Yes, Virginia, HFT and Liquidity are Not All They Are Cracked Up to Be

You may have seen a big outbreak in the academic literature and business media of defenses of liquidity for liquidity’s sake, evidently prompted by increased interest in and in the EU, implementation of transaction taxes as a way to tame speculation and secondarily raise revenues.

And other efforts to curb market manipulation are underway. For instance, the SEC and the FBI are joining forces to try to rein in predatory HFT, but are finding it an uphill battle due to the fragmentation of information:

While some of the alleged tactics are the same as those used in the past, the new challenge is how they are executed. Algorithm-driven trades are issuing thousands of quotes in microseconds across numerous trading platforms, including “dark pools” that are off-exchange.

“Unless there is evidence of intent to manipulate, it’s hard to derive intent from raw trading data,” said one government official.

To bring a manipulation case, the government would have to find a way to show a person programmed the code to manipulate trades….

Among the challenges authorities face is a lack of comprehensive market data to work from. Until recently, the SEC had relied on official trade confirmations from what is known as the consolidated tape, a record of all transactions executed that does not include cancelled orders.

Since 2011, high-volume traders are required to report their equity trade data to the SEC. The SEC began using tools from Tradeworx, a high-frequency trading group, at the end of last year to provide it a real-time glimpse into stock trading.
But a plan to build a so-called consolidated audit trail is not expected to be completed for several years. The SEC’s team of quantitative analysts will help decipher the data received from the high-volume traders.

Even with these impediments, however, people who are knowledgeable about HFT practices can describe how it is in fact detrimental. A useful piece by former Goldman derivatives expert Wallace Turbeville explains in depth why it is extractive. Some analysts have suggested that HFT is destructive in volatile markets by withdrawing activity when it is most badly needed, but adds liquidity in normal markets. Turbeville debunks the idea that it plays a positive role even when markets are functioning normally.

A big reason for Turbewille’s clear, negative reading is that (mirabile dictu!) he defines the purpose of financial intermediation as facilitating real economy activity. From the start of his article at Demos:

The prior article in this series points out that the most important social purpose of the financial markets is to facilitate the movement of funds from (a) holders that seek investment opportunities to (b) businesses and governments who need to put investment capital to work in productive ways and to individuals who seek to borrow for their current needs. This function is referred to in this series as “Capital Intermediation.” The article describes findings that the cost of Capital Intermediation has increased significantly over the 35 years of financial market deregulation in the United States, despite advances in information technology and quantitative analysis that intuitively should have increased efficiencies in the process. Instead, Capital Intermediation has become less efficient.

The article starts by defining three type of traders: Value Investors (even momentum investors who operate on longer timeframes than HFT types can wind up being Value Investors), Market Makers and Information Traders (HFT and algos). As Turbeville explains:

It is obvious that a market participant is a liquidity provider only if the prices he or she quotes can be relied upon by other market participants, specifically Value Investors and those Information Traders who at the time are acting as liquidity takers. A price quote that appears on a screen is useless as a source of liquidity if it is not available when it comes time to transact. An Information Trader provides meaningful liquidity when his or her quoted prices represent levels that are reliable and meaningful to the participants who are liquidity takers. Sometimes an Information Trader provides such quotes and sometime it does not. When it is active, but not providing such quotes, it is a liquidity taker..

Therefore, Information Trader volume in the deregulated markets sometimes lowers transaction costs and at other times causes transaction cost to skyrocket. But in all cases it is designed to extract value well in excess of the bid/ask spread for the trader who generates the volume. And even when it reduces the cost of a given transaction, if it does not, on a net cost/benefit basis, make Capital Intermediation more efficient, it serves no social benefit. The evidence discussed below strongly suggests that high-speed algorithmic Information Traders extract far more value than they add, burdening the Capital Intermediation process….

The point is made eloquently by Eric Hunsader, the founder of Nanex, a high speed market date feed service.

