Home > #OWS, finance, hedge funds, rant > High frequency trading: how it happened, what’s wrong with it, and what we should do

High frequency trading: how it happened, what’s wrong with it, and what we should do

October 1, 2012

High frequency trading (HFT) is in the news. Politicians and regulators are thinking of doing something to slow stuff down. The problem is, it’s really complicated to understand it in depth and to add rules in a nuanced way. Instead we have to do something pretty simple and stupid if we want to do anything.

How it happened

In some ways HFT is the inevitable consequence of market forces – one has an advantage when one makes a good decision more quickly, so there was always going to be some pressure to speed up trading, to get that technological edge on the competition.

But there was something more at work here too. The NYSE exchange used to be a non-profit mutual, co-owned by every broker who worked there. When it transformed to a profit-seeking enterprise, and when other exchanges popped up in competition with it was the beginning of the age of HFT.

All of a sudden, to make an extra buck, it made sense to allow someone to be closer and have better access, for a hefty fee. And there was competition among the various exchanges for that excellent access. Eventually this market for exchange access culminated in the concept of co-location, whereby trading firms were allowed to put their trading algorithms on servers in the same room as the servers that executed the trades. This avoids those pesky speed-of-light issues when sitting across the street from the executing servers.

Not surprisingly, this has allowed the execution of trades to get into the mind-splittingly small timeframe of double-digit microseconds. That’s microseconds, where from wikipedia: “One microsecond is to one second as one second is to 11.54 days.”

What’s wrong with it

Turns out, when things get this fast, sometimes mistakes happen. Sometimes errors occur. I’m writing in the third-person passive voice because we are no longer talking directly about human involvement, or even, typically, a single algorithm, but rather the combination of a sea of algorithms which together can do unexpected things.

People know about the so-called “flash crash” and more recently Knight Capital’s trading debacle where an algorithm at opening bell went crazy with orders. But people on the inside, if you point out these events, might counter that “normal people didn’t lose money” at these events. The weirdness was mostly fixed after the fact, and anyway pension funds, which is where most normal people’s money lives, don’t ever trade in the thin opening bell market.

But there’s another, less well known example from September 30th, 2008, when the House rejected the bailout, shorting stocks were illegal, and the Dow dropped 778 points. The prices as such common big-ticket stocks such as Google plummeted and, in this case, pension funds lost big money. It’s true that some transactions were later nulled, but not all of them.

This happened because the market makers of the time had largely pulled their models out of the market after shorting became illegal – there was no “do this algorithm except make sure you’re never short” button on the algorithm, so once the rule was called, the traders could only turn it all of completely. As a result, the liquidity wasn’t there and the pension funds, thinking they were being smart to do their big trades at close, instead got completely walloped.

Keep this in mind, before you go blaming the politicians on this one because the immediate cause was the short-sighted short-selling ban: the HFT firms regularly pull out of the market in times of stress, or when they’re updating their algorithms, or just whenever they want. In other words, it’s liquidity when you need it least.

Moreover, just because two out of three times were relatively benign for the 99%, we should not conclude that there’s nothing potentially disastrous going on. The flash crash and Knight Capital have had impact, namely they serve as events which erode our trust in the system as a whole. The 2008 episode on top of that proved that yes, we can be the victims of the out-of-control machines fighting against each other.

Quite aside from the instability of the system, and how regular people get screwed by insiders (because after all, that’s not a new story at all, it’s just a new technology for an old story), let’s talk about resources. How much money and resources are being put into the HFT arena and how could those resources otherwise be used?

Putting aside the actual energy consumed by the industry, which is certainly non-trivial, let’s focus for a moment on money. It has been estimated that overall, HFT firms post about $80 billion in profits yearly, and that they make on the order of 10% profit on their technology investments. That would mean that there’s in the order of $800 billion being invested in HFT each year. Even if we highball the return at 25%, we still have more than $300 billion invested in this stuff.

And to what end?

Is that how much it’s really worth the small investor to have decreased bid-ask spreads when they go long Apple because they think the new iPhone will sell? What else could we be doing with $800 billion dollars? A couple of years of this could sell off all of the student debt in this country.

What should be done

Germany has recently announced a half-second minimum for posting an share order. This is eons in current time frames, and would drastically change how trading is done. They also want HFT algorithms to be registered with them. You know, so people can keep tabs on the algorithms and understand what they’re doing and how they might interact with each other.

Um, what? As a former quant, let me just say: this will not work. Not a chance in hell. If I want to obfuscate the actual goals of a model I’ve written, that’s easier than actually explaining it. Moreover, the half-second rule may sound good but it just means it’s a harder system to game, not that it won’t be gameable.

Other ideas have been brought forth as to how to slow down trading, but in the end it’s really hard to do: if you put in delays, there’s always going to be an algorithm employed which decides whose trade actually happens first, and so there will always be some advantage to speed, or to gaming the algorithm. It would be interesting but academically challenging to come up with a simple enough rule that would actually discourage people from engaging in technological warfare.

