately jump to some far-fetched nefarious explanation when the simple and most probable answer is staring right at them? Is it because the truth is boring?

Your broker fucked up, simple as that.

The broker routes the trade according to your order ticket. What else would you have the broker do? Why do you people always go with "somebody fucked up or fucked you over" when something you don't understand happens?

"even if the HFT shops are making good money, they’re still making less than the big broker-dealers used to make back in the day. But that’s not a calculation Lewis seems to have any interest in."

The profit margins of information arbitrage in the past are not relevant to the profit margins of information arbitrage in the present if one's goal is simply to diminish or eliminate opportunities for information arbitrage.

"gut instinct-driven" market making? Say what now?

I have been a market maker for a very long time and this is the first I've heard of Goldman's (or any market maker's who actually stays in business) "gut instinct-driven" anything. What kind of silliness are you imagining? This is pure nonsense.

In the very very old days, when computing power was pretty much non-existent, market makers had to do a lot of guessing. Because they knew their computations of fair value had a greater degree of uncertainty they made wider markets to compensate them for the large risk they were taking to make a market at all. However, this is a long gone era that was disappearing as I entered the business (a long time ago. Before the internet became a thing).

Also, you seem to be entirely unwaware of the fact that Goldman got out of the floor market making business entirely and is actually one of the largest HFT operations in business. Goldman's "Sigma X" is huge and does enormous amount of volume. So, while I agree entirely that HFT is unfairly maligned, it's pure fantasy that smaller HFT shops are somehow undercutting something to do with gut instinct at investment banks.

The reason spreads have narrowed is because our ability to compute fair value in real time has improved, electronic markets have increased competition and the use of algorithms have lowered the cost of making markets tremendously (our algorithms replace several expensive traders and developers and programmers are cheap in comparison to a good trader). The "old school" traders you're imagining have disappeared long ago and Goldman is no longer making markets in stocks. Being a designated market maker in stocks is a regulatory and tax nightmare. HFT guys perform the same function without having to submit to the regulator as fully as a designated market maker. And that also lowers costs, allowing firms to provide liquidity on thinner margins. As usual, it's competition that is the giver of all goodies, but you have the players all wrong, Tyler. The big i-banks all have huge HFT operations. And there's not a whiff of "gut instinct" in any of them.

How do you compute the fair value of IBM or Microsoft or a railroad in real time using a high speed computer?

Is the value of real estate fluctuating wildly by the second? Are the computers tallying up the marginal value of each sale in the seconds after the deal is final?

Or are we talking of the abstraction of derived value in the shares of stock which increasingly are driven by the gut value of millions of people who are tossing cash into 401Ks and IRAs?

When I heard of computers calculating fair value I think of the computers calculating the value of mortgages in decaying cities backed by decaying houses using the history of mortgages on suburban and sustained communities. If the computers were looking at the Streetview photos from Google, maybe I wouldn't immediately thing gigo.

Millions of people are pouring money into stocks and bonds as part of savings programs with Wall Street bankers doing virtually nothing to promote the building of new capital assets. So, money is chasing scarce assets, the fixed pool of stocks and bonds, and that merely inflates those asset prices. Only by looking at the real assets from which the stocks are derived from can you possibly determine how the price inflation should be distributed. No computers can do that determination. A corporation which fires all its R&D staff to boost profits is worth less, not more, because it will become increasingly obsolete because it is not investing to offset the depreciation that is inherent in all capital assets.

So, what I'm getting from your comment is that you have no idea how to value securities and because you have no idea how to do even a basic valuation you assume nobody does.

And based on your fantasies you want more regulation of something you have literally zero understanding of. Great. Do you also advise neurosurgeons and airline pilots because you mistake your fantasies for reality in those professions?

Real estate values would fluctuate rapidly if pricing was continuous. But how often does a house come on the market compared to a share in a company?

And consider how often homebuyers/sellers get screwed because residential real estate is an illiquid market.

After reading most of the comments on the two posts relating to HFT , I am still not clear whether HFT is a zero-sum game. If it is , who are the losers ? Its hard to believe that as much losses are there by losing HFT firms compared to those gaining.
If it is not a zero-sum game , is the nett positive amount coming up only because of a market that is moving up?

