A modest proposal to remove some of the more ridiculous waste (and some corruption) from our financial markets.

Given the massive ignorance, not just of you’re average Joe (Sorry I think that’s now ‘Joe Six-pack’) but of experts, I think we should be particularly on the lookout for ‘no-brainer’ reforms.  Simple things that we can do than generate gains and for which it’s very difficult to imagine substantial costs.  For a long time I’ve thought that so much resources in the financial sector go into trying to get the scoop on others – trying to beat other investors to the punch to buy or sell assets before the news spreads to others that it’s a major source of inefficiency.  If so there’s something that one could do fairly simply.  One could only trade every now and again, say for a couple of weeks every three months on a stock exchange and require all listing firms to have a high degree of information out in the marketplace before markets opened.

A bit radical? Well yes, I wouldn’t try the experiment in quite such a bold way. I expect such a change would generate far more gains than losses, but I can’t prove it.  And it’s undoubtedly the case that markets being open 24/7 enable various risk management strategies that may be important for the real economy.  But one might move in this direction more incrementally in two ways that I think would produce far more gains than losses.

  1. Where there’s evidence of insider trading – for instance where there are major and unexplained movements in the share-price in directions which are understood after some company announcement, a firm should be put onto a regime in which their shares are not traded in some substantial period before any major announcement that can be anticipated – like results.
  2. The race for priority has escalated to such a ridiculous degree that a huge number of hedge funds and trading houses have moved important IT facilities as physically close to exchanges as possible.  The incentives driving this wasteful race for priority could be dealt with, with minimal downsides it seems to me with staggered randomised priority.  Buyers and sellers would send in orders as they do now.  But they would go into one minute batches.  Thus whether your order arrived in the first or the last second of a one minute segment of time, it would go into the pool of buyers and sellers in a random order for that minute (one might delay it by a minute also just so there are no remaining benefits from relocating your IT to slightly improve your chances every now and then of ‘making the cut’ into an earlier minute block.) Thus every minute the system would spew out matches and generate pricing information as it does now.  So you get rid of some rising costs, and it’s hard to believe that you wouldn’t have all the benefits of trading that we enjoy today.

Postscript: It is in the interests of the exchanges themselves to bring about PSR (That’s Periodic Staggered Randomisation for the uninitiated) as explained in this subsequent post.

    This entry was posted in Economics and public policy. Bookmark the permalink.
    Subscribe
    Notify of
    guest
    24 Comments
    Oldest
    Newest Most Voted
    Inline Feedbacks
    View all comments
    Ronan
    Ronan
    11 years ago

    One of the principles that underpinned some of the old stock and futures exchanges was that of ‘one market’. Many exchanges had, and many still have rules that govern where it’s traders could do business. It was not fair if, for instance, in a wheat commodity pit that some traders would do business away from the floor and not be subject to its rules nor reporting requirements. To force the participants to engage in a trading pit and not at the local cafe ensured that opportunity was available for all to participate in the price discovery process, and a wider group of opinion is meant to produce fairer pricing (herding instincts aside).

    In this age of elecronic trading, while (most) orders might go through a central matching system, for some traders to make decisions at the speed of light while others have a delay of several seconds to minutes is unfair to market partcipants that simply cannot compete on these speeds. It creates an asymetric advantage to the participants with the most resources to re-locate their servers to the buildings next to exchanges. Co-location seems to destroy any concept of a level-playing field and ‘one-market’.

    The other advantage a 1 minute auction system would bring is cooler heads in the event of crashes and mini-spikes. The most recent ‘flash-crash’ in the States may not have been fully averted but at least time would have been available for participants to make more reasoned decisions – and not sell $20 P&G stock for 10c. And with more and more automated trading, these sudden liquidity spikes/holes will probably become more and more prevalent.

    Great idea of 1 minute batches.

    Bill

    Dave
    Dave
    11 years ago

    Can you explain why you think that financial markets are inefficient? Competitive markets generally promote efficiency, unless there is some source of market failure such as externalities or public goods. What do you think would cause market failure in this case?

    SJ
    SJ
    11 years ago

    Dave, here’s a pointer in the right direction.

    SJ
    SJ
    11 years ago

    Actually, the previous day’s one was probably better.

    Dave
    Dave
    11 years ago

    Nicholas,

    thanks for your response. I’ll take that as a “no”. If it is not possible to explain the source of market failure in a few sentences, I would suggest it is not a “no brainer”.

    SJ

    thanks for the pointers, but these address a rather different point: that HFT creates instability, rather than simply that it is a waste of resources.

    Dave
    Dave
    11 years ago

    Trading firms invest in high-frequency trading (HFT) infrastructure because it is profitable to do so. In the short-term, trading is a zero sum game, so the HFT profits are coming from other traders with inferior or no HFT resources. Why are these other “non-HFT” traders paying the HFT firms in this way?