The only sure-fire way to make people think harder about trading so quickly and so often is a simple tax on transactions, often referred to as a Tobin Tax. This would make people have sufficient amount of faith in their trade to pay the tax on top of the expected value of the trade.

And we can’t just implement such a tax on one market, like they do for equities in London. It has to be on all exhange-traded markets, and moreover all reasonable markets should be exchange-traded.

Oh, and while I’m smoking crack, let me also say that when exchanges are found to have given certain of their customers better access to prices, the punishments for such illegal insider information should be more than $5 million dollars.

Categories: #OWS, finance, hedge funds, rant
  1. October 1, 2012 at 10:07 am

    “If I want to obfuscate the actual goals of a model I’ve written, that’s easier than actually explaining it. Moreover, the half-second rule may sound good but it just means it’s a harder system to game, not that it won’t be gameable.”

    “Obfuscating” and “gaming” nevertheless impose costs which will REDUCE unfairness. Re the Tobin Tax — Do nothing until we can do everything?


  2. October 1, 2012 at 10:08 am

    I disagree. Gaming increases unfairness because only the insiders know how to game things.


    • October 1, 2012 at 10:30 am

      I think you’re saying — The harder it is to game, the more advantage insiders have. That’s true, but irrelevant, if the opportunity is taken away.


      • October 1, 2012 at 10:33 am

        But the opportunity isn’t taken away with a complicated new rule. That’s the problem.


        • October 1, 2012 at 10:47 am

          The aim should be to open the game to as many as possible. Extending the minimum time does that.


  3. HFTer
    October 1, 2012 at 11:01 am

    No way your profit estimates are right — where do you get them? Is it for equities only?
    Tabb thinks HFTs make $1.8B (0.003% of its traded volume): http://banking.senate.gov/public/index.cfm?FuseAction=Files.View&FileStore_id=268afb1b-4c5e-4ec0-8028-fc840b973843
    Also, this is probably just trading profits, not net profits, so multiplying this by 4 to get IT costs would be wrong.

    As for transaction taxes, SEC already collects a tax of about 0.00125% (if memory serves). (Please compare it to the HFT profit above.) So do you want to make this tax larger? Add a new tax? How are you going to handle “bona-fide hedging” — is it exempt? If not, how would you delta-hedge options? How large would the tax be? Why wouldn’t it be just passed along to the “little guy”, pension funds and all?


  4. Ivan
    October 1, 2012 at 11:02 am

    > It has been estimated that overall, HFT firms post about $80 billion in profits yearly, and
    > that they make on the order of 10% profit on their technology investments. That would
    > mean that there’s in the order of $800 billion being invested in HFT each year.

    These numbers seem too high by an order of magnitude. Where do they come from? I have seen estimates of $7-8 billion in yearly profits which is still high.
    What you are saying is that around 5% of US GDP is invested into HFT each year. This just cannot be true, HFT is way smaller than this and does not employ lots of people.


  5. October 1, 2012 at 11:07 am

    My numbers may be off, I got them from an insider but I don’t have corroboration. Still huge money even if I’m off by 10x, but yes I’d like to get it right. This stuff is hard to ferret out because, of course, nobody comes out and says how much money they made. And I’m talking about everyone in HFT, which is a lot of people (and of course also depends on how you define HFT- does double digit millisecond instead of microsecond turnaround count? I’d say yes).

    In terms of the tax, yes I want to make it larger. It would be passed to the little guy, of course, but since those little guys barely ever trade compared to HFT’ers, it won’t be very important to them. In terms of how large it should be, that’s up for debate, but we could of course starts small and get larger. As long as we do it for all markets simultaneously so you guys can’t arb the taxes.


    • HFTer
      October 1, 2012 at 12:17 pm

      You can look at Knight’s numbers, since they are public. (It will be a vast overestimate, since Knight has several businesses in addition to its HFT operations.) For example, take 1500 people at Knight, times 20 — I’d think that would be an absolute upper bound of total number of people in HFT, including lawyers and secretaries. (Not a very tight one — with 1.8B profit estimate, it suggests an average salary of $60K, and that’s before the “hefty” co-location fees. Surely the average HFT guy is paid better than a median Chicago teacher.)


  6. jeremiah757
    October 1, 2012 at 11:43 am

    As a matter of fact, the exchanges collect rents from the HFTs to colocate their servers within fiber-cable distance of the exchanges’ own servers. There is strong evidence that HFTs use the speed advantage to engage in unlawful manipulation. Hunsader is pretty much the authority http://www.nanex.net/Research/QuoteStuffingBanned/QuoteStuffingBanned.html

    He has also found that the exchanges put their premium customers on special networks that send them the quotes earlier. Why hasn’t the SEC cracked down? Hunsader’s expression – bought and paid for. Check him out.