It's a zero-sum game. The HFT traders are parasites who add nothing to the system: during normal times they suck money out of the losers, and just add to "noise" , while during volatile times, when you need their participation, they go away. Pure parasites. Arguably they are not even neutral parasites, as evidenced by the flash crash of 2010 (which some people say they caused; others say it was one rogue trader, Knight Capital I believe).

Those "parasites" provide almost all of the liquidity that you rely on to trade your little portfolio and to keep cost capital and volatility down. If you eliminate those "parasites" then you will see inefficient, volatile markets and less willingness to invest long term. Oh, and more and larger bubbles. But please, by all means don't try to understand the economics of financial markets and go ahead and rely on what "some people" say about things you don't bother to even try to understand before you decide to burn them down.

Ray - re:parasites. I seem to recall you saying you were a lawyer?

I'm not a lawyer, but I sometimes play one on the internet.

HFT are at best zero-sum and at worse a net drag on society. They should be lightly taxed, to cut down on the churn.

So, Ray doesn't understand that the market makers will simply widen their market and pass along the tax to him. What Ray is really saying is he'd like to pay more for the stock that he's buying and get less for the stock that he's selling.

My bet is Ray is not a rich man.


Any one trade is zero sum. Trading as a whole is not.

First of all, the liquidity provided by HFT is a net positive because it reduces volatility, decreases transactions costs and reduced cost of capital for firms. Second, losing on one trade can be a good thing. Suppose you have a two-legged trade where you are long an undervalued stock and short a correlated overvalued stock. In other words, you have a hedged position. You may lose on the long but gain more on the short than you lost on the long. In that case, you are a net winner even though you lost on one leg of your trade.

Here's one point I don't understand: @HFT Trader & others maintain that to an infrequent retail investor with long horizons, any downsides of HFT's strategies are so small that they are almost irrelevant.

But, at the same time, HFT is said to provide liquidity advantages that benefit these same small investors.

Why is this an asymmetric argument? i.e. If I trade infrequently enough why does the tiny bit of liquidity or fractional cent of improved price discovery matter to me?

What part of this does not compute for you exactly? A tiny downside against a large gain is a net gain for customers. The alternative is no HFT and much larger transactions costs, higher risk, more volatility and larger cost of capital - so, a HUGE downside. The large net upside is better than a huge net downside. Pretty elementary.

You are screaming bloody murder that you're being forced to pay fractions of a penny to liquidity providers (because you completely nonsensically think that this valuable thing should be provided to you for free) when the alternative is paying whole percentage points for liquidity. Since you're so into empirical evidence, have a look at illiquid securities and see what kind of price you pay for liquidity there. Watch as slight changes in the size of orders drives the price up and down by several percentage points. Have fun losing several percentage points to exit or enter a position.

Joe Schmoe could trade 100 shares of MSFT just fine with the same spread and liquidity without the existence of HFT. I don't see where this amazing improvement in spreads and liquidity has done anything for him.

Any look at financial markets over the time period that HFT has been around should dispel the idea that they have had a meaningful effect on "volatility" or "price discovery".

Liquidity itself is just choice. Choice itself is not a good, its only good if it allows the actual item being selected to be better. If additional choice doesn't lead to better decisions then its meaningless, and there can be costs associated with all that extra volume of transactions that are negative. Having algos battle the shit out of each other doesn't help Joe get a better price on his 100 shares of MSFT, nor does it help him figure out if MSFT is the right price, nor does it make MSFT go up over the next ten years for when he retires. To the extent that ease of transaction causes Joe to frantically trade his assets it's probably even a negative.

"Joe Schmoe could trade 100 shares of MSFT just fine with the same spread and liquidity without the existence of HFT. I don’t see where this amazing improvement in spreads and liquidity has done anything for him."

Well, if you, from your completely uninformed vantage point, don't see the benefit, then none exists, right? You seem to not grasp the benefits of competition.

"Any look at financial markets over the time period that HFT has been around should dispel the idea that they have had a meaningful effect on “volatility” or “price discovery”."

Over the past twenty years since algorithmic trading (the heart of HFT) has been in widespread use, the market has become far more liquid and transactions costs have dropped precipitously. The fact that you no longer have to cross a giant bid/ask spread to trade means exactly that volatility has fallen. Does that mean that prices don't move? Of course not. Information is what drives valuation, but the lower bid ask spread has incrementally lowered volatility. By definition. Does liquidity mean that prices move to fair value faster? Yes. That means the probability that you, Joe Schmoe, will buy on stale valuations is lower. If you don't see the benefit in that then I'm very sorry for you because you seem to prefer a world where I stood in a small crowd on a trading floor and raped your order. The advent of algorithmic trading has reduced my margins by making the market more competitive and more efficient - all to the benefit of customers, mainly Joe Schmoe. My loss (of huge margins) is your gain. I'm sorry for you that you bemoan that fact.