In summary, HFT algos reduce the value of resting orders [for example, a Market Maker’s orders] and increase the value of how fast orders can be placed and cancelled. This results in the illusion of liquidity. We can’t understand why this is allowed to continue, because at the core, it is pure manipulation.29

One simplified way to think about the world prior to HFT was that one of the reasons in over the counter markets (and HFT is moving into OTC markets like foreign exchange) is that the market-makers had an incentive to be aggressive (which meant offer good prices most of the time to customers) because they would gain information advantages. It was all network effects: more customers would deal with you, including the ones that tended to be good market bellwethers; you’d be better able to place inventory or move a big trade discreetly. And having a broader view of the market (in bond king Salomon’s ad lingo, “market intelligence”) most importantly allowed you to manage your inventory better. HFT traders not only syphon off information, they introduce noise into trading:

Trading tactics are dictated by information. Some information is in the form of earnings reports, crop yields, political events and similar data. These types of information are directly related to the fundamental value of businesses and assets and are used to adjust the price of securities and derivatives to objectively sound levels. (As discussed above, fundamental value refers to the intrinsic value of a stock, bond or derivative based on available information.37) It has long been known that asymmetrical distribution of information that is directly relevant to fundamental value can distort the intermediation function of the markets.38

Another type of information relates to non-fundamental motivations that are predictive of market behavior by individual participants in a market. Tactics based on this type of information are driven by the expected short-term behavior of other market participants based on specific stimuli, not based on generally available information on supply and demand and company data. If, for example, a trader knows that a buyer who is relatively insensitive to price is currently in the market, he or she can corner the market for the specific category of securities sought by the buyer and exploit this price insensitivity.

This type of asymmetry can affect prices broadly as well as securing profit for an algorithmic trader in a single set of trades. This happens when the trading provides misleading signals as to fundamental value. The signals are generally not intentional; rather they arise from the perceptions others in the market have when they see the trades executed by an algorithmic trader using aggressive tactics.

Here is how it works: There is an assumption that trading is motivated by views on fundamental value. Traders must operate under this assumption to protect themselves from unforeseen large price moves based on fundamental value information. As a result, a reported trade will be interpreted as a statement about the buyer’s or seller’s perception of fundamental value, at least until other information contradicts this assumption. A trader with a view as to fundamental value will be forced to question his or her view when he or she sees a trade that is not in line with his or her perception of fundamental value. Trading behavior based on non-fundamental information is indistinguishable from trading based on fundamental information to other market participants (unless the trader is using software designed to detect an HFT tactic, which sometimes happens). Trading against the non-fundamental trader entails risks that he or she has superior fundamental information, and since no trader can afford infinite losses, the fundamental trader will, at a minimum, change his or her price of indifference or perhaps exit the market.39

And HFT now dominates equity trading:

It has been estimated that today 73% of equity trading volume is a result of algorithmic and high frequency trading.3 HFT has changed fundamental characteristics of markets. It has been estimated that at the end of the Second World War, the average holding period for stocks was 4 years. By the turn of the millennium, it was 8 months. By 2008, the average holding period declined to 2 months. And it has been estimated that, at least for actively traded shares, it had declined to 22 seconds by 2011.4 Obviously, trading that churns the market has increased tremendously.

A market with an average holding period of 22 seconds has nothing to do with investing and everything to do with profiting from churn.

Turbeville provides some eye-popping examples of how predatory some of the strategies are, such as:

The best indication of the level of aggressive algorithmic strategies is provided in a study of the practice of “Quote Stuffing.”34 This is a tactic in which HFTs flood an exchange or other transaction-matching venue with quotes to buy or sell in order to slow down the venue’s processing times. HFTs do this so that aggressive tactics can be implemented without intervention from other traders. They also employ Quote Stuffing to slow down one exchange so that price differences between that exchange and another that is operating normally can be exploited.35 Therefore, when quote stuffing occurs, it is highly likely that an aggressive tactic is underway. The study reaches the following conclusion:

We find that quote stuffing is pervasive with several hundred events occurring each trading day and it impacts over 74% of US listed equities. Our results suggest that, in periods of intense quoting activity, stocks experience decreased liquidity, higher trading costs and increased short term volatility.

Quote stuffing operations have been estimated to “consume 97% of computer system resources that the whole market has to bear.”36

Now that you understand how HFTs can use quote stuffing to displace other participants, let’s look at one of the ways they make money:

A classic aggressive strategy involves hunting and trapping “whales.” It is a good example of aggressive HFT strategy.