    • Jonathan
      October 1, 2012 at 5:31 pm

      Thanks for the reference. But their subtitle undermines their credibility.

      “Rule change proposed by the NYSE marks the beginning of the end of quote manipulation”.

      Do they really think quote manipulation is going to end?


  7. Kaleberg
    October 1, 2012 at 2:16 pm

    The problem with HFT is that it is simply a wire scam. Any market where one party gets advanced information, even if it is only a microsecond in advance, is simply running an old fashioned wire scam. Everyone knows this, and it erodes trust in the markets as a pricing mechanism. Wire scams are popular in revenge fantasy movies like The Sting or Queen of Hearts, because they provide a plausible mechanism for duping an otherwise sophisticated investor, but we shouldn’t make them central to the operation of our capital markets.

    There is also the issue of what it means legally to make an offer to buy or sell which cannot be accepted by another party. Most of the traffic on HFT exchanges consists of bids which are of extremely limited duration for the purpose of making them impossible to accept. If Walmart put an item on sale 100 feet from the cashier’s but it was only valid for 50 nanoseconds (50 light feet), there would be a serious outcry about the fraudulent nature of the sale. Once again, fraud is not something we want at the heart of the marketplace.

    Despite the theory that there is no reasonable way to deal with these problems, we already know of several. There is no reason that the market cannot place a minimum duration on bids to prevent fraudulent offers. There is no reason there can’t be algorithmic specialist agents who are given an income stream in return for maintaining an orderly bidding process. These, and a number of other methods, have a proven track record in maintaining credible markets. I have nothing against private algorithms or deception, but a system based on simple fraud does not inspire confidence.

    P.S. Given the vast and increasingly vast inefficiency of our capital markets, it might be time for some restructuring.


  8. None
    October 1, 2012 at 3:43 pm

    If you don’t like the price, don’t make the trade. Every share i’ve ever bought I thought was a bargain.

    The HFT algorithms increase liquidity, that’s a fact i’m afraid, and that means YOU get to buy cheaper and sell for more than you’d otherwise have had to at that point in time. That they allegedly make 80 billion a year is of no consequence, they have done it offering to buy or sell at a higher or lower price than anyone else at that point in time, ie they have saved the other party money.

    It would be interesting to see how much they have saved the counterparties, that may do something about the continued whining about them.

    Anway, these HFT things remind me very much of LTCM. Companies running a bunch of algorithms which make a bunch of money until one of their assumptions goes wildly wrong. If they temporarily cause large drops in the market thats a buying opportunity and if they cause a spike that’s a selling opportunity. As long as I do not have to help bail them out when they go bust…


    • jeremiah757
      October 1, 2012 at 4:28 pm

      Enhanced liquidity is one of the myths HFTs use to justify their scam. Research has disproved it. See UI study at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2066839

      To quote the study – no improved liquidity, efficiency, or social benefit. Just quote stuffing and congestion.


      • HFTer
        October 1, 2012 at 5:40 pm

        Did you even read the study?
        It only claims that Nasdaq’s nanosecond speed upgrade in April-May led to more quoting, but not more trading, liquidity, etc.
        It does NOT claim that HFT have no social benefit. In fact, they explicitly acknowledge that “the increase in trading speed from second to millisecond may
        still have social benefit by creating new trading opportunities”


    • Jonathan
      October 1, 2012 at 5:27 pm

      The system as a whole has no liquidity. Except for IPOs, etc. there are just as many shares of Apple outstanding today as there were yesterday. Only the price and owners have changed. Some people have more shares and less cash. Other people have fewer shares and more cash.

      This is not to say that any single investor’s ability to buy or sell is irrelevant.

      HFT provides liquidity second-by-second but not day-to-day or even hour-by-hour since they look to reverse their trades very quickly.

      Today’s buyers (those holding more shares today than they did yesterday) paid more for those shares than today’s sellers because the intermediaries extract a price. This is not necessarily true every day (because the intermediaries take some risk) but it must be true on average.

      Of course, this was true 50 years ago before there were HFT traders (at least not computerized ones). One question is whether the amount they extract is larger than the previous regime of market specialists extracted. Also important is whether that system was more or less stable and more or less fair. Also, most relevant is whether a proposed alternative would be better or worse on these dimensions.

      I don’t have an answer to any of these questions.


      • Auntie Nomen
        October 2, 2012 at 9:01 pm

        The premium market makers charge per share is basically the spread. Spreads in US equities used to be over $0.10/share. Now they’re typically 3-10x smaller. This is a direct result of the huge increase in competition amongst market makers & liquidity providers, which is in turn a direct result of the introduction of high-speed computerized trading. These days, a few programmers in Jersey City can compete with Goldman-Sachs’ equity market making operation.

        Frankly, I don’t think HFT deserves anything resembling the amount of attention it gets. The 2008 financial crisis was caused by humans making huge dumb bets with instruments like CDOs and CDSs that do not trade at high speed on electronic exchanges. We are still suffering economically because those bets are still unwinding. You can fix every single problem HFT might conceivably cause, and it will make essentially zero difference to our economic woes.

        To make matters worse, most of the proposed fixes would actually increase the amount Mom & Pop pay to market middle men: Most of them are designed to drive high frequency market makers out of business, by raising the cost of doing business. (HFT players make a fraction of a penny per share traded. A $.01 tax amounts to a few hundred percent tax on profits.) When the little guys are gone, the big guys are going to use their oligopoly to extract rents, exactly as they did in the old days.


    • Harry
      October 2, 2012 at 12:45 pm

      What a rubbish argument. Ill give you a better one. If the game is rigged, dont play it. Thats my attitude to stock market speculation now. Which if it were the general attitude of the public would eventually result in the market becoming ever smaller and of less practical use for capital raising.

      And it is not a fact that HFT algorithms increase liquidity. Prove it. Indeed you might want to think about defining liquidity before you make that point. Increasing turnover in good markets and widening spreads in bad markets is not a meaningful improvement in liquidity. In fact a better definitition of liquidity might include some assessment of the total costs of trading. Having a group take 20bn out of transactions is not liquidity. Its a tax.

      Finally “front running” is a crime. So tell me why computerised front running should not be a crime? Packet sniffing type algos strike me as clearly illegal.

      In the past the axiom was that all participants should have equal access to information. Why should that no longer be the case?

      Your comment seems to me to be poor logic unemcumbered by any facts.


  9. Jonathan
    October 1, 2012 at 4:52 pm

    I think the issues of stability, fairness and trust are more important than cost.

    Does HFT increase or decrease market stability? It seems to me that it may decrease volatility much of the time but increase the frequency of extreme moves that are not related to news. A sharp move in reaction to a big news event is perfectly reasonable but if it occurs in the absence of news is disturbing. The issue is not so much the risk of one algorithm going haywire but the interaction of multiple algorithms faced with unusual events. It creates a system that is impossible to understand, even by the participants.

    The second issue is fairness. are some investors taken advantage of in the current system?

    The third issue is whether it increases or decreases investor trust in the system.

    These are obviously related. Sharp market moves for unknown reasons will erode investor confidence. So will visible unfairness. But they are not the same thing. Opacity may erode confidence even if it does not increase inequity.

    I don’t claim to have a great answer to this. I don’t even claim to understand the issues very well. But that is not just my problem. I am an investment professional (but not involved in HFT at all). I made a serious effort to understand the strategies and issues involved. Ultimately, I came to the conclusion that I couldn’t understand it well without access to insiders Which means that regulators can’t either.

    Of course, no system is perfect. And any system will extract a price for transactions. But I do think that there are changes that would be more stable, fair and transparent. I’ll post some ideas later.


  10. Jonathan
    October 1, 2012 at 5:41 pm

    Entertaining, perhaps enlightening, video on the topic


  11. October 1, 2012 at 7:48 pm

    Chris Sparrow’s blog post “The Failure of Continuous Markets”:



  12. October 1, 2012 at 8:23 pm

    The whole argument of the benefits of liquidity strikes me as having very weak foundations. Two counterarguments:

    1.) Per Nanex, a recent NY Times article included a chart showing the collapse in trading spreads over the past decade or so, which was presented as a benefit of HFT. As Nanex noted, however, the bulk of the collapse occurred prior to 2006, which means that a non-trivial portion of it was due to electronic trading more broadly. So it’s a bit misleading to claim that HFT was the primary driver of these savings.

    2.) As Cathy noted above, liquidity is not a constant. The new DMMs (digital market makers, like Knight Capital) are for some reason not subjected to the same requirements that applied to traditional market makers, who were required to maintain two-sided markets for each stock they covered. Without that limitation, DMMs are free to disappear from the market when it most needs liquidity, as happened during the Flash Crash. McKenzie’s piece in the LRB is particularly good on this:

    If the only thing that HFT has to offer is liquidity when they feel like it, that’s a pretty sad tradeoff for the creation of a new systemic risk.


    • None
      October 2, 2012 at 12:16 pm

      “If the only thing that HFT has to offer is liquidity when they feel like it, that’s a pretty sad tradeoff for the creation of a new systemic risk.”

      What is this “systematic risk” of which you speak ?
      Some guys following an erroneous algorithm, that generally makes prices better for everyone but that can occasionally cause a temporary market drop/spike if they have enough leverage and their algo goes haywire ?
      Thats not a systematic risk for anyone other than the guys with the algorithms, or the ones lending them money for their leverage. It’s a buying/selling opportunity of major proportion for everyone else.

      Your point 2) is very weak. The requirement to maintain a two sided market is meaningless, it just means as a market maker you set your spreads so wide that noone trades without the price swinging even more dramatically in one direction so as far as a guarantee of providing liquidity or stability when needed, its even worse than useless. Secondly, DMMs didnt disappear from the market during the flash crash, at least one of them was actually causing it (again to the detríment of the people running the algorithms, and to the enrichment of those who were buying while it was selling).


  13. Bindicap
    October 1, 2012 at 8:26 pm

    Your “How it happened” makes a good point that bringing latest technology to bear on a problem is natural and inevitable. I think it’s missing the dismantling of the old specialist system and the deliberate structure of the National Market System and best bid/ask rules. Those are a big driver for the high frequency market making. I don’t understand these very well myself, so I’d be interested to see some good discussion. Craig Pirrong has some useful articles.

    Your “What’s wrong” is hobbled by bringing up only examples where HFT didn’t actually do anything wrong. You do kind of acknowledge this with “two out of three times were relatively benign”, but the third non-benign case wasn’t HFT, as you basically say yourself…. It’s odd.

    And people already pointed out your profit/investment figures are magnitudes off. Whatever insider fed you those figures is leading you way astray. It should have been obvious….

    “What should be done” is on pretty shaky footing given this. The minimum order time is actually kind of interesting as a requirement. It forces a minimum amount of liquidity provision on participants, and is effectively a small tax. It would push participants out faster as volatility (and the effective tax) increases. I think there are a lot of practical issues, but it will be interesting to watch the German experience. I don’t know if they have the same order structure and multiple venues as US equity markets (which is a driver of the situation now). You are very dismissive of this, though. But you think a transaction tax is sure-fire, even though it seems far more fraught.

    I don’t know what to think of all the discussions like this in the media. There was a period were it was very clearly driven by old school market makers trying to fight for turf through the media. There is also resentment by very large funds that the market structure favoring small investors either requires them to have very clever buy/sell tech themselves, or go through dark pools, or live with the market reacting to their signals. They might have valid points, but it comes across as a marketing campaign long on innuendo and without much regard for solid info.


  14. Jonathan
    October 2, 2012 at 10:08 am

    Why are we so addicted to speed?

    Germany has proposed that orders need to persist for 1/2 second. That is a step in the right direction. But it still wouldn’t let humans back in the market. 500 milliseconds is a long time to my computer, but faster than I can click a mouse. Why not 10 minutes?

    I can, somewhat, understand the objection to having an order forced to stay in the market for 10 minutes, but there are other ways to slow things down. There could be market clearing auctions periodically. “Quantized markets” as described in an article I reference below.. Having the market pause for 10 minutes seems radical, but we live without liquidity for something like 60 hours every weekend. What is the harm of forcing people to wait 10 minutes to trade? They lose the opportunity to dump their stock on the less-quickly-informed investors? That is not a harm. This is a serious question. If anyone thinks there is a significant problem with 10-minute market pauses, I’d like to hear it.

    The current structure, and even the German proposal, means that institutional trading must be mediated by computers. .Call me a luddite, but it seems to me that it would be a good thing if we had a market humans to participate in directly if they wanted to..

    So, a tax might work. But I think a market that traded at discrete times would be better. I raised this at the OWS Alt-B meeting on Sunday and someone who is better informed than I am suggested that there were problems with this and I hadn’t thought it through fully. He was right that I don’t understand all the technical aspects. So, I am delighted to see that David L. has posted this link to Charles Sparrow’s “The Failure of Continuous Markets” from the Tabb Forum and HFT Review.

    So, maybe there are flaws in the proposal but I’m not alone. Now, admittedly, Sparrow proposes 1-second intervals. But this is odd because he says “We want to assume that all market participants can react to information and implement their responses before a given time”. One second surely does not accomplish that.

    Now, 10-minutes is probably too long. I’d be delighted with 1 minute. I raise 10 to make a point. People don’t really need to “implement their responses” within the interval. If they need more time they can withdraw from trading. So, 1-minute would be ok with me, too.


    • None
      October 2, 2012 at 12:28 pm

      “But it still wouldn’t let humans back in the market.”

      Comments like that really make me scratch my head in wonder “Do these people know anything about the market ?”.

      HFT’s IMPROVE prices for humans (and everyone else). They offer to buy at more and sell for less than others. If it was any other way their orders would not get filled.


      • Jonathan
        October 2, 2012 at 12:47 pm

        Yes, I know a little about how markets work. I’ve been an investor, professionally and personally, for 31 years.

        If you are trading 1000 shares, you are right. If you are trading 1,000,000 you are wrong. HFTs look for patterns in orders that suggest a large institution is buying or selling. Then they front-run them. This increases the institutions execution costs. To avoid this, institutions use algorithms to try to try to disguise their trading. Of course, this has a cost, too, but a smaller one. All institutions execute their orders via computer algorithms. If they didn’t their execution costs would be higher because the HFTs would sniff them out and take them to the cleaners.

        So, the trigger to the May 6th flash crash was a large institutional order. It was not an HFT but it was executed into a computer system. Apparently it had some flaws. But no complex algorithm will be perfect.


        • Harry
          October 2, 2012 at 2:17 pm

          The argument is wrong for even 1000 shares.

          Consider an investor who seeks to sell 1000 shares at market. A packet sniffer works out that there is a large institutional seller. There is an advantage in selling before that seller trades. Arguing that any buyer is advantaged ignore the other retail sellers who were disadvantaged by the use of asymmetric information.


        • Jonathan
          October 2, 2012 at 2:30 pm

          I agree there are various subtle ways that retail is exploited. But it seemed too complicated to get into (so complicated that it would take time to work out fully myself).

          In addition, Maybe I’m being naive but it seems to me that retail traders on balance are clearly better off than they used to be — though not so much because of HFTs.

          Not so clear for institutions, which includes most individual investors via mutual funds, pension funds, etc.


        • Harry
          October 2, 2012 at 3:27 pm

          You mean because spreads and commissions have narrowed? Very true, but I dont see why a large proportion of that gain should be appropriated by computerised illegality. If the information is equally available then fine. But it seems to me that information is not equally available and the exchanges are selling trading advantage to the highest bidder. Thats a rigged game.


        • Jonathan
          October 2, 2012 at 3:53 pm

          I agree.

          I don’t mean to defend the current system. Just note that ripping of retail traders is not high on its list of sins.


        • None
          October 2, 2012 at 4:01 pm

          Have you ever actually bought any shares ?
          You ask for a quote, and get one with a time limit, then place your order if you are happy with the quoted price and dont place it if you are not happy with the quoted price. The best price may not have been offered by an HFT, but if it was it was offering a better price for YOU for that trade than anyone else. In what way is that bad for YOU ?


        • None
          October 2, 2012 at 3:55 pm

          Apologies for the quip then.

          “If you are trading 1000 shares, you are right. If you are trading 1,000,000 you are wrong”

          Your original comment was about letting humans back in the market. You seem to accept that on the smaller scale it’s better for humans (which makes your original comment a little odd) but now the complaint is that it prevents large movers from sneaking their large positions into and out of the market.

          As for being wrong when trading 1,000,000 I disagree anyway. Noone is being “taken to the cleaners”. In theory HFT shave fractions of a cent off prices and are closed positions at the end of the day. Noone gets taken to the cleaners, they make their money by moving the price where they think it’s going fractionally before it gets there and doing it extremely frequently. So far, the only noticeable price fluctuations have been when something goes wrong, with the loss making party being the HFT owner not the third parties who took advantage of the bizarre prices.


        • Bindicap
          October 2, 2012 at 7:29 pm

          Thanks — you’ve been really active in this thread, and I could not get where you were coming from before.

          I think the US markets are more friendly towards retail investors by design. We get the tightest spreads and lowest commissions ever from venue competition and order rules.

          But the same forces jumping in to serve retail flow are watching very carefully not to get burned by large, informed investors who will make their books wrongly balanced while the market runs against them.

          If you are a large institution just balancing your allocation per formula or processing redemptions, it seems unfair because maybe in the past you could toss in some large orders and get filled with small slippage. But in the super competitive, anonymous market we have now, aggressive market makers deduce this activity from order book history and step back our of fear you are Galleon or Einhorn teeing up.

          So large institutions need their own smart trading tech. Or dark pools. Or suck it up.

          Or they can argue that this market structure is not good and should be changed. But I’d like to see the argument made openly. I’ve gotten jaded by reading posts and discussions like this that don’t seem to do that.


        • Jonathan
          October 5, 2012 at 1:07 pm


          Sorry, I don’t have time to really to reply.

          Essentially, I would say it isn’t as benign as HFTs “step back from fear” as much as “attempt to exploit.” Not that I’m arguing Galleon needs protection, but others might.

          This might be helpful


    • Jonathan
      October 2, 2012 at 4:45 pm

      I want to clarify what I mean by “bringing humans back into the market.

      By “letting humans back into the process” I wasn’t thinking of retail investors (who are, of course human) but meant to allow humans to be more directly involved in trading for institutions, such as mutual funds, pension funds, etc. They constitute the bulk of the “long-term” trading (by long-term here I mean anyone with a horizon longer than one day).

      Right now, those institutions all trade via computer algorithms because to do otherwise would be quite costly.

      Now that I write that, I realize it isn’t immediately obvious why it would be a good thing for humans to be more directly involved. And, in any case, it is likely that many, perhaps almost all, institutions will use computer systems even in the slow markets I am advocating. I do think there is still a benefit. A system with a mixture of humans and machines will diversify its flaws. As Lou Puls points about below the algorithms are likely to become quite similar if they aren’t already (in his words “aping the same fallacious assumptions, techniques and choices.”). This is because the HFT community is not that big and undoubtedly people will move from one firm to another so techniques will get shared. A homogeneous system is less stable. Also, the focus on computer trading will be more on intelligence and less on pure speed as it is today. Finally, there will be less of an unproductive arms race where all of these firms pay the exchanges for co-location and pay the computer manufacturers for the latest chips just to keep up with everyone else who is doing the same thing.


  15. tw
    October 2, 2012 at 10:57 am

    The HFT conundrum is problematic because most of the time there is no real definition of what it is. There are arb, stat arb, etf, rebaters, market making, etc and each one has a different strategy. 

    If you look at Peter Bernstiens: Capital Ideas Evolving, he talks with the head of Goldmans HFT desk. He noted that they need volume, technology and cheap money. In an era of free money, expanding technology, this series of strategies is going to rule no matter what.

    There will be many different solutions coming forward to deal with this new advancementhe it the big funds need to be part of the solution, as is the case in Australia. Set times for bids, holding periods etc don’t address some of the underlying issues with the general market productively.

    For example, we don’t know how much HFT is due to eft movement. Etf’s ironically are supposed to mimic the underlying, when in fact they are more influential and may be open for manipulation. Because of extensive computer activity ai, dark pools, internal trading and multiple trading pools have been adopted to gain control and advantage. Rebates from exchanges to generate bids and offers have opened the door for “spoofing” and manipulation to drive trading to passive orders to generate rebates on other venues. The strategies seem to rely heavily on the fact that humans can’t keep up and regulators will never be able to make a credible case due to complexity, speed and volume data. In other words, they own the market.

    The speed of movement means that often trades occur before quotes actually appear. It also means that as orders are broken down into smaller increments, that 2000 share bid you see will be cancelled after 500 is filled, and the other 15- 100 share bids are removed….or false advertising.

    It is ironic that computers are used to act in a fashion that human traders can still be penalized for, but due to the speed and volume of information, it is impossible for regulators to follow them, while it is easy to phone the trader, registered with your organization, whose volume of transactions is small.

    There are many problems, but a few potential solutions are available.

    1) many liquidity pools, but one central market place.
    All trades should be done in one location, with liquidity pools competing for business to trade on the exchange. 

    2) end of for profit exchange model.
    Eliminating the for profit model and contracting out management of the exchange with a mandate for innovation and “level playing field”

    3) dark pools and internal trading banned
    These are opaque non transparent options that result from HFT domination and will no longer be necessary.

    4) all traders doing above a certain volume on the exchanges must be registered as pros with the appropriate regulators: this would bring current computer traders into the formal regulatory structure

    5) all exchange traffic will be put through the regulators black box, before it hits the exchange. Fees will be levied based on traffic, with those generating the most traffic paying the highest fees. This will allow regulators access to the information involved, and with appropriate funding they can mine data in a more timely manner to figure out what happened, when and by whom, using the same software as the big funds.

    6) rebates banned for any non registered individuals, and for those charged with infractions.

    7) for every ten cancellations, there must be one transaction, or a fee is charged and no rebate will apply (ratio of transactions to cancellations as proposed in Germany)

    8) leveraged efts: banned

    9) any efts that do not meet a certain volume/activity threshold in a quarter, or three consecutive months, are to be delisted.

    10) any firm found to be in gross violation of securities laws to have their regulatory certification pulled for one year plus a fine. 

    11) sub penny trading eliminated

    I am not in favor of a Tobin tax. This is counter productive and allows for and encourages regulatory arbitrage. I am not certain that time limits on “orders” will necessarily be effective either, but adjusting the rebate rules and having a ratio of transactions to cancellations will automatically balance out the market…..or maybe it won’t.

    I would also suggest that the rise of HFT has occurred in an environment of historically low interest rates. What will their influence be when these adjust or if they are “normalized”? I think this question is an important one as well, since these strategies require cheap money as a significant part of their operations.




  16. October 2, 2012 at 12:06 pm

    It seems too sweeping to generalize the main problem of HFT to be algorithmic instability, but it’s a good candidate, given that whole books are written on the vagaries of computational numerical instability.

    Modern HFT is largely based on standard solution techniques borrowed from the partial differential equations of fluid dynamics. When one deals with a multidimensional state space where infinitesimal variation is necessarily estimated by a CHOICE of appropriate finite differences for the immediate conditions, it is critical that the choice does not itself create instability.

    When the frequency of trading is the prime determinant of such choices, rather than the state space conditions, good luck trying to avoid greater probability of horrendous instabilities – they become little Black Swans in ever increasing hordes of flocks overwhelming any attempt at representation of reality.