"Liquidity itself is just choice."

So, what you're saying is that you don't know what liquidity is.

" Choice itself is not a good, its only good if it allows the actual item being selected to be better."

Ummm....the "item" you're "selecting" is a security and a share of, say, Apple is just like any other share of Apple. It's not like selecting edible apples at Whole Foods where you might choose one that is more ripe or relatively less bruised. Market makers compete with each other to trade with you. The only way to do that is to better the market of our fellow competitors and, in the end, you are not compelled to trade with any of us unless we offer you the quantity you want at the price you're willing to pay.

Having "algos battle the shit out of each other" to compete for your business benefits you enormously because it lowers the price of trading for you. Do you have something against paying less? Do you prefer to be abused by a crowd of colluding market makers on an exchange floor? I'm curious why you think paying a lot more to be abused by a crowd of market makers is so much better for you? Take me through that logic of yours, please.

"...nor does it help him figure out if MSFT is the right price,"

Um....more opinions in the form of more market participants and more bids and offers ENSURES that you're trading at the right price. The more opinions you factor in, the more correct you can assume the price to be. Where you don't get that benefit is in illiquid stocks where there are few liquidity providers because it's not worth their time to write the algos.

If you are so stupid that the liquidity alone drives you to frantically trade in and out of a security, then I suggest you need to either not trade at all because you can't handle it or you have a very strange self-control problem and need to be locked up in a mental institution where people who have some impulse control can make sure you don't hurt yourself. The financial markets don't exist to protect morons from themselves. Although, the existence of all these traders you so hate ensures that if you do manage to lose control of yourself, you'll lose minimally.

@Methinks: "The fact that you no longer have to cross a giant bid/ask spread to trade means exactly that volatility has fallen" - and why is this a big deal to Joe Schmoe? The supposed drop in volatility (which btw is not really true, Google Great Moderation and note this ended around 2008) benefits big players, not small ones. Joe, who trades a few times a year, was/is perfectly OK with Merrill Lynch taking $25 commission to trade his stocks, even before the Big Bang reduced commissions. Please, quit the strawman arguments. As for bid/ask, last time I tried to buy over $100k in an ETF, I got burned since the transaction was too big for the relatively illiquid ETF I was buying, and ended up buying it in chunks at rather outrageous bid/ask spreads, and this was just a few years ago. So much for your much vaunted "liquidity".

"You seem to not grasp the benefits of competition."

And you seem to think competition is "magic". I've worked in lots of industries where more competition lower prices. In fact one where it was a big driver of prices going way up (because without effective bargaining power the supplier can rape you and you have to pass that along to the consumer).

Even the article mentions things like travel industry pricing. As someone who worked on models for that industry and made pricing decisions I can tell you for a fact that I did projects designed to rip customers and competitors off. Once you get some data you find that there are a lot of good ways to manipulate and take advantage.

Margins fell because the internet became a big thing amongst other factors. We wouldn't be trading 1/32nds if HFT didn't exist, so don't take credit for something you didn't do.

"more opinions in the form of more market participants and more bids and offers ENSURES that you’re trading at the right price"

This is fucking retarded. Did "more opinions" keep tech from being a bubble or housing from being overpriced?

*I’ve worked in lots of industries where more competition *doesn't* lower prices.

Gee, I don't know, Ray. Why is it a big deal when you pay less for housing, food, and education? Since your argument is that it doesn't matter to you how much you pay for anything, then I assume that you're as happy paying $100/oz. for your steak as you are $0.10/oz for the same steak. And, of course, when gasoline prices spike, you cheer and claim that you are unaffected. Only the big guys who eat a lot of meat and drive 18-wheelers are impacted by price changes. The bid/ask spread is what you pay to trade or what the money manager you hire to trade your retirement pays to trade.

You also have a very pedestrian view of volatility. In your estimation, incrementally lowering the volatility inherent in higher bid/ask spreads is useless because narrowing the spread doesn't remove all of life's uncertainty and the changes to valuation that brings. Since things still change even if spreads tighten, the tightening is totally useless, according to you.