It is initiated by a “pinging” operation. This involves placing multiple orders designed to detect the presence of a market participant with a large position that it is in the process of accumulating or liquidating. “High frequency traders employ pattern recognition software to detect large institutional orders sitting in dark pools or other liquidity venues.”41 When it pings, the algorithmic trader places orders to buy and/or sell in an array of prices inside the bid/ask spread. If the potential target market participant starts walking through the orders, it becomes clear that a large position is in play that could consume all of the liquidity on one side of the market (bid or ask). Traders sometimes refer to the potential target as a “whale.” Often, a whale intends to buy or sell a large block for structural reasons and is relatively less sensitive to price than other buyers and sellers. If the trader’s order strategy reveals price insensitivity, the whale becomes a target.

For example, an investment fund might have experienced large redemptions from its investors that require liquidation of investments under the terms of the fund. The fund is compelled to sell promptly and is relatively insensitive to price. It is a properly motivated whale.

Once a whale is sighted and its price insensitivity is confirmed, it becomes clear that dominating the other side of the market will give the algorithmic trader control of the bid/ask pricing mechanism. In our example, the algorithmic trader would meet the offers of all of the bidders for the shares that the whale seeks to sell, becoming the sole and dominant purchaser in the marketplace. The algorithmic trader establishes a narrow band of absolute market power controlling the particular securities or derivatives that the whale seeks to buy or sell. The whale will be compelled to either abandon the market or transact at the price demanded by the algorithmic trader.

In an instant, the algorithmic trader cancels all “pinging” orders and buys the market in order to extract as much value as possible. It will also likely flood the trading venue with orders to create congestion and slow other market participants who are also watching and might intervene. The whale must transact at the price required by the algorithmic trader to accumulate or liquidate the position, as appropriate. When the transaction with the whale is accomplished, the quote stuffing orders are all cancelled.

In the final step, the algorithmic trader liquidates the position acquired from the whale immediately, but at a new bid/ask spread reestablished at market levels that are more competitive. The algorithmic trader intends that its profit from moving the bid/ask spread to take advantage of the whale’s price insensitive motivation will exceed the potential loss from the operation if the whale detects that it is being taken advantage of. For this reason, the strategy is most useful if all of the steps can be executed quickly, before detection is possible.

By displacing the entire side of a bid/ask spread, the HFT eliminates all meaningful depth of buying or selling interest for a short time period. As a result, the observed spread in the market is meaningless. Various studies that examine the narrowing of bid ask spreads in the modern marketplace miss the crucial point that HFT tactics like the one described are missing the point that the narrowing is an illusion.42 If the study draws the conclusion that HFT narrows spreads, it is simply misleading.

And the prevalence of this sort of strategy turns the logic of investing on its head. From a 2011 speech by the Bank of England’s Andy Haldane, The Race to Zero:

As of today, the lower limit for trade execution appears to be around 10 micro-seconds. This means it would in principle be possible to execute around 40,000 back-to-back trades in the blink of an eye. If supermarkets ran HFT programmes, the average household could complete its shopping for a lifetime in under a second. Imagine.

It is clear from these trends that trading technologists are involved in an arms race. And it is far from over. The new trading frontier is nano-seconds – billionths of a second. And the twinkle in technologists’ (unblinking) eye is pico-seconds – trillionths of a second. HFT firms talk of a “race to zero”. This is the promised land of zero “latency” where trading converges on its natural (Planck’s) limit, the speed of light….

This has added a new dimension to the “adverse selection” problem in economics – of uninformed traders suffering at the hands of the informed. Being informed used to mean being smarter than the average bear about the path of future fundamentals – profits, interest rates, order flow and the like. Adverse selection risk meant someone having a better informed view on these fundamentals.

Adverse selection risk today has taken on a different shape. In a high-speed, co-located world, being informed means seeing and acting on market prices sooner than competitors. Today, it pays to be faster than the average bear, not smarter. To be uninformed is to be slow. These uninformed traders face a fundamental uncertainty: they may not be able to observe the market price at which their trades will be executed. This is driving through the rear-view mirror, stock-picking based on yesterday’s prices.

It would actually be pretty simple to end the parasitic role of HFT, say, by requiring all orders to remain open for a second. It’s astonishing to see the SEC blandly permit the rise of this practice and now be scrambling to try to curb it. But no one seems to question predatory finance until it has become so well entrenched that it’s difficult, both politically and practically, to uproot it. Until we solve the problem of cognitive capture, however, we are certain to have finance become even more predatory.