Thursday 1 November 2012

Algorithms, an abuse of a good instiution

I don't normally read a Sunday paper as the Saturday one takes me all weekend to digest, but a friend  (dubbed a "Thoroughly good egg" by a reader of this blog and designated a "member of the Liberal √©lite" by his Tory MP) has alerted me to an article in the Observer a couple of weeks ago in which the writer, Simon Neville, reveals  how traders now operate on the US Stock Market.

Apparently 75% of all trades are initiated by algorithms on computers.  The shock statistic, to me at any rate, is that the average purchase is held for just 22 seconds (my emphasis).

I was taught that the "invention" of the limited liability joint stock company was as important to the industrial revolution as the invention of the steam engine, in that it enabled capital to be brought into economic and industrial development without endangering the entire assets of the investor.   Stock exchanges  are a natural extension of the concept, in that they enable stock to be sold without delay, though perhaps at a loss, should the investor need access to  his or her capital.

All good ideas are open to abuse, and it has long been the case that stock exchanges function as gambling casinos for those with wealth.  It has long been known that investors can act irrationally (the fall in the share prices when Harold Wilson caught a train to visit his sick father was a good example from my earlier teaching days)  but it is surely outrageous that the gambling can now be devolved to machines capable of acting with such speed.

I am insufficiently inducted into the mysteries of share and asset dealing to have precise solutions as to how to return stock exchanges to their original function, but something should be done.  For starters I suggest either a whopping capital transactions tax, not just the faction of 1% advocated by Tobin, or regulation that would require stock to be held for a minimum period, say six months, before it can be re-sold.

More informed suggestions welcomed.


  1. Whilst agree regarding the problem, I don't think that your solutions are realistic. A slow trading market would simply die - the whole concept of the modern market is based around fast trading and market reactions. Similarly with tax - big tax on trades would simply make me shift to a more liberal market. If a systems output moves too sharply and modally the best thing to introduce to it is more chaos and non-linearity. I think this would solve many of the problems, and can be achieved simply by opening the markets up - making changes to increase the number of small trades by individuals rather than brokerages or larger traders. A limited open API for the LSE would go a long way to doing this - we should make market trading as appealing, popular and simple as possible.

    If the nation gambled more on its industry than on random balls flying out of a tumbler we'd all be a lot richer, there would be greater market stability and there would be no need to fudge the system for our own piece of mind. Also, I'll wager that our economic crisis would disappear pretty quickly too. Access to the system holds every part of the process back - why do we need companies like Kickstarter when we've got a perfectly good system for this sort of thing already? I think it's because the market is inflexible, inaccessible and daunting for the individual.

  2. The article I quote refers, I think, to the New York stock exchange but I suspect much the same sot of thing is going on elsewhere. So the changes I suggest would have to have agreement from at least the other major exchanges. Although my suggestion of a six months embargo before shares can be resold is perhaps somewhat over the top, I think slowing down the market can only help to get away from the gambling aspect and back to its original function. However,as I admit in the post, I am not an expert in this area and welcome suggestions from those better qualified.

  3. This is not the case - it's not the speed of a system that controls its volatility, it's how modal/chaotic the inputs are and how many of them have non-linear interactions. Slowing down the market would stop it from being reactive and kill it; it has to be able to react quickly or it's pointless. The idea that any exchange would accept anything like this is madness and it will never happen (it makes the markets a poor method of trading, there would be much better systems for the sort of market you describe). Reactions are important and necessary to all living systems - your brain is mainly set up for subconscious reaction; any system that has reactions that are far slower than the events that interact upon the system is inefficient and doomed (in a Darwinian sense). What you term the "gambling aspect" is a necessary function of a market that can adapt quickly to the inputs; if they didn't gamble the market would collapse.

    The problem with computer driven trading is how modal the system becomes; the computers are trading for thousands of people simultaneously, and so the thousands moves as one; that is the dangerous mechanism. It's not the computer trading that causes problems, it's that they're actioning trades on behalf of many people and that creates market modality that causes strange interactions between the competing systems. The answer is not to impede the system as a whole, but change what generates the actions - get more people trading and less computers.

    The system simply needs more chaos.

  4. I think we have different view of what the market is for. In my view it exists to bring sellers, who have something to offer, together with buyers, who have need of what is offered. In this case, in the simplest terms, what is offered is capital from investors with liquid capital to spare, which entrepreneurs need to finance their real investment.

    That the resulting shares can be exchanged is a useful device to enable assets to remain more liquid. It is hard to see how the gambling aspect, ie simply exchanging the assets when they have increased in value rather than providing new capital, can be eliminated entirely, but I am looking for measures to reduce this to a minimum. If not taxation or regulation, when what?

    Getting more people to trade (how?) would indeed reduce the individual profits to be made, but stock exchanges would remain gambling casinos rather than sources of real capital.