I'm sorry. I didn't realize that you were hired by Joe to be his agent. That high commission and a large bid/ask spread eats into Joe's returns. It means he gets less for the effort of trading and his retirement egg doesn't grow as fast or as large and every Joe I've ever met cares very deeply about his own portfolio. If he didn't brokers wouldn't bother competing for customers by lowering commissions and they wouldn't bother competing for them on execution either. But they do. Huh. Why do you suppose that is if Joe doesn't give a damn if his return is severely damaged by commissions and crossing spreads?

I don't think you understand what "straw man" means, dearie. But, what's most hilarious of all is that after all of your assertions that Joe Schmoe doesn't care about being screwed by high spreads and commissions, you end your comment by bitching about getting screwed by huge spreads resulting from illiquidity. So, you don't care about liquidity but you scream bloody murder when there isn't any around. What happened? I thought only the big boys suffered because of that. All I can say is I hope that was my ETF you were trading.


Competition is not magic and it doesn't have to be magic to lower prices. In an undifferentiated product like securities the only way for competitors to compete with each other is price. If my competition offers you a better price, you'll trade with him instead of me. So, I have to better his price to entice you to trade with me instead and stay in business. It's not magic, it's elementary economics.

While your meandering through the travel industry is both confusing and mildly amusing, you end up claiming competition doesn't lower prices and then you destroy your own argument by claiming that competition from the internet lowered prices for consumers despite the industry's best efforts.

As for the internet bubble, I'll let you chew on this (probably forever): New technology, fat tails. And as for the housing bubble, it's hard to imagine a market more manipulated, less liquid and more impossible to short than real estate. There's nothing in those two bubbles that contradicts anything I said, but it does show your general lack of understanding of markets and elementary economics in general.

An empirical question: In Sept. 2013 Italy introduced a transaction tax on sub-second HFT positions. Is there any empirical analysis of market performance before & after this discontinuity? What do the HFT insiders think about the law? Did it have any impact?

I find very elegant & often temptingly convincing all the pro-HFT arguments, but maybe this is a unique opportunity to inject some empiricism into the debate?

liquidity was reduced in the affected securities, spreads widened and more trading moved to other exchanges around the world. You see, there's still no such thing as a free lunch. Liquidity providers merely pass along that transactions tax to customers - whether those providers are HFT or not. Period. You seem to be particularly attached to the fantastic and laughable idea that providing liquidity (as well as an exchange for you to send orders to) is costless. This is quite an amazing assumption. For an adult.

"For an adult" is not a complete sentence child. And your spittle is showing in your venomous replies. It's an end of an era for you--time to get a job that actually contributes something to society rather than the Wall Street job you have now.

Congratulations. You know how to formulate a complete sentence even if the content is laughable. You can't understand why anyone would not write a complete sentence because style eludes you, but that's the least of your problems.

Please continue to air your hatred of things you don't understand. I do very much hope to find you on the other side of my trades.

It certainly is likely that it is true that HFT has led to decreased transaction costs and more efficient operations, while enabling some scalpers to extract rents from practices that don't serve the publics interest.

Our take, unfortunately behind paywall


Can you elaborate more on this quote from your article:

The market structure that enables the HFTs and provides us with their benefits may also be one that risks technological calamity. The good news has been that regulators began to focus on this potential problem last year. ....Real work is necessary to improve and safeguard a complex and still reasonably new system.

What exactly are the potential problems & what's the nature of the improvements & safeguards we need?

My optimism about the regulator's ability to deliver on the hope of a less fragile market is very far below yours, Cliff. Long experience tells me the process will be hijacked and the regulator will in the end a.) not really understand the problem and b.) will be captured by large interests that will result in anti-competitive regulation. That is the history or regulation and that is my experience running a regulated firm. As soon as regulators get involved, my expectations drop significantly. Costs will go up, liquidity will go down - particularly where liquidity is most scarce arlready.

Btw, both Levine and Salmon great on this topic

Levine and Salmon are indeed doing an excellent job. In other news, the TABB Group is out with its take on HFT: If you and HFT Trader haven't gotten this into the skulls of the skeptics, I don't know what will, but it's worth a shot.

Don't know about the others, but @HFT Trader & you made me change my mind.