    Any claims of liquidity advantages are an illusion akin to the inevitable increase of entropy as energy becomes more and more unavailable: such EFT-derived liquidity is unavailable liquidity, and the entropy rooted in exponential instability can only be highly destructive. Only Tobin taxation seems capable of regulating such structural illusion.


    • Jonathan
      October 2, 2012 at 1:24 pm

      My concern is not a single algorithm becoming unstable. I don’t care if an HFT blows themselves up.

      My concern is about interaction among multiple algorithms, the fragmented market structure, the inconsistencies of rules between venues and between stock index futures and stocks, actions taken by market participants, exchanges or regulators as the crisis begins.

      This type of “system accident” is discussed in a different context in the book “Normal Accidents” by Charles Perrow. The key conditions are “tight coupling” and complexity. This seems to describe the market very well. Such accidents are still rare but when they occur catastrophic and hard to control.


  17. October 2, 2012 at 1:36 pm

    Interaction between multiple faulty algorithms becomes far more systemic and probable, especially when the algorithms are aping the same fallacious assumptions, techniques and choices.

    Indeed these are relatively rare and highly consequential, Taleb’s definition of a Black Swan.


  18. October 2, 2012 at 3:47 pm

    Changing the rules is clearly the way forward, with the German example a good first step. The trick is to change the rules so they reward volume and the length of time a quote is held open. Section 9 of the paper ‘Pricing, liquidity and the control of dynamic systems in finance and economics’ discusses such a system in detail:



    • None
      October 2, 2012 at 4:54 pm

      “The trick is to change the rules so they reward volume and the length of time a quote is held open.”

      Reward in what way ? Government subsidies for quotes open 3 months or longer ? I think its better we reward the cheapest quote by closing the deal, if we want it.

      Extracts from the paper:
      “This paper does not however require belief in the modelling of YMTU, it simply requires a
      disbelief in the efficient market hypothesis.”

      Yes the market is very inefficient that is why those academics make so much money taking advantage of its inefficiencies.

      “But the reduced spreads have been accompanied by increased frequencies of

      Amazing, it’s almost as if the latter caused the former.

      “It is trivially obvious that if spreads are halved, but traders are forced to trade three times more frequently, then overall trading costs have been increased by 50%.”

      It’s trivially obvious that the halving of spread does not force a trader to trade more often, it just makes it cheaper when they do trade. The implication that offering a cheaper price to the trader for a trade is actually costing him money is bizarre.


  19. Jonathan
  20. October 8, 2012 at 5:24 pm

    In the devious world of HFT, here’s a sneaky trick to get traders to reveal their positions, setting up for further strateegery:



  21. papicek
    October 9, 2012 at 10:02 am

    Make every trader enter a capcha every time they want to place a trade. Problem solved.


  22. papicek
    October 9, 2012 at 10:51 am

    I might also share this little snippet on individual investors bailing on stocks:



  23. November 16, 2012 at 12:43 pm

    people on the inside, if you point out these events, might counter that “normal people didn’t lose money” at these events. The weirdness was mostly fixed after the fact, and anyway pension funds, which is where most normal people’s money lives, don’t ever trade in the thin opening bell market.

    But there’s another, less well known example from September 30th, 2008, when the House rejected the bailout, shorting stocks were illegal, and the Dow dropped 778 points. The prices as such common big-ticket stocks such as Google plummeted and, in this case, pension funds lost big money. It’s true that some transactions were later nulled, but not all of them.

    Was that an HFT-related crash or just a crash?


  24. November 16, 2012 at 12:53 pm

    Putting aside the actual energy consumed by the industry, which is certainly non-trivial, let’s focus for a moment on money. It has been estimated that overall, HFT firms post about $80 billion in profits yearly, and that they make on the order of 10% profit on their technology investments. That would mean that there’s in the order of $800 billion being invested in HFT each year. Even if we highball the return at 25%, we still have more than $300 billion invested in this stuff.

    And to what end?

    Is that how much it’s really worth the small investor to have decreased bid-ask spreads when they go long Apple because they think the new iPhone will sell?

    Quoting from http://www.ft.com/cms/s/0/2de2b29a-9271-11de-b63b-00144feabdc0.html:

    In the early 1950s, the “finance” sector (not counting insurance and real estate) accounted for 3 per cent of all US wages and salaries; in the current decade that share is 7 per cent.

    So the US puts 7% of its human resources toward finance. Of course not all of this is in markets where HFT penetrates. But I would guess the social benefit of lower spreads doesn’t accrue to retail spec traders, rather the cost of the system as a whole is reduced. Like a better oiled machine: less friction.


  1. October 4, 2012 at 4:58 pm
  2. October 9, 2012 at 8:53 pm
  3. April 9, 2013 at 3:30 am
Comments are closed.
%d bloggers like this: