Dust flux, Vostok ice core

Dust flux, Vostok ice core
Two dimensional phase space reconstruction of dust flux from the Vostok core over the period 186-4 ka using the time derivative method. Dust flux on the x-axis, rate of change is on the y-axis. From Gipp (2001).

Tuesday, May 1, 2012

It isn't the speed; it's the illusion of liquidity

At last the Toronto Star has an opinion piece on high-frequency trading deploring the condition of the markets. Unfortunately the piece focuses on the microsecond advantages that some well-heeled entities have created for themselves by laying millions of dollars of fibre-optic cable, and misses the most important problem--the ability of some players to place and remove bids at inhuman timeframes of thousands per second.

Historically, the person with the best access to information has an advantage over the other participants. I don't view this as being unfair--it's the way it has always been. Furthermore, when those who have an information advantage act on it, they add information to the market, so that the time-advantage of advance knowledge dissipates quickly through the mechanism of the market.

The real problem with rapidly creating and cancelling orders is that it strikes at the fundamental trust underlying the market. When you decide to purchase a stock, you normally do at least a cursory check of liquidity. If, for instance, you look at ABC on the exchange, and the best bid price is $8.00 and the best ask is $2.00, you know there is no liquidity, and it is a stock to be avoided. If the bid is $8.00 and the ask is $7.99, and there are reasonable numbers of shares on offer, then there is liquidity and you should be able to get in or out without too much trouble.

What happens when the entities that have made the best offers have no intention of filling their orders? A casual inspection of market depth might convince other market participants that there is a lot of liquidity available for that particular stock, but the liquidity is an illusion which lures market participants into unsuitable investments and traps them there. Inevitably, the entrapped participants will leave the market and never return.

A store might advertise a product at less than half the usual price, along with a disclaimer that the price may change without notice. We recognize that there may be contingencies that may force the store owner to raise the price. The supplier may call and say there is no more supply. But suppose the contingency that caused the store owner to raise prices was a customer entering the store to buy the product. So the article is advertised as being on sale for $1, but if anyone tries to buy it, the store owner declares a contingent increase in price to $2. The article is for sale at $1, unless someone tries to buy it. How long before the customers stop going to the store? The advertisement may not have been a legally binding contract, but it does create an expectation.

The same expectation is created by the illusion of liquidity. This illusion causes great harm to a motivated buyer (or seller) of securities, particularly if they are motivated to buy or sell a large order. Imagine that you are managing a pension fund, and you find yourself with a million shares of a junior mining stock to sell. You look at market depth, and lo and behold, someone is willing to buy a million shares at, say, 12 cents. You decide that that is an acceptable price; but when you try to fill the order, it is immediately cancelled (perhaps 1000 shares are successfully sold at 12 cents) and replaced by an offer to buy a million shares at 11.5 cents.

Now what? You might wait and see if any more of your offered shares are picked up, but if they are not,  you have a problem. Institutions don't have cheap trades. Your shares are in the form of a certificate in a safe--for you to complete the sale of 1000 shares you have to take the certificate from the safe, deliver it to the transfer agent; have the transfer agent split the share certificate, sending one certificate to the new owner and returning the balance to you; then you have to return the certificate to your safe, but this process has to be audited (i.e. witnessed by a lawyer, who verifies that the certificate is for the correct amount, and has been placed in the correct compartment in the correct safe). All of this costs more than $7.99 that a discount brokerage might charge. So chances are that you will be criticized for a $120 trade.

So you try again at 11.5 cents, but once again only 1000 shares are sold before the buy order is cancelled and replaced by one at 11 cents. Well, you can see how this goes. You keep trying to sell and the price keeps falling until there is real market interest; perhaps at 6 cents--or lower. Had you known that you would have to sell at 6 cents you might never have started. You were lured into a bad trade by the illusion of liquidity.

The practitioners of HFT claim they are supplying liquidity to the market, but this is not the case. They are only creating the illusion of liquidity, and this illusion has drawn a lot of money into the market--money which would not have entered the market without the illusions. Unfortunately, drawing this money into the market is politically advantageous for North American governments who can use it as evidence for an improving economy--consequently there is no interest among the political class for a fair solution to the HFT problem. It is not a problem for them. But it is a problem for everyone else in the market, even, ironically enough, for the high-frequency trading firms.


  1. Excellent post. We should recognize, though, that this same issue threatens markets without HFTs. Consider markets with few participants... If a single participant writes a large number of bids, they may fully intend to execute at that price but a single event impacting the participant could repeal the volume near the execution.

    This is common in heavy commodity markets. The "market" solution to this is increased speculation. More speculative participants in a market mollifies the impact of non-market events (such as those on single participants). This isn't a new concept, either, it is long understood that more participants promotes fluidity even when volatile.

    We face a threat to this reality though. The United States presently seems intent on blaming speculators for our fiat-induced bubbles. As a result many leaders are seeking to inhibit speculation, reduce market participation and put up barriers to entrench the current large players.

    I wonder if a more appropriate solution would be for exchanges to track the volume related to uniquely margined accounts. This would then give participants transparency to whether the liquidity is "real." If we knew some measure of "ability to fill" then we could address this problem to some degree.

  2. I was going to write something on possible remedies.

    I don't think measuring the ability to fill an order is helpful--because usually these characters are completely capable of filling the orders they place. The problem is that they have placed the orders without any intention of filling them--and I don't know how we could measure anyone's intent.

    One of the solutions is to reduce the speed at which orders can be cancelled. Nanex has proposed that there be a minimum duration for an order--even were that duration to be a second, we would see an improvement in the behaviour of the market.

    Provided the retail market returns . . .

  3. Not solely based on the ability to fill (for certainly these players have the margin!) but also on the concentration of offers originating from unique accounts.
    For an over-simplified example consider how analytics tracks web traffic. You get values for time spent on pages, quantity of visits but also stats on unique visits.

    I contend that if I could see some metric of volume/account concentration that I could get a more accurate picture of the number of independent participants and the depth of their pockets. When I brought up ability to fill I was thinking ahead of my idea: that one simple way around it would be to create multiple small accounts and thus trick other algos into thinking there was "real" volume... that's where a metric of margin plays out. If one house opened 10,000 trading houses then each account would have more obstacles than a single account of the same balance provided the clearinghouse instituted a delay on balance transfers.

    I don't know that a bid lifetime is the solution but it would certainly be interesting. If you saw Dr. Sean Gourley's TEDx talk it addresses the minimum human decision times... Perhaps something could be made sufficiently small to limit the algos but that doesn't address your main point here: that the *existence* of the HFTs is not inherently the real problem.

  4. I think I understand your point better here. In a market with a healthy retail participation, if you saw a million shares on offer, chances are they would be a thousand or so orders from different participants. So if you could see whether the bid was entirely placed by a single entity as opposed to thousands of entities that would give an indication of the "reality" of the liquidity. I am not so sure that it would be too expensive for the traders to create thousands of fictitious accounts, as this is a problem that could be solved through programming.

    IIRC, RBC's Thor program was written to try to take advantage of myriads of orders that were scattered across the different exchanges, and to tailor lags for each order so that they would all ideally be exercised simultaneously. It seems to me then that the HFTs are already splitting up block orders into lots of smaller artificial entities.