Rahul, I appreciate your curiosity and civility. It's fine that people remain unconvinced of the *benefits* of HFT. Frankly, most people I've met in the business think the social utility is pretty small. We're in it for money and intellectual challenge, not spiritual satisfaction (I apologize to other HF traders reading this whose views I'm mischaracterizing). However, I hope that I've convinced a number of people that the *issues* as commonly discussed are either irrelevant or exaggerated. Electronic trading isn't perfect but is a huge improvement over the non-computer status quo. At some point later today, I'll get around to trying to answer your question about risks and problems of the industry (there are a fair number).

In other news, it appears the article I linked to is no longer public. A summary can be found via a quick Google search.

The article mentions things like travel industry pricing. As someone who worked on models for that industry and made pricing decisions I can tell you for a fact that I did projects designed to rip customers off. Once you get some data you find that there are a lot of good ways to manipulate and take advantage. In fact that was basically the point of what I was doing. Whereas before the consumer surplus got split more randomly now I was trying to soak up as much of it as I could for us at the customers expense. Believe me, at no point in these meetings did anyone ask if this was good for the customer.

"Electronic trading isn’t perfect but is a huge improvement over the non-computer status quo."

How did we go from "HFT" to "no computers". It seems like all the claimed benefits of HFT are mostly just the result of the internet being introduced to finance. These benefits would be in place even with fewer HFT shops.

@AHFT Trader--time to update your resume. The forces on the Hill are counterattacking and you are in the same position Drexel Burnham Lambert was in the late 1980s when they tried to argue their greenmail and LBO activity was socially beneficial. Well it turns out it was not: 50% of all mergers and LBOs fail, which means it's a coin toss, and further every corporation has fat that can be trimmed, but not by loading a corporation with junk bonds and forcing them to fire people to pay off the bonds. End of an era, dinosaur. And don't let the door hit you on the way out.

The "forces on the hill" aren't touching their donors on Wall Street and they're not as willing to destroy American financial markets to please ignorant peasants like you as you'd like to imagine. HFT is going nowhere and after the flash crash, the SEC required all designated market makers to become HFT shops. Did you get that? Broker Dealers in the business of providing liquidity are now required by your beloved regulators and monkeys on The Hill to use algorithms to post markets. Mandatory (also not a sentence!). The technology is here to stay. You luddites are not going to win this war on technology either. You're just going to have to grit your teeth and suffer the benefit of not getting screwed by enormous bid/ask spreads where HFT operates. Woe is you.

Also, great article. Always nice to see measured answers to complex questions.


I used to work in an industry where I experienced the same thing. Pricing strategy was designed price discriminate, and we tried to find ways to maximize that - usually focusing on switching costs and segments that. It creates an environment where if you are a consumer you should give no loyalty to a company, because the company knows you have become loyal and will charge you more than they would otherwise.

There's also Vanguard's statement from 2010:

"While Vanguard does not engage in [high-frequency] trading, we recognize that such trading has a positive impact on the markets at large, including longer term investors. Such arbitrage trading enables investors to get a fair price across market centers. Vanguard believes that the market structure changes facilitated by the Commission's various regulatory initiatives and the "knitting" together of the marketplace by "high frequency trading," have led to a significant decline in transaction costs for long-term investors over the past ten years through increased liquidity and tighter bid-ask spreads. We conservatively estimate that transaction costs have declined 50 bps, or 100 bps round trip [over the last 10-15 years]. This reduction in transaction costs provides a substantial benefit to investors in the form of higher net returns."

How much of this "100 bps" (seems too high) over the last "10-15 years" (why such a big gap?) is due to technology such as dark pools, the internet, the major exchanges doing away with floor traders and/or automating order matching, and how much is due to HFT? I would say 90% is from the former and only 10% from HFT. Time for a Tobin Tax on HFT. If it's so socially beneficial, then it can survive a small transaction tax, can't it?

I'd reverse your numbers.

Hong Kong has a hefty stamp tax on share transactions, ten basis points for the buyer and seller each.

How do the median spreads on liquid HK stocks compare to the median spreads of liquid US stocks? To liquid stocks that trade in other Asian markets like Tokyo and Sydney? Any thoughts on why HK spreads might be so different?

You don't have to sign up for membership in their website, except for the free products you want to get.
Confidence will bloom through sheer knowledge of selected
product's quality and usability, but also through well-targeted recipients.

net freebie forum is one of the most addictive on the web.

Comments for this post are closed