  5. 1) What evidence is there that the stock exchanges aren't performing their "original function"? As far as I know if you have spare capital you can buy shares and bonds, and if companies need capital they can emit them. If you are a long-term investor, the fact that computers are trading that asset a million times a second while you are holding it should be immaterial to you. And when you do finally decide to trade, you'll always have a counterparty there ready to take the other side*.

    2) I have yet to find a definition of gambling and of investment which sets them apart. It's certainly not the holding period. Losing or making lots of money in 1 second has exactly the same end result as doing it in a year, it just feels different. Every investor who believes he is buying undervalued assets looks like a gambler to someone who believes they are correctly valued and like an idiot to someone who believes they are overvalued. That's what a market is - if on balance the participants agreed with each other the price would be elsewhere.

    3) You lament the irrationality of human investors. How then can you be upset about unemotional computers making decisions based on statistical observations?

    I have a lot more to say about the subject but I'm interested to hear what you think about what I've said so far so I'll leave it at that for now.

    * (Note that the alternative is an illiquid market where you have to turn yourself into a trader by waiting for that liquidity, perhaps in the meantime hedging yourself in more liquid markets in order to protect against adverse price movements. That there are firms dedicated to taking the position off your hands because they can handle the risk better than you while waiting for an opposite trade is just an instance of division of labour and specialization. Given the number of transactions that can occur daily in a highly traded underlying - think EUR/USD - the fact that these firms hold on to their positions for such a short average time is unsurprising.)

    1. Meanwhile...

  6. Thanks for your comment, Eric, and thanks for directing me to the FTAlphavile site, which I've read with interest. Clearky algorithms are the coming thing, particularly now they have their own diminutive - algos.

    I agree that all investment is made with a view to making a financial gain (your point 2) but there are two ways of achieving this. One is backing a good idea which you think will lead to a stream of income in the (long term) future; the other is, as you mention, to acquire an undervalued asset and to flog it to make a capital gain at the earliest possible moment. The first is socially and economically useful, the second has only limited value (arbitrage etc).

    Clearly today's markets are much more concerned with the second rather than the first motive, and the use of computer algorithms increase this predominance. Short-termism has for years been the major obstacle to genuine economic development in the UK; make a quick profit and move on rather than support long-term investment. This emphasis on the short term means that managers are far more concerned with maintaining the current share price, by whatever means, in order to maintain the value of their share options or avoid a take over, than they are with the long run viability of their enterprises.

    Hence I advocate in the need for measures to shift the market to providing capital for and supporting enterprises likely to provide a stream of income over a long period, rather than the search for a fast buck. Higher taxation or regulation seem to me to be two ways of achieving this, but I'm open to suggestions

  7. FTAlphaville posted this link today which is a much better overview of HFT than I anything could write.

    I don't really understand your objection to buying mispriced assets in order to make a capital gain. If you are buying an undervalued asset, then the person selling it to you presumably considers it overvalued (or in any case, for whatever reason, they'd prefer to have those X pounds rather than the asset). Even if this isn't useful for society at large it's certainly useful for them! But actually it is useful for society at large! Exploitation of mispricings is what causes the price to reflect the information. And this has real economic value because if the prices are wrong, how can people make optimal decisions about how to allocate resources? We see this every day - goods that are consumed too much or too little because the price transmission mechanism is broken, sometimes because of laws or taxes. Impeding it doesn't make the prices better.

    In any case, aren't the 'two ways' you mention in reality the same thing? If you're interested in the long-term revenue stream (way #1), what you're subconciously doing is you're estimating of what that will be, you're discounting it, you're calculating an NPV and you're comparing it with what you have to take out of your wallet today. For it to be worth the bother, you'll want that NPV to be at least what you're paying upfront for the investment - either that or you have to really really like the underlying business or the people running it. Investing is about foregoing purchasing power today in exchange for more purchasing power in the future. Anything else is charity (which is not a bad thing, but not what we're talking about here) or stupidity.

    So if you're comparing your evaluation of the asset's NPV with its market price... well isn't that just way #2? Capital gain can result from the market realizing that you were right about the business and its future flows and incorporating it into the price. I would suggest that an investor who buys an asset because he considers it undervalued will sell it when he considers it correctly valued, not at the 'earliest possible moment'. The holding period is of no consequence to a rational investor - all that counts in any given moment is: which does he value most, the asset or the pounds he could get by selling it?

    Regarding your other points I reject that the speed at which money is made is what makes it right or wrong. I think that the reason this association is made is that people are afraid of instability. So OK - taxation which makes trading unattractive, longer holding periods, regulation, probably make the system more stable. But is that really true? Or do they just constrain it so that when it does finally break down the impact is even more violent (finance is full of examples of that happening). There is, as anonymous pointed out, a trade-off between stability and the ability to react.

    So anyway I apologize for the long ramble - perhaps you were looking for were suggestions on how to achieve your aim rather than a discussion about its merit!

  8. The comment below is from Jaime, who Emailed it to me because for some reason the algo (sic) which protects the security on this site refused to allow him to post it himself.