    I think there are 3 possible answers:

    (1) The non-HFT firms benefit from the HFT service. That is, the HFT firms are providing a valued service which the non-HFT firms are voluntarily paying for.

    (2) The non-HFT firms are behaving irrationally: eg they are competing in the HFT “race” but keep losing. Should we regulate to protect these firms from themselves? Or is there some societal damage associated with this irrationality? (Aside from the point about market instability, which is a different issue.)

    (3) The non-HFT firms are being forced to pay for the HFT service, even though they don’t want it. This would obviously be a flaw in the market design. But what causes the flaw? Is it in the electronic market clearing infrastructure? Is it in the information disclosure regime? Why aren’t the non-HFT firms complaining about the flaws and asking the market operator or regulator to fix them?

    Understanding these issues looks pretty subtle and complex to me. Perhaps it is a “no brainer” only to those who are not curious about market design.

    Tel
    Tel
    11 years ago

    Why make the proposal modest?

    I suggest only one round of clearance every day: the bids sit in queue until at a randomly determined moment somewhere during business hours we have an instantaneous event and all bids that can execute do so, then it’s back to queuing until the next day.

    No one can possibly believe that buying and selling in the space of a few hours represents a real systemic investment in the overall economy, so anything that didn’t get done today can safely wait until tomorrow. Genuine long term investors would have no problem with this, and any exchange with a higher proportion of genuine investors is a more attractive exchange for business to float on because their share price will be more stable and their share holders more level headed.

    Also, on an unrelated note, it looks like computer scientists are on the cusp of mathematically disproving the efficient market hypothesis.

    http://www.freedom-to-tinker.com/blog/appel/intractability-financial-derivatives

    The Princeton guys claim that all derivatives trading is subject to the NP problem because sneaky correlations built into the instrument in such a way that the buyer cannot feasibly calculate a price. They use it to explain a problem with CDO’s but my feeling is that a bigger problem with CDO’s was that the buyer wasn’t even given the information to begin with.

    Warfare based on complexity escalation is nothing new, consider the pricing of mobile phone calls for example.

    Last year there was a paper that said the the EMH is equivalent to saying P=NP

    My feeling is that it isn’t even as difficult as this. You are suggesting that market participants have knowledge of not only the local region around the present state, but knowledge of the entire possible state space and all connectivity therein. If it can be proven that EMH is computationally unfeasible even under this condition of infinitely extensive knowledge then that is interesting from a theoretical point of view, however, in a practical sense it’s irrelevant.

    I argue that the best any civilization can hope for is knowledge of the local region around the present state (plus some historic data) implying that the system as a whole must be a gradient follower (give or take a few gifted individuals who see a fraction farther than the others do) and gradient followers perpetually need to deal with the issue of a relative minima in a non-linear world. Don’t worry about NP because we aren’t even close to knowing the problem itself.

    Of course, once EMH has been laid to rest once and for all, the Bureaucrats will jump up and tell you they can do it all better (that’s what they always say) and you will have exactly the same problem of limited knowledge but with fewer brains thinking on the issue and exploring a vastly smaller region around present state. Gone from bad to worse.

    Tel
    Tel
    11 years ago

    But the more trading you have instantaneously clear, the more you create incentives to watch the pot and change bids before the system does the trading, which is time consuming and may encourage gaming.

    The randomized trigger time over an 8 hour window during the day should discourage most of the gaming. With a “fair split” clearing algorithm, the incentive for tweaking bids is low (the algorithm picks the single price that maximizes the volume of trade, then offers that price to all). I do see people put zany bids into the overnight queues on the ASX for other punters to ponder over; then yank them out at the last minute and replace them with a sensible bid. I’m not sure how effective the tactic is but someone must believe in it. The ASX already randomizes start of trade, but only over a narrow window (10 minutes or something).

    What additional benefits does your rule confer over mine?

    I agree that investors who do nothing other than search for what Jacques calls microarbitrage are weighing the system down, and discourage genuine investors.

    My proposal puts casual investors (who have a real job) on an equal footing with people who have all day to stare at tickers. These casual investors bring more information from more sources into the marketplace and they make better shareholders because they buy based on intrinsics rather than the dynamics of the hour. These are the type of shareholders that firms should be looking for.

    In other words, actively making life difficult for short-term day traders makes the exchange better for everyone else. In terms of the economy as a whole, price discovery drawn out from a day to a few weeks isn’t going to be a problem (and drawing it out most likely will improve price stability). When I look at the way day prices jitter around, I have great difficulty believing they represent any physical basis in the real world.

    Dave
    Dave
    11 years ago

    Nicholas,

    I’m not sure from your comments whether you understand arbitrage. It involves buying one product and selling another related-product (or the same product, in a different market) close to simultaneously, so as to profit from the price difference.

    If markets only clear once per minute, you have a minute to wait between discovering that you have bought and then being able to sell to complete the arbitrage. Since prices can move in this time, “micro” arbitrage becomes impractical. It is not just a matter of arbitrage being delayed by one minute.

    Your bid price on one product must reflect the market price of the other product, so your concept of having some static preference (in post 14) does not apply. The textbook has moved on.

    In the terms of the P=NP discussion, one can think of each market clearing as one iteration in solving a complex optimisation problem. So, slowing down the rate of clearing may lengthen the “solution” time and prevent the market from finding its equilibrium. Or, put another way, “market clearing” really means clearing bids and offers across multiple, related products. It is not feasible to do this in a single iteration. Micro-arbitrageurs are part of this complex clearing process, which requires many steps to complete.

    dorinny
    11 years ago

    a huge number of hedge funds and trading houses have moved important IT facilities as physically close to exchanges as possible.

    I only learned about this last semester, and its pretty extraodrinary isn’t it? That real estate can be so inflated due to split second differences in data transfers. I’m not sure that a randomised priority is the solution – couldnt they just find a way to make the data transfers faster, regardless of where they are coming from?

    Where there’s evidence of insider trading – for instance where there are major and unexplained movements in the share-price in directions which are understood after some company announcement, a firm should be put onto a regime in which their shares are not traded in some substantial period before any major announcement that can be anticipated – like results

    How would this work? Don’t the results have to be released before the insider trading can be realised? And why penalise all shareholders of a company due to the opportunistic (and illegal) action of one party? I think a more realistic solution to the problem of insider trading is to make information even more transparent for listed companies – the faster ‘inside’ information can become available to the public, the better.

    If you’re prepared to sell at a given price, you let the market know and see if anyone is prepared to relieve you of your assets at the price you’ve stated you’d like to get rid of them.

    Yes, but the price is not necessarily determined by what you are prepared to sell it at. Have you ever tried selling a house?

    Simple things that we can do than generate gains and for which it’s very difficult to imagine substantial costs.

    Here’s one – ASIC banned short-selling (covered) during the GFC, but then lifted the ban in May 2009 – Why? They recognised that it was a problem before, so why did they lift the ban? Particularly as our markets (Aust & global) are still relatively volatile.

    SJ
    SJ
    11 years ago

    couldnt they just find a way to make the data transfers faster, regardless of where they are coming from?

    No, you can’t make the slow transfers faster. It takes about 1/10 of a second for a signal to travel half way around the earth, and there’s no way to speed that up. The only way to level the playing field, if that’s your goal, is to slow down the transfers from places that are closer to the exchange. The exchange could update its information once per minute, as Nick suggests, but even once per second would be an improvement.

    dorinny
    11 years ago

    Nicholas, what I meant by my question was related to your suggestion that,

    “a firm should be put onto a regime in which their shares are not traded in some substantial period before any major announcement that can be anticipated”

    – my point was that once the “insider” has committed the crime, the damage has been done, so to speak.. so what would be the point of restricting the future trading? or in the case of preventing insider trading by imposing preemptive restrictions, how would the company know that insider trading was about to occur? Maybe I’m not reading this right..?

    And I do know what insider trading is, Nic, I’m just suggesting that releasing company info ASAP once it becomes known by managers is a good way to diminish the opportunities of insider trading.

    “The only way to level the playing field, if that’s your goal”

    Does it need to be a level playing field? If a business decides to set up shop in a small lane, they might not make as many sales because they don’t get the foot traffic or exposure that they would on a more busy road, but then they don’t have the same fixed costs, do they? Not all business is created equal :)

    As for arbitrage of any sort, I truly feel sorry for any trader who tries to make a profit in this way – what a boring career path that would be.

    Dave
    Dave
    11 years ago

    Nicholas,

    “you put in two orders with the arbitrage margin and you can only make money”.

    I don’t understand what you mean. For example, suppose there are related products A and B, and you wish to buy A and sell B if the price of B exceeds the price of A by $1 or more, but otherwise you wish to do nothing. I don’t understand how you can excecute this trade in a single-shot market.

    “some period during which risk is taken”

    That is my point. The period currently can be very short, because of high-frequency trading. Your proposal would substantially lengthen the period.

    I agree that small pricing errors will not affect efficiency. But it is not clear to me how these pricing errors add up. Hence my comment about an iterative solution. You might have small price changes in each iteration, but still move quite a long way in aggregate to the final prices.

    Dorinny,

    I think that most micro-arbitrage would be undertaken by computer. Hence the vary fast speeds. I imagine that programming these computers would be a very interesting career path.

    Mr middle
    Mr middle
    11 years ago

    Fabulous idea. Perhaps the writer ought to trial test blind purchasing at the supermarket and let us know how that works for him.

    trackback

    […] a recent post I noted the massive investments that are going into moving the servers of traders for hedge funds […]