    I hesitate to join this debate, but when Eric says "anything else is
    charity or stupidity" there is a third possibility: gambling or
    speculation, ie putting your resources (or others') into some project
    ignoring or deluding oneself about the balance of present costs and
    future benefits. I think this comes pretty close to the definition of
    speculation that I am looking for. And of course it is not just in the
    financial sphere that this happens.


  9. No need to apologise for the “long ramble”, Eric: what you’ve said is very enlightening, and I’m particularly grateful for the link to FTAplhaville, which, as you say, gives a very clear explanation of High Frequency Trading. (Incidentally, could you explain how to put a link in a comment? I know where the facility is when writing a main post, but it does not seem to feature in a comment, but you seem to know the trick.)
    I think we need to distinguish between “real investment,” which is putting up money for some project which will enhance the utility of humankind (even if in a frivolous way, as in, say “pushpin” parlours) and “speculative investment,” which is the transfer of ownership of that piece of investment in the hope that it will appreciate in value. Whoever owns the title to the investment once the “real investment” has been made is of no consequence, as you argue, but only provided that the speculation does no harm. (You seem, by stating “(e)ven if this isn’t useful to society at large it’s certainly useful to them” to be aware of Adair Turner’s comment that much of the activity of the financial sector is “socially useless.”)
    However, I think it does harm in at least three ways:
    • A point I’ve already made: It forces managers to concentrate too much on the current value of their company as perceived by the speculators, rather than its long term potential. Hence too much manoeuvring to avoid a hostile take-over or a drop in the value of their share-related bonuses. Among other things, valuable research and development, which does not produce an immediate return but which may improve the quality of our lives I the future, is neglected. This has long been one of the curses of British industry.
    • It siphons off very able people (you?) into this essentially socially useless sector rather than enticing them into life enhancing occupations. I have just bought “Meme Wars” by Kalle Lasn (published Adbusters this year I think) and, opening it at random, read this quotation from a student at Oxford: (my lecturer)asked, “So how many of you want to work in the Civil Service when you’re older?” . . . . . .No…arms were raised. “So how many of you want to go work in the city: Invest, trade, move money and make money?” Arms shot up all around me. (Can’t give you a page number as the book doesn’t have them.) As a good Liberal I was never a great admirer of Harold Wilson, but his comment that “One of our problems in this country is it is too easy to make money rather than to earn it” clearly remains valid, today, probably moreso.

    • As your FTAlphaville link acknowledges, “High Frequency Trading causes instability” And, although the article doesn’t go on to say this, when the trading is in derivatives, credit default swaps and such-like which the traders don’t understand, that instability can cause a world recession and misery to millions.

    I think I’ve said just about all I can on this topic, but please feel free (and “anonymous” as well) to have a last word if you wish.

  10. The discussion could continue forever really, so I agree we should stop here. Just to reply to your last post...

    If I've understood correctly you feel that investment should really have some moral quality to it - i.e. you invest a) to maintain the value of your wealth (possibly making a reasonable return without exaggerating) but also b) to put that capital at the disposal of projects which will benefit society at large. On the other hand you view investing which is purely aimed at making money without entering into the merit of the underlying investment as something which needs to have brakes applied to it because it incurs costs to third parties. Actually I personally believe that statistically (obviously someone makes lots of money from gambling by pure chance) the market already punishes the second class of "investors" anyway. As the famous investor Benjamin Graham said, "speculation is neither illegal, immoral nor particularly fattening for the wallet".

    Re the "socially useless" comment... what I meant was if I'm hungry and I eat a sandwich, then that's useful for me. The fact that it doesn't benefit society at large isn't a reasonable argument for stopping me from eating it. The same should apply to exchanging an asset for another with a consenting counterparty. It seems to me a sound principle for a free society that unless there is enough evidence that something has externalities, it should be freely allowed.

    I personally don't believe such evidence exists for HFT. I know the FTAlphaville link says "HFT causes instability" but to my knowledge this isn't in any way the academic consensus. Actually from what I've read there seems to be more evidence that the presence of algorithmic traders reduces volatility - it's the irrational humans that screw up the markets! This makes sense, as trading algorithms are emotionless in their reasoning, they don't enter trades unless there's a statistical reason to. Human trades depend on things like testosterone/sugar levels, fear of judgment, stress because they have lost money etc, all things which should be irrelevant. As the HFT article says, algorithms can have their disadvantages as they do things which are evidently stupid to a human. On the other hand humans do a lot of things which are evidently stupid to an algorithm (i.e. stupid when examined by statistical analysis).

    In any case the whole Tobin tax/HFT discussion is very political - it's such an esoteric part of the economy which I'm convinced affects everyone much less than is popularly believed. If, as it seems it will, the tax does get implemented in the Euro area, I would not be surprised if it is watered down to the point where the impact is minimal (it will probably only be paid by long-term investors - there's already talk of intraday trades being exempt). All the politicians care about is the tax revenue, so they will extract it from those they can and will be hesitant to do anything which could damage European financial sector, which isn't exactly in a great state.

  11. Thanks for your final word, Eric. I love the Benjamin Graham